What is Kufi?

This overview is mainly intended to give students and collectors a general understanding of the history and features of the Kufi scripts. At the same time, I feel it is essential to closely follow this with a more in-depth discussion, and to dispel some myths and misconceptions that are running amok around the web. For this reason there is a Short Version offering the essential history, followed by an optional, more technical Closer Look.

The Short Version

Round scripts of the late tradition (left) vs. Kufi of the early tradition (right)

The history of Arabic calligraphy is made up of two traditions. We are most familiar with today are the cursive scripts that belong to the later tradition: Naskh, Thuluth, Nastaliq, Diwani etc. But the earlier tradition, which revolutionised Arabic writing and turned it into an art, is the Kufi (or Kufic) tradition: a family of scripts underpinned by a shared, strongly geometric structure and precise grid. How did Kufi arise, and why did it die out?


6th-century Arabic inscriptions

Old Arabic was originally written in Nabataean, but by the 6th century, under the influence of Syriac, the letters had evolved into a form we can recognise. There was no great need for writing, however, until the Qur’an was revealed, and particularly after the Prophet’s death in 632: to preserve the sacred text became a matter of urgency. The earliest known Qur’ans are written in the secular hand that was then available, where “hand” means each scribe wrote in their own handwriting. What we now call Hijazi (and Islamic sources called jazm) is not properly speaking a calligraphic script: it has no formal rules to govern its design and unify its appearance from one scribe to another. Looking at these early examples, one can see that all the focus was on efficiency (recording the information) with hardly any aesthetic considerations.

An example of so-called Hijazi text

But preserving the text in writing was not enough: it needed to be beautiful, worthy of the divine message it carried. In Islam, Beauty and Majesty are the two faces of God, and a well-known hadith says: “God is beautiful and He loves beauty.” Within a few short decades, this consideration would give rise to the Quranic scripts reserved for the writing of the holy book, and to the art of Arabic calligraphy.

A New Identity

At the very end of the 7th century, the reign of caliph Abd al-Malik saw a complete rebranding (to use a modern but appropriate word) of the Islamic empire to give it its own distinctive identity. Arabic became the official language, new coinage was minted, and the first distinctly Islamic monument, the Dome of the Rock, was built in Jerusalem. Inside the Dome, a mosaic inscription displays the earliest known example of Kufi, possibly the prototype for all that followed (I discuss this more below, under Manuscript Kufic). There we can see an elegant mature script, deliberately proportioned, with a consistency of appearance thanks to a small number of shapes that echo each other from letter to letter. There was no transition period, no gradual evolution between the primitive jazm and this fully developed Kufi: it appeared virtually overnight, so that we can be sure the script was redesigned purposefully. Other inscriptions above the north and east doors, engraved in copper, show this same transformation, with the metallic support allowing for even more pronounced geometry in the letters, some of which are based on perfect circles. In both cases, there is a clear underlying grid onto which the letters are mapped. Kufi was the first Arabic script to be made consciously beautiful.

Details of the mosaic and one of the copper plates at the Dome of the Rock

From then on, all Qur’anic manuscripts closely followed suit. Formal training was now clearly taking place because Manuscript Kufic became the norm, even if that didn’t exclude a small degree of stylistic and/or personal variation. A template called mastara was now used so the calligraphers (no longer mere scribes) wrote in straight lines. The system of proportions binding the letters extended to the whole page, defining the relationships between lines, margins and page size. The binding format also changed to landscape, possibly chosen to stand out from the other book traditions (which used a portrait or square format), but also more suited to the horizontality of the script.

While Manuscript Kufic (كوفي مصحفي), in its diverse iterations, was the product of pen and ink on parchment, the new script also developed into the many constructed styles (كوفي هندسي) designed to be rendered in hard materials, sometimes on a large scale: bricks, wood, metal, stucco and so on. We also see ceramic-specific variations, dictated by the material requirements of applying glaze, and others on textiles.

