Rug history

The Turkish Knot and the Persian Knot

The symmetrical Gördes knot and the asymmetrical Senneh knot are structures, not nationalities. A 2025 peer-reviewed study shows both names are wrong, Ottoman court rugs were tied with the Persian knot, and no measurement supports the claim that either knot wears better.

Mehmethan Özer · October 20, 2026 · 21 min read
Tezgâhta elle düğüm atan dokuyucu — Gördes ve Sine düğümü

Last updated: October 2026

Short answer: Rug weaving uses two basic knots: the symmetrical knot (commonly called Turkish or Gördes) and the asymmetrical knot (Persian or Senneh). Both geographic names are wrong. A peer-reviewed study published in 2025 in the journal Yedi showed that the name "Gördes knot" rests on a false etymology, that "Senneh knot" is factually incorrect, and that the term "physician's knot" has no basis at all. And there's more: Ottoman court rugs were tied not with the Turkish knot but with the Persian knot. Museums around the world already avoid the geographic names. They say symmetrical and asymmetrical.

One of the lines you hear most often when buying a rug is this: "This rug was tied with the Turkish knot, so it's stronger." Both halves of that sentence are shakier than they sound. In this article we explain what the knot physically is, where its name came from, what has actually been measured in which parts of Anatolia, and how much any of it has to do with quality, drawing mostly on academic sources published in Türkiye. If you'd like a close look at the knot structure of our handwoven rugs, you can browse our handwoven rug collection.

Two Knots, Two Structures

Annette Ittig's entry on knotted-pile techniques in the Encyclopaedia Iranica defines the two structures this way:

The symmetrical knot is wrapped around two warp threads, and both of its ends come up between that pair, mirroring each other. The knot grips the pair of warps.

The asymmetrical knot is tied to the same pair of warps, but only one end comes up between them; the other emerges after passing over a single warp. That gives the knot a direction in which it opens: open to the left or open to the right. The left-opening variant is the more common.

This description is not an academic abstraction; it is the language museum records actually use. The Victoria and Albert Museum's record for the Ardabil rug (museum no. 272-1893), the most examined rug in the world, describes its pile structure as "asymmetrical knot open to the left."

A small but important warning: Öztürk's 2025 study finds an error in most of the knot diagrams in Turkish sources. In drawings of the symmetrical knot, the crossing is shown below the pile ends, when it should be above them. In other words, some of the diagrams we have been seeing in textbooks for years are drawn wrong.

When Did These Names Appear? Far More Recently Than You'd Think

Rugs have been knotted in Anatolia for at least eight hundred years. The knot types have had names for less than a century.

Öztürk's chronology runs like this: no formal name was given to the knot types until the end of the nineteenth century. The first widely used pair of names appears in 1916. The symmetrical/asymmetrical terminology shows up no earlier than 1922, and it only becomes widespread after the 1970s.

So "Turkish knot" is not a Seljuk term; it is a label born in Western rug literature of the early twentieth century. That alone does not make the name wrong, but it does show how recent, and how external, the naming is.

Where Does the Name "Gördes Knot" Come From? An Etymological Mistake

The spine of this article is a peer-reviewed paper by Bahadır Öztürk of the Dokuz Eylül University Faculty of Fine Arts, published in the journal Yedi (no. 33, 2025, pp. 163–176). The paper's central finding is blunt: "some expressions widely used in the naming, such as 'Gördes, Sine, Hekim knot,' have been found to be entirely wrong."

The old name of Gördes is not "Ghiordium." In the historical record between 30 BC and AD 640, the town appears as Iulia Gordos. Western writers built a connection between "Ghiordes" and Gordion, the Phrygian capital, on nothing more than a resemblance in sound, and tied the rug knot to the legend of Alexander the Great and the Gordian knot. In Öztürk's words: "The name Gördes has no connection whatsoever with the city of Gordion/Gordium, where Alexander the Great cut the legendary knot."

Geography settles the matter anyway: Gordion lies within the Polatlı district of Ankara, 330 kilometers from Gördes as the crow flies.

Gördes Did Not "Invent" the Symmetrical Knot

The second problem with the name is historical, and Ottoman production figures make it measurable.

