A palace celebration is in full swing. Women carry bowls, musicians play, attendants lean toward the newly married couple, and a staircase connects two richly furnished rooms. Under the crowd, two carpets compete for attention. One has a spacious network of stars and linked frames. The other is smaller in the picture, with a much denser, flower-like pattern. Both have red borders filled with white, letter-like ornament.

This is not a photograph of surviving rugs. It is a painting in a Persian manuscript made in Baghdad in 1396. Yet its floor coverings offer an unusually engaging way into an elusive chapter of carpet history: the Jalayirid period, before the celebrated carpets of the Safavid courts.

Where the Jalayirids fit in Persian carpet history

The Jalayirids ruled parts of Iraq and northwestern Iran after the fragmentation of Ilkhanid power in the fourteenth century. Baghdad and Tabriz became important centres of courtly painting and manuscript production. Artists carried forward earlier decorative traditions while developing the richly articulated interiors that make these books so absorbing to look at.

For carpets, however, the evidence is uneven. No physical rug securely identified as Jalayirid is available in the group examined here. The relevant manuscript paintings therefore preserve something that the surviving textiles cannot show directly: how patterned floor coverings appeared within an imagined courtly environment.

That distinction matters. A Persian-language manuscript painted in Baghdad belongs to the history of Persianate art, but it does not establish that every rug shown was woven in Iran. Objects, patterns and artists moved across political boundaries. Here, “Jalayirid carpets” means carpets represented in paintings associated with that artistic setting, not a securely identified group of excavated or surviving weavings.

The illustrations span several kinds of book, including Kalila and Dimna, the poems of Khvaju Kirmani, and Nizami’s Khusraw and Shirin. Together they offer more than one spectacular interior. They reveal recurring ways of organizing a carpet, framing its edges and placing it beneath people.

A carpet built from repetition, not one central medallion

Many of these painted carpets depend on a repeat unit: a small arrangement of lines and shapes that can extend across the field. Four-armed interlaces connect to neighbouring forms. Flowers sit inside geometric compartments. Stars recur at intersections. The eye moves across a network rather than being pulled toward a single, dominant centre.

The Persian term vagireh describes this repeating unit. It is helpful to think of it as the building block of the design, not necessarily as one isolated motif. A flower alone may not be the full unit; the connecting frame and the spaces around it can be just as important.

This differs from the familiar composition in which one large medallion anchors the middle of a rug. Several early paintings do show repeated octagonal compartments, but multiple repeated compartments are not the same arrangement as a single central medallion. The distinction is explained more fully in our guide to Persian rug medallions.

Nor does a geometric framework exclude flowers. In the wedding scene, floral and star-like forms already occupy angular networks. “Geometric” and “floral” are useful descriptions of different aspects of the same design, not necessarily separate stages through which every carpet must pass.

Two carpets in one celebration

Look first at the larger carpet on the left of the wedding painting. Its field is divided by pale connecting lines, with pointed stars and smaller ornaments distributed through the framework. Bodies and vessels obscure much of it, but enough remains visible to follow the repeat across several intervals.

Its broad red border gives the crowded group a definite enclosure. Beyond that border, blue patterned surfaces belong to the surrounding interior. The carpet is not simply another patch of decoration: it marks a particular area within a room that is already dense with ornament.

On the right, the carpet beneath the attendants has a tighter rhythm. Small, many-pointed forms crowd together, creating a finer-grained surface. Its script-like border is clearly related to the left-hand carpet’s, but the fields are not interchangeable. One room can accommodate related framing devices and quite different internal patterns.

A third covering is partly visible beneath the raised seat on the right. Cushions and the seated figure hide most of its field. It is tempting to complete the missing design in the mind, but the painting does not supply enough information to do that confidently.

The differences also affect how the scene reads. The left carpet gathers a busy cluster of guests into one broad area. The smaller right-hand area has its own edge and concentration of activity. Carpets help separate social spaces without walls, while their related borders keep the whole interior visually connected.

The border that looks like writing

The white ornament around the wedding carpets resembles angular lettering. This is usually described as a pseudo-Kufic border: decoration recalling the strong upright strokes, turns and interconnections of Kufic script. “Pseudo” matters because the resemblance does not mean that the band contains a readable sentence.

A genuinely legible inscription can carry a name, prayer or dedication. A script-like ornament can instead borrow the visual authority of writing while functioning as a repeat. Unless the letters can be read securely, translating such a border would produce a message that the image has not established.

These borders recur across the manuscript examples, but they are not all identical. Their proportions, connecting elements and treatment of corners differ. Some are narrow and tightly organized; others give their shapes more room. Smaller guard bands can introduce another rhythm between the main border and the field.