End of an Era

Kufi was a purely Qur’anic script, the only one for 300 years, never used for anything else and indeed not suited to anything else. Its function was beauty and praise, not practicality. The laborious writing of it was an act of devotion that used up as much parchment and precious pigments as could be afforded (within the limits of appropriate austerity). Meanwhile, it was cursive Arabic that was used for everyday needs, and as the empire grew, so did its paperwork. The scribal profession grew in importance and developed its own “round scripts”, quick to write, clear to read, and highly functional. As the social landscape changed, so did the relationship to writing. Literacy rose, but this likely only applied to the widely available round scripts, and as the demand for personal copies of the Qur’an rose, these pocket versions were written in those more familiar forms of writing. Society moved on from the ethos and original burst of inspiration that gave birth to Kufi, and new forms of round scripts were designed to rival it as Qur’anic scripts. From the 10th, they gradually took over, and by the 12th century the early tradition had vanished completely: as they were replaced, the old Qur’ans were put away in mosque caches, never to be seen again. In this way Manuscript Kufic completely vanished from sight and memory for centuries. Only the architectural inscriptions remained visible and those constructed Kufi styles continued being used, but in the absence of a living line of transmission, they soon stopped being understood on their own terms and fell into decadence.

A Closer Look

It’s important to note, before anything else, that all names attached to specific scripts were coined long after the scripts were used, and while they are useful shorthands, we must not be too attached to these names or read meaning into them. There are many outdated myths and misunderstandings that get endlessly perpetuated online, top of the list being the idea that Kufi originated in the city of Kufa. The only reason for this myth is that a modern researcher misinterpreted the term as found in the writings of 10th-century chronicler al-Nadim, and it’s been impossible to get rid of this notion ever since.

Other, less misleading ways to refer to these scripts have been proposed, such as “Abbasid scripts” (because that is the period in which they flourished, and they declined together) or “rectilinear scripts” (a misnomer, as they are not made up of straight lines). I’m content to keep the simpler word Kufi, which is now too widespread to retract, as long as it is clearly defined and re-focused. (On a personal level I’m trying to avoid “Kufic” because there’s no good reason for this ending and it sticks in my throat. After all, we don’t say “Hijazic” or “Maghribic”.)

Kufi is not one specific style, but a family of scripts that all share the same underlying principles and are mapped on a similar grid (unlike the script, the gris is rectilinear). They constitute the early tradition of Arabic calligraphy, which is fundamentally different to the later tradition that is made up of the round scripts. The latter is known in Arabic as الخط المنسوب , usually translated as “proportioned writing”, but as they were neither the only nor the first to be proportioned, this is better translated as “standardised writing”. In contrast, the principles that guide Kufic calligraphy are more internal, and the outer appearance of the script can support much more variation without turning into something else. It is misguided to approach Kufi from the point of view of the later tradition and to project aspects of round scripts onto it, as too many do. For instance, the famous device of using the calligraphic dot as a basis was devised to formalise the round scripts, and is entirely irrelevant to the earlier tradition. Kufi needs to be studied and practiced on its own terms, and for later-tradition calligraphers that may require some unlearning first.

Manuscript Kufi

Manuscript Kufi variant known as D1, in my opinion the highest peak of this style, the most accomplished, with a fully developed script grammar.

I began using the term “Manuscript Kufi” on my own initiative to refer to the styles written on parchment (as opposed to architectural styles), but it happens to reflect the existing Arabic term كوفي مصحفي. A number of stylistic variations exist, that have been catalogued by François Déroche in a system of groups and subgroups designated by letters and numbers (e.g. B1, D4). This helpful system is the norm today, but it’s constantly being refined, and at this time it doesn’t take into account differences that may be due to nothing more than the scale of the text, or differences I would call cosmetic. By this I mean that if a scribe, through long habit, cuts the corners of ن and writes it more rounded than usual, this will affect several letters and the overall texture of the page, but it’s not a distinct style, just a personal quirk. Kufi was never at any point super-codified and strictly formulaic in the way the round scripts were designed to be. Hence the apparent proliferation of styles: individual variation was perfectly tolerated, as long as the essential rasm (how the letters are written) was adhered to.