According to production figures for 1884, Gördes was a secondary center in the Anatolian rug trade of the time: Gördes 25,547 m², Oushak 83,795 m². The gap is more than threefold.

The scale of Oushak becomes even clearer in a study published in Belleten: at the start of the twentieth century Oushak had 2,000 looms and 8,000 workers, and between 1885 and 1896 output had risen from 75,000 meters to 250,000 meters. In 1910 the Rug Syndicate alone employed 15,000 workers.

Was the difference between the two centers only one of volume? A 1917 Ottoman report on the rug trade, published in the journal Arış, says something sharper: Oushak work was known for its durability and was chosen for commercial spaces such as hotels, shops and staircases, while Gördes work was comparatively coarse and went mostly to the domestic market.

In other words, a knot type was named after a town that was neither its largest nor its finest producer.

"Senneh Knot" Is Even More Wrong

The name "Senneh," given to the asymmetrical knot, is also factually wrong. In Öztürk's words: "Not every Iranian rug, and not even every Senneh rug, is woven with the asymmetrical knot." The study identifies a symmetrically knotted Senneh rug dated 1880.

Beyond that, a source from 1914 writes that Shiraz rugs were woven "usually with the Senna, sometimes with the Ghiordes knot." Even within Iran, then, the knot type varies from region to region, and even from workshop to workshop.

There is a second problem as well: Senneh lies within Iran's borders, but "the city's people are not Fars or Persian"; it is a Kurdish provincial center. The name "Persian knot" therefore misdescribes the structure, the geography and the community all at once.

What Is the "Physician's Knot"? Nothing at All.

The expression "physician's knot" (hekim düğümü), which turns up now and then in Turkish sources, has no basis whatsoever. In Öztürk's words: "Not only do physicians not use a single type of knot in surgery, but none of the knots they do use resembles the symmetrical knot."

Ottoman Court Rugs Were Not Tied with the Turkish Knot

This may be the most unsettling part of the article, and the source is a Turkish scholar.

Bekir Deniz of Ege University, in his study "The Sources of Anatolian-Turkish Rug Art" published in Sanat Tarihi Dergisi, describes the Ottoman court rugs this way:

"These rugs, which use silk, wool and cotton in the warp and weft and wool and cotton in the knots, were woven, unlike Anatolian rugs, with the Persian knot (Senneh)."

— Bekir Deniz, Sanat Tarihi Dergisi, XIV-1, 2005

The same study states that Oushak rugs were woven "in wool and with the Turkish knotting technique." So even within Anatolia, the two lived side by side: symmetrical in folk weaving, asymmetrical in the court workshop.

The name "Turkish knot" leaves out the Ottomans' most magnificent rugs. It is hard to find a better example of how wrong a name can be.

Why Is Geographic Naming Misleading?

This debate is not confined to Türkiye. The Encyclopaedia Iranica declares the older terminology problematic outright:

"This older terminology is problematic, because both knot types are used throughout the Near and Middle East, regardless of geopolitical boundaries or the linguistic or ethnic affiliation of the weavers."

— Annette Ittig, Encyclopaedia Iranica

Öztürk arrives at the same conclusion: Turkic communities in Iran used the asymmetrical knot, while the Iranian regions of Shiraz and Heriz used the symmetrical one. A knot type belongs to a workshop tradition, not to a nation. The title of the exhibition the MAK museum in Vienna mounted in its 2002–2003 season reflects that choice at headline level: KNOTS SYMMETRIC_ASYMMETRIC.

Anatolia's Oldest Known Rugs

The concrete ground beneath the terminology debate is the oldest Anatolian rugs that survive. Five of them have verifiable records:

Object Museum / Inv. No. Date Structure in the record
Konya Alâeddin Mosque rug Museum of Turkish and Islamic Arts (TİEM), inv. 681 13th century Gördes (symmetrical) knot, wool on wool; 519 × 294 cm
Rug with animal figures The Metropolitan Museum of Art, inv. 1990.61 14th century, attributed to Türkiye Symmetrically knotted pile, wool; 165.1 × 138.4 cm
Marby rug Swedish History Museum, inv. 96378_HST 1300–1420 Gördes (symmetrical) knot; 160 × 112 cm
"Star Ushak" rug The Metropolitan Museum of Art, inv. 58.63 Late 15th century, Oushak Symmetrically knotted pile, wool; 430.5 × 232.4 cm
Rug (Central Anatolia) The Metropolitan Museum of Art, inv. 2009.458.2 15th–16th century Symmetrically knotted pile, wool; 236.2 × 188 cm

TİEM 681 is one of eight rugs (four complete, four fragments) found in the Alâeddin Mosque in Konya by F. R. Martin in 1905–1906, and it entered the museum on March 31, 1911. For Met 1990.61, the museum's own curatorial note says it is one of three rugs of such an early date to survive complete.