Similarly, an ornamental “endless knot” is a drawn interlace without an obvious free end. It is not evidence for the type of knot used to attach pile to a carpet’s foundation. The same word can refer to a visual motif and a weaving operation, but the two should not be confused.

When the border begins to curve

Another painting in the 1396 Khvaju manuscript shows the Sasanian ruler Nushirvan with his counsellor Buzurgmihr. Although the subject belongs to an earlier historical world, the illustrated furnishings are part of the painter’s own visual vocabulary. The picture is not a record of sixth-century carpets.

The orange carpet has a field of linked angular forms, flowers and stars. Its brown border is especially interesting. At first it belongs to the same broad family as the letter-like bands of the wedding scene. Seen closer, however, parts of the ornament branch and curve rather than remaining rigidly straight.

This makes the border a useful place to observe change. A stable repeated field can coexist with a less angular surrounding band. The difference need not involve replacing the whole design: a relatively small part of the composition can carry a new kind of movement.

It would be too strong to call this the first invention of floral carpet borders, or to conclude that the artist personally designed a rug for a weaving workshop. What survives is a painted choice. Its value lies in showing curving ornament alongside established geometric organization, within one carefully composed image.

Carpets, cushions and covers are not the same thing

These interiors are layered. A broad carpet may support a cushion, a raised seat, another textile or a figure’s clothing. An exposed strip at the edge can be all that remains visible of the lower covering. Without noticing the overlap, it is easy to assign a cushion’s design to the carpet beneath it.

The distinction is particularly clear in Farhad before Shirin, a folio from a manuscript of Nizami’s romance dated approximately 1400. Shirin occupies the centre, flanked by other seated groups. Related carpet patterns extend beneath the three groups, while additional coverings and cushions interrupt their surfaces.

The outer carpets appear closely related in colour, while the central arrangement differs. This suggests coordinated furnishing within the painted interior. It does not prove that two physical rugs were woven as a matched pair. A painter could repeat a pattern for compositional unity, just as actual furnishings could share a design.

Nor can the painted surface settle whether a covering was pile-woven or flatwoven. A striped textile might recall a jajim or kilim, but resemblance is not structural analysis. The image gives us placement and visible pattern more readily than it gives us technique.

A shared language of ornament

Look away from the floor for a moment. The walls, doors, archways and clothing also carry repeated patterns. Stars, linked compartments and vegetal ornament circulate across the interior. The carpets make sense within a larger world of decorated surfaces, rather than appearing as a separate art with an entirely independent vocabulary.

This relationship helps explain why a carpet repeat can resemble architectural ornament without being a literal copy of a particular tiled wall. Geometry offers arrangements that work on many surfaces. Painters can also coordinate those surfaces to make the whole setting feel ordered.

The direction of influence is harder to establish. A resemblance alone cannot tell us whether a textile designer borrowed from architecture, an architectural designer borrowed from textiles, or both drew on a broader repertoire. It is more convincing to describe the shared forms than to invent a one-way origin story.

What these paintings let us recover

The strongest evidence is close to what we can actually see: repeated fields, distinct borders, layered coverings, contrasts of pattern density and the placement of people on carpets. The pictures also show how a rug could define a ceremonial or conversational space within a larger interior.

The weakest conclusions are those that require information the painter could not supply. Pigment is not a dye sample. A person’s painted height is not a measuring tape. A detailed border does not reveal pile density, fibre content or workshop address. Even an apparently plain field may reflect simplification rather than a genuinely unpatterned rug.

Dates need equal care. One Kalila and Dimna manuscript has a text dated 1343, while its illustrations are assigned to a later period. The date of copying a book cannot automatically date every image or design inside it.

With those limits in mind, the Jalayirid paintings become more interesting, not less. They preserve a varied visual world before the better-known surviving Safavid carpets, including the narrative-rich Sanguszko carpets. Their fields offer repetition, their edges offer variation, and their layered interiors show carpets actively shaping a room. We cannot recover the lost wool from a painted page, but we can recover something of the choices that made a carpet worth looking at.

Sources

• Davood Shadlou. Aesthetic Analysis of Carpet Designs in Illustrated Manuscripts of the Jalayerid Period (737–814 AH). 2026.
• Eleanor G. Sims. Carpets viii. The Il-khanid and Timurid Periods. 1990, updated 2017.
• British Library. Khvaju Kirmani manuscript, Add MS 18113, 1396, folios 45v and 91r.
• Smithsonian’s National Museum of Asian Art. Farhad brought before