An egregious exemple of gestural variation. For whatever reason, the pen was held very flat angle instead of the required 45º. The result appears like a totally different style, but it isn’t; there’s nothing novel in the shape of the letters proper.

Although it’s handwritten, Kufi is far from cursive. It is a script that aspires to retain an architectural quality on the page, and in its most accomplished styles such as D1, a great deal of the writing time is spent touching up and re-shaping parts of the letter away from calligraphic strokes. Right-angle corners are added where the pen would leave a 45º angle, and curves are filled in where there would only be the flat of the pen. The spaces inside and between connected letters is made as narrow as possible, even if this means filling it in, so that it looks like the white lines are cut out of the black ink, as they would be out of stone or similar. Less advanced styles are specifically marked by the absence of such finitions*. It is as if the particular style seen in the Dome of the Rock mosaics, where the medium naturally called for these shapes rather than calligraphic strokes, was the prototype that always remained the ideal model to emulate, regardless of the amount of extra work involved. The result is an unusually constructed calligraphic script, but also an extraordinary stillness of the page, where all movement is cancelled and fullness is held in balance with emptiness.

*(This is not to dismiss “less advanced” styles. The creativity, experimentation and sheer joy of making that characterise early manuscripts are an example to us all.)

In practice, this result requires tremendous self-discipline and skill on the part of the calligrapher, and yes, inner stillness that is not attained overnight. Even as Manuscript Kufic reached its zenith at the hands of calligraphy masters, it was already starting to change as more junior, more time-pressured calligraphers allowed themselves to take shortcuts that would gradually transform the script. These shortcuts, which start out as simple gestural changes, can be observed in some surviving manuscripts. They are very interesting because they are the exact same shortcuts that I can see in students just beginning to learn this script, and they are very simply the natural tendencies of the hand, which looks for more natural ways for the pen to travel from point A to point B. Much of Kufi apprenticeship is to overcome these tendencies in order to make less intuitive, more architectural strokes. Before the self-discipline is formed, or after it has been relaxed, the shortcuts reassert themselves, and this is what we’re seeing here.

And as, in the East, a much less formal form of Kufic became popular for secular texts, enshrining these changes in daily use, they lead in a natural progression to the script we know as Eastern Kufi.

The style in this undated manuscript (BNF Arabe 377) appears to be exactly halfway between Manuscript and Eastern Kufi. The tightening of spaces saves a lot of parchment and therefore money. The rounded letters are now written pointed, which is both easier and quicker: could this have been the original cause of the change? This type of transitional style needs to be studied in detail for more insights.

Mashriqi (Eastern Kufi)

Fully formalised Mashriqi (left), and the variant known as “Qarmatian” (right)

It is in Persian lands that we find the emergence of more casual, dynamic Kufi formalised into a Qur’anic script, and there that it flourished from the late 11th to the 12th century. So even though it was not entirely reserved to that part of the world, we are on safe ground calling it Eastern or Persian Kufi, as it’s also known. It’s frequently paired with lush ornamentation that betrays quite a different stylistic sensitivity from the majestic austerity, relieved only by geometric carpet pages, of Abbasid manuscripts. In Déroche’s system it is called “New Style”, which says nothing and lumps the many variations into only two subgroups (I would be more inclined to make it a group H, with G grouping the curious in-between styles). Another proposed name that is both bizarre and erroneous is “broken cursive”, which rests on the broken assumption that it derives from a round script. In reality it is firmly Kufi, in an extreme variant of the Manuscript form that nevertheless keeps to the same grid. To put it differently: If you understand the structure and script grammar of the former, the switch is merely a question of translating the forms, with no need to relearn from the ground up.

The so-called Qarmatian/Karmatian script, famous for its striking angular appearance against a lush background of palmettes and scrollwork, is actually known from the surviving folios of a single manuscript. It’s not so much a script of its own as a Mashriqi variant, where small changes on the main script have maximum visual. Perhaps it was specifically designed for this particular copy of the Qur’an, the size of which makes it clear that whoever commissioned it was very important and/or very wealthy. In any case, the arbitrarily-attached label “Qarmatian” has long been discredited, and attempting to derive any significance from it is a waste of time.