About the Marby rug, honesty is called for: the Swedish History Museum records it as tied with the Gördes knot, but gives its place of production as "Armenian Highlands, Anatolia." The museum record does not support the statement "the Marby rug is a Turkish rug"; it does support "it is of Anatolian origin and symmetrically knotted."

Note: None of these objects has a knot-density figure in its museum record. Museums state the structure; they do not give a count. Treat the knot counts quoted online for these rugs with caution.

What Has Been Measured in Which Parts of Anatolia?

So far we have talked about terminology. Now for the real question: which knot is used in which part of Anatolia, and how densely?

We compiled the answer not from what sellers say but from Arış, the journal published by the Atatürk Culture Center, and from university journals. Every row in the table below is based on measurements taken on real objects:

Region Knot Measured density Source
Oushak (16th c., Ankara Vakıf Eserleri Museum, 15 rugs) Symmetrical (Gördes) 17–21 × 21–26 / dm² Özkavruk & İleri, Arış 7, 2012
Oushak (Ayaş mosques) Symmetrical (Turkish/Gördes) 30 × 30 per dm² Arış 22, 2023
Yağcıbedir (Sındırgı) Symmetrical — "The knots are Gördes (Turkish) knots" 735 – 1,232 knots/dm² Atiş, Arış 20-21, 2022
Taşpınar (Aksaray) Symmetrical (Turkish) TSE 1967 report: average 1,750 knots/dm²; today 30×35/dm² Arış 10, 2014
Hereke Symmetrical (Turkish/Gördes) Silk 10,000 knots/dm²; wool 3,600 knots/dm² Demirarslan, Art-e 25, 2020
Ladik (Konya) Traditionally symmetrical — production has now shifted to the Persian (asymmetrical) knot 40 × 50 Karpuz, Arış 7, 2012
Karapınar (Konya) Symmetrical (Turkish); all 28 examples wool No figures in the publication Arış 5, 2011

This table tells us two things. First: the overwhelming majority of measured Anatolian examples are symmetrically knotted. The name "Turkish knot" may be wrong, but measurement supports the view that the symmetrical knot is the predominant choice of the Anatolian tradition.

Second, and more interesting: the tradition is not fixed. Emine Karpuz's field survey of Ladik documents that even in a center where the symmetrical knot is "traditional," production has moved to the Persian knot. Karpuz describes this as a loss; what is being abandoned, in the study's words, is "a sound knot form." In Ladik the warp is now 100% cotton, and the weft wool is a blend of New Zealand, Karaman and merino.

Which knot a rug is tied with depends not on the region it comes from, but on the year and the workshop in which it was woven.

So Does Knot Type Affect Quality?

This is a place where we need to be honest, because the answer is far less certain than sales talk makes it sound.

The claim that "the symmetrical knot is stronger" does appear in an official Turkish source. The Ministry of National Education's MEGEP module Halıda Kalite Hesaplama (Calculating Rug Quality) states that rugs tied with the Turkish knot are sturdier, and that the Senneh knot cannot offer the same sturdiness because one end of the knot is left free. Karpuz's Ladik study reaches a similar assessment.

But neither of these is a textile-science study based on measurement. We could not find a peer-reviewed study, in Turkish or in English, that measures abrasion resistance, pile loss or breaking strength by knot type in handwoven rugs. The textile engineering literature in Türkiye is focused largely on machine-made rugs.

So we can report the claim as "this is how the Ministry module and field studies assess it"; we cannot say "it has been scientifically measured."

The claim that "the asymmetrical knot allows finer detail and curvilinear designs," on the other hand, has no source at all. It could not be verified in any museum record or in the peer-reviewed literature. It is a trade myth.