Maghribi (Western Kufi)

Maghribi is quite a diverse family in its own right, with a journey very similar to Mashriqi: like the latter, it derived from Manuscript Kufi, transformed by the need for speedier writing of secular texts, before one or more variants were re-formalised enough to be suitable Qur’anic scripts. But here the transformation consisted in a softening of angles into curves, written with pointed pens (cut in the softer local reed) rather than the hard-edged Eastern chisels. Also, unlike Mashriqi, Maghribi did not disappear after a brief flowering: it is to this day the native script of North Africa, where the round scripts never took hold. This is the most cursive of Kufi styles, but the grid, as well as the letters’ ascendency, can still be detected behind the extravagant curves. One particularity of Maghribi that must be noted is that ق and ف are pointed differently: ق has only one dot above, while ف has its dot below the line.

Several names are used to designate different variations of Maghribi, but not always consistently, especially when you compare Arabic and Western sources. For now I’m leaving them out.

Square Kufi

Also known as the Masonry Script or Kufi Banna’i, This most fashionable of styles appears deceptively simple, and as a result is routinely seized upon and horribly mangled in contemporary attempts. My advice to students is to ignore any modern design they may encounter, and stick strictly to historical ones, until they have the necessary discernment. Despite its extremely stylised nature and the unusual positions in which letters can end up, Square Kufi still adheres faithfully to the essential rules of the script. Without breaking them, it must also fill the space in a precise way, making it one of the most challenging styles to practice—unless you use one of the permissible “cheats” that past artisans had at their disposal. Square Kufi first appeared, in primitive form, in Ghazna, Afghanistan, between the late 11th century and the early 12th. It’s only a century later, under Mongol rule, that its popularity was boosted—by the official seal of the Great Khan, of all things. Its design in Phagspa script was so similar that it instantly imbued Square Kufic with an aura of authority. From there, it underwent a twofold evolution: on the one hand, the monumental style spread westward from Afghanistan and became especially prominent in Persia. On the other, designs copied on paper, took the quick route to Egypt, where the style took root and started spreading eastward, typically on smaller objects, until they met in the middle.

Notable about this script is that, at its height, its subject matter was purely religious, while its legibility was at its lowest. Longer texts were typically written in a spiral, so that most of the text was sideways are upside-down, and the viewer was evidently not intended to read the inscription at all. Given the geographical location, they may not have been expected to necessarily speak or read Arabic in the first place. The visual impact and presence of the text were probably all that was needed for it to fulfill its function, much like talismans that are worn hidden in a sewn pouch, never to be read, or prayers buried in the foundations of a house.

Other Styles

The Kufi family is very far from being fully catalogued, and as I hinted earlier, it would be misleading to attempt to categorise every slightly different instance separately. I’ll add to this section as clear groupings arise. Note, however, that floriated, foliated, knotted and so on, are not styles of Kufi, despite being too often listed as such. They are decorative treatments that are applied to the rasm, and can be stripped from it without losing the text. They are basically medieval Zoom filters.

Notes for Readers of Modern Arabic

Modern readers often don’t realise that, in the early days of Kufi, Arabic spelling was quite different from what it is today. One must be prepared to encounter some archaic features and not imagine they are mistakes.

  • The (lack of) use of Alif: Medial Alif, representing the long sound â inside a word, was not used at all in pre-Islamic times, and took a long time to become the norm. In early texts, the sound is either implied, or carried by a ي or و. This is why in early Qur’ans, “prayer” is written صلوة as opposed to modern صلاة, and اله is written اليه. It’s also why الله and هذا are still spelled this way instead of the more logical اللاه and هاذا —such common words would have been too ingrained to ever be “updated”. Alif al-wiqâya (third person plural ending, e.g. فعلوا) can also be absent. This omission of Alif is referred to as scriptio defectiva, as opposed to today’s scriptio plena.
  • Final Ta: ة did not originally exist, and a ت was used instead: for instance “year” was spelled سنت before it became سنة. At some point this was replaced with ه without pointing. The pointing that gave birth to the ة (which doesn’t even have full-letter status) seems to have been a form of ihmâl, a reminder of its dual pronunciation (ه or ت depending on what follows it).