What can be defended is this: what determines the detail resolution of a rug is not knot type but knot density. The Encyclopaedia Iranica shows this indirectly but clearly, in the jufti example below.

Jufti: The Knot That Really Deserves Attention

If any structure in the knot debate truly affects quality, it is the jufti, also known as the "false knot." As Iranica defines it, jufti is a knot tied over more than two warps.

Its effect is documented too: jufti reduces production cost and material use, but it lowers knot density, coarsens the design and can lower quality. When you question a rug's value, the real question to ask is not whether the knot is "Turkish or Persian" but how many warps it is tied over.

How to Read a Knot Count — and How Not To

Knot density matters, but marketing it as a single number is misleading. The V&A's own Ardabil record puts it best:

"The knot count, as is typical, varies across the structure of the rug, so the figure given of 340 knots per square inch is an average."

— Victoria and Albert Museum, Ardabil rug record

The system used in Türkiye is different. According to the MEGEP module, quality is written as two numbers, such as 26×33, 30×30 or 40×40. The first number gives the count of warps/knots in 10 centimeters (width), the second the count of pile rows in 10 centimeters (length); their product is the number of knots per square decimeter. 26×33 = 858 knots/dm², or 85,800 knots per square meter.

The same module's classification: 250,000–500,000 knots/m² medium, 500,000–1,000,000 good, 1,000,000–4,000,000 rare, above 4,000,000 excellent. The 10,000 knots/dm² given for Hereke silk rugs in the table above corresponds to one million knots per square meter, in other words the start of the "rare" band.

A caution on comparison: in the Ardabil rug the warps are depressed. In asymmetrically knotted weaves with depressed warps, only one warp shows on the surface, so a count taken from the surface and a structural count can differ. Knot-count comparisons across different knot types and different warp levels should therefore not be taken at face value.

Why We Don't Talk About Some Regions

Milas, Kula, Döşemealtı, Isparta, Kayseri, Bünyan and Yahyalı are missing from the table above. The reason: we could not find an academic study that measures knot type and density for these regions and that we were able to verify. There are studies of motifs and composition; there are no technical measurements.

This is not our shortcoming; it is the state of the literature. Ebru Çatalkaya Gök's "Atlas of Publications on Handwoven Rugs in Türkiye," published in Arış, puts numbers on it: 1,992 scholarly studies were surveyed, and those naming a region came down to 559 (227 articles, 215 conference papers, 23 books, 94 theses). The most studied provinces are Isparta (42), Oushak (34), Konya (33), Milas (22) and Döşemealtı (19). By contrast, no academic study at all was found for 22 provinces.

And the most ironic part: we could not find measured structural data for Gördes itself. The studies of the region's rugs focus on motif and composition; knot type and density are not given. The town that gave its name to a knot type is one of the regions whose knot structure is least documented.

If a seller tells you "this knot is used in this region," you have every right to ask for the source. Regional rug traditions are legally defined in geographical indication registrations; in an upcoming article we will look at those records.

The name itself travels, too. The Art Institute of Chicago's rug no. 1992.385 has “a technique known as ‘Ghiordes knots’” in its materials field, and the Cleveland Museum of Art's rug no. 1964.125 has “Ghiordez knot: wool” in its technique field: two American institutions use the name of the same western Anatolian town in two different spellings, and in the records of rugs that were not even woven in Anatolia. We look at these records separately in our article “The Names That Shaped Rug Design”.

Anatolia's Own Pile Tradition

The world of knots is not just a symmetrical–asymmetrical pair. In a study published in Yedi, Sedef Acar and Ecem Tosun place Anatolian pile floor coverings within the international classification and, alongside knotted pile, define pulled weft wrapping, rod-looped wrapping, tülü and filikli structures. The study's finding is striking: "pile products other than rugs rarely appear in international publications."

A concrete example: Ülkü Küçükkurt's research on the filikli tülü of Seydiler in Afyonkarahisar, published in Milli Folklor, identifies three distinct knotting techniques across 11 woven examples and 20 informants: the Turkish knot, the ördekbaşı (duck-head) knot and the atlamalı (skipping) knot. In the tülü of Yıldızeli in Sivas, "a row of pile weaving is made using the Gördes knot technique"; the plain-weave band between the pile rows is typically 7–9 centimeters in the region.

So Anatolia's repertoire for making pile is far wider than two knots, and most of that repertoire is not yet in the international literature.

The Field's Own Self-Criticism

One last note: this scattered state of affairs is on scholars' own agenda as well. H. Örcün Barışta's study in Arış identifies the methodological scatter in research on rugs and flatweaves and proposes, as a remedy, a standard observation form separate from the museum inventory card. In other words, the reason there is no clear answer to "which knot in which region" is not only that the data has not been collected, but that what has been collected is not in a common format.

Expert View

"The question I hear most in the showroom is 'Is this a Turkish knot?' My answer these days is: 'A symmetrical knot, yes, but that name is the wrong name. I can tell you why, if you like.' Most customers are surprised at first, then interested. Because nobody has ever told them that the grandest Ottoman court rugs were tied with the 'Persian knot.' We don't have to sell a rug on the nationality of its knot; we have truer and more interesting things to tell."

— Mehmethan Özer, Halıcızade

Frequently Asked Questions

What is the difference between the Turkish knot and the Persian knot?

The structural difference is this: the symmetrical knot (commonly called Turkish or Gördes) wraps around two warp threads, and both of its ends come up between that pair of warps. In the asymmetrical knot (Persian or Senneh), one end comes up between the two warps and the other passes over a single warp, which gives the knot a direction: open to the left or open to the right.

Why is the name "Gördes knot" wrong?

According to a peer-reviewed study published in the journal Yedi in 2025, the name rests on a false etymology: the historical name of Gördes is Iulia Gordos, not "Ghiordium," and the town lies 330 kilometers from the Phrygian capital Gordion. In addition, Gördes was a secondary center in 1884, producing about a third as much as Oushak, and according to a 1917 Ottoman report its work was comparatively coarse.

Which knot were Ottoman court rugs tied with?

According to Bekir Deniz's study published in Sanat Tarihi Dergisi, Ottoman court rugs, unlike Anatolian rugs, were woven with the Persian (Senneh) knot. Oushak rugs, by contrast, were woven in wool with the Turkish knotting technique. The two techniques lived side by side in Anatolia.

Is the symmetrical knot really stronger?

The Ministry of National Education's MEGEP vocational training module and some field studies assess it that way. But there is no peer-reviewed textile-science study that measures abrasion, pile loss or breaking strength by knot type in handwoven rugs. The main factors that determine a rug's durability are yarn quality, knot density and weaving discipline.

Which knot allows a finer design?

There is no verifiable source for this distinction; it is a trade myth. What determines detail resolution is not knot type but knot density. The Encyclopaedia Iranica notes that the jufti knot, tied over more than two warps, lowers density and coarsens the design.

What is the knot density of Yağcıbedir rugs?

According to Didem Atiş's collection study published in Arış, the knots in Yağcıbedir rugs are Gördes (symmetrical) knots, and density was measured at between 735 and 1,232 knots per square decimeter. The warp, weft and pile are entirely wool.

Which knot is used in Hereke rugs, and how dense is it?

According to Deniz Demirarslan's study published in Art-e, Hereke rugs are woven with the symmetrical (Turkish/Gördes) knot. The measured density is 10,000 knots per square decimeter in silk rugs and 3,600 knots in wool rugs.

Does knot count alone indicate quality?

No. Even the V&A's record for the Ardabil rug states that the figure of 340 knots per square inch is an average and that the count varies across the structure of the rug. In addition, in weaves with depressed warps a count taken from the surface and a structural count can differ, so comparisons across different knot types should not be taken at face value.

Which is the oldest Anatolian rug?

The oldest Anatolian rugs we have are the eight pieces found in the Alâeddin Mosque in Konya in 1905–1906; one of them is registered at the Museum of Turkish and Islamic Arts as inv. 681: 13th century, 519 × 294 cm, Gördes knot, wool on wool. The 14th-century rug with animal figures at The Metropolitan Museum of Art (inv. 1990.61) is, in the museum's own words, one of three rugs of such an early date to survive complete.

If you'd like to browse our handwoven rugs: Handwoven rugs · Hereke rugs · Yağcıbedir rugs · Oushak rugs · Rare pieces

Related reading: Handwoven vs. Machine-Made Rugs · What Is a Loom-Woven Rug? · What Is a Hereke Rug? · The History of Turkish Rugs · Why Is a Handwoven Wool Rug the Most Valuable Kind of Rug?