Other aspects of the script that have completely changed since then:

  • Diacritics (تعجيم): It’s completely untrue that the pointing system differentiating letters didn’t exist before the 8th century—it is already present in the earliest Hijazi examples, only it’s used selectively at first. Gradually, at the same time as other optional features become ingrained, full diacritics become the norm in manuscripts (unlike in ornamental and architectural styles, where they are always left out). This hard-to-kill myth is closely entwined with the equally mythical idea that pointing was created some time after the Islamic expansion, so that foreigners could read Arabic, because the word for “non-Arab, barbarian” is supposedly at the root of تعجيم. It isn’t! (And knowing that pointing was already used in pre-Islamic texts, this interpretation falls apart entirely.) The etymology derives from عجم “pip, fruit kernel”, referring to the shape of the diacritics: little dashes, not the dots we know today. Ta’jîm simply means “pipping” the text.
  • Vocalisation (حركات): In its earliest form, it consisted of red dots differentiated solely by placement, very typical of Manuscript Kufic. Other dot colours indicated alternative readings. Without going into details, Mashriqi and Maghribi systems start to look like today’s, but that’s never completely achieved until the later tradition.
  • Numerals: The Indian system of numerals was imported into Arabic in the 9th century. They are not native to the script and were never universally adopted: in the Mashriq they became the Eastern Arabic numerals (١٢٣٤٥) but in the Maghrib they evolved differently, into the North African Arabic numerals (12345), which Europe imported via Andalusia. Prior to this, the letters themselves were the native numerals: ا to ط stood for 1 to 9, ي to ص for 10 to 90, etc. The numeral sequence is remembered in the mnemonic sentence ابجد هوز حطي كلمن سعفص قرشت ثخذ ضظغ. This system survived in some specialised applications such as astrolabes, long after the Indian system prevailed.

Further Reading

Birth and early days of Kufi:

George, Alain. The Rise of Islamic Calligraphy. London : Saqi, 2010.

Déroche, François. Qur’ans of the Umeyyads, A First Overview. Brill, 2014.

Cataloguing Manuscript Kufi variants:

Déroche, François. The Abbasid Tradition: Qur’ans of the 8th to the 10th centuries AD. London: Nour Foundation in association with Azmimuth and Oxford University Press, c1992.


Saint Laurent, Béatrice. “The Identification of a Magnificent Koran Manuscript”, Les Manuscrits du Moyen-Orient, Varia Turcica, ed. François Déroche, Bibliothèque Nationale de France and Institut Français d’Etudes Anatoliennes d’Istanbul (Istanbul, Paris, 1989), vol. VIII, pp. 115-123, p.115, note 2.


O. Houdas, “Essai sur l’écriture maghrebine”, in Nouveaux mélanges orientaux, IIe série vol. xix., Publications des Langues Vivantes Orientales (Paris 1886)

Afa, Omar and Mohammad al-Maghrâwi. Al-Khaṭṭ al-Maghribī: tārīkh wa-wāqiʿ wa-āfāq الخط المغربي : تاريخ و واقع و آفاق . Casablanca : Manshūrāt Wizārat al-Awqāf wa-al-Shuʾūn al-Islāmīyah, 2007

van den Boogert, N. Some Notes on Maghribi Script. Manuscripts of the Middle East (MME) 4 (1989): 30-43.

Square Kufi:

Majeed, Tehnyat. The Phenomenon of the Square Kufic Script: The Cases of Ilkhanid Isfahan and Bahri Mamluk Cairo. Thesis, St. Hugh’s College, 2006.

al-Naqsh, Maḥmūd Māhir. Khaṭṭ-i bannāyī. Tihrān : Surūsh, 1370 [1991 or 1992].

For a fuller list of sources see my Bibliography.