Sources — Türkiye: Bahadır Öztürk, "Halıda Türk-Gördes Düğümü, İran-Sine Düğümü: Doğru Bilinen Yanlışlar", Yedi, no. 33 (2025), pp. 163–176, DOI 10.17484/yedi.1564035 · Bekir Deniz, "Anadolu-Türk Halı Sanatının Kaynakları", Sanat Tarihi Dergisi, XIV-1 (2005), pp. 79–103 · Elvan Özkavruk & Berna İleri, "Ankara Vakıf Eserleri Müzesi'nde Bulunan Bir Grup Etütlük Madalyonlu Uşak Halısının İncelenmesi", Arış, no. 7 (2012), pp. 86–101 · Didem Atiş, "Avukat Servet Örmen Koleksiyonundaki Yağcıbedir Halıları", Arış, no. 20-21 (2022), pp. 53–73 · "Taşpınar Örneklik Halıları", Arış, no. 10 (2014) · "Ayaş Camilerinde Uşak Halıları", Arış, no. 22 (2023) · Emine Karpuz, "Lâdik (Konya) Halıcılığının Günümüzdeki Durumu", Arış, no. 7 (2012), pp. 42–73 · "Karapınar (Konya) Yöresi Göbekli Halıları", Arış, no. 5 (2011) · Deniz Demirarslan, "İç Mekânda Kullanım ve Teknik Özellikler Açısından Hereke Halısı", Art-e Sanat Dergisi, vol. 13, no. 25 (2020), pp. 395–421 · Ebru Çatalkaya Gök, "Türkiye El Dokuma Halı Yayın Atlası", Arış (2019), pp. 42–51 · H. Örcün Barışta, "Halı ve Düz Dokuma Yaygılarla İlgili Araştırmaların Kazanımları, Sorunları…", Arış, no. 6 (2011), pp. 28–39 · "Osmanlı Halıcılığının I. Dünya Savaşı Yıllarındaki Durumuna Dair Rapor", Arış, no. 24 (2024) · "Osmanlı Devleti'nin Son Döneminde Uşak'ta Halıcılığın Makineleşme Süreci", Belleten, vol. 71, no. 260 (2007) · Sedef Acar & Ecem Tosun, "Anadolu Havlı Yer Yaygılarının Yapısal Açıdan İncelenmesi ve Bir Sınıflandırma Önerisi", Yedi, no. 18 (2017), pp. 61–73 · Ülkü Küçükkurt, "Afyonkarahisar'ın Seydiler Beldesi'nde Bir Kültür Mirası: Filikli Tülü Dokumaları", Milli Folklor, vol. 17, no. 132 (2021), pp. 286–300 · "Sivas Yıldızeli İlçesi Tülü Dokumalarının Özellikleri", Arış, no. 6 (2011) · A. G. Öngen & K. D. Atiş, "Anadolu Selçuklu Halı Tasarımında Zemin-Şekil İlişkisi", Hars Akademi, no. 5 (2022), pp. 1–16 · Republic of Türkiye Ministry of National Education, MEGEP — Halıda Kalite Hesaplama

Sources — international: Annette Ittig, "Carpets iii. Knotted-pile carpets: Techniques and structures", Encyclopaedia Iranica · Victoria and Albert Museum — The Ardabil Carpet, museum no. 272-1893 · Discover Islamic Art / TİEM — Konya Alâeddin Mosque rug, inv. 681 · The Metropolitan Museum of Art — Confronted Animal Rug, inv. 1990.61 · The Metropolitan Museum of Art — "Star Ushak" Carpet, inv. 58.63 · The Metropolitan Museum of Art — Carpet, inv. 2009.458.2 · Statens historiska museer — Marby rug, inv. 96378_HST · MAK Wien — KNOTS SYMMETRIC_ASYMMETRIC exhibition, 2002–2003

M Mehmethan Özer Koordinatör, Halıcızade

Looking for a rug like this?

Describe it in a sentence: the room, the colour, the size. We will narrow the collection for you.

Ask our specialists
A letter, once a month

New rugs and one story. Once a month, never more.

You can unsubscribe with one click.

Keep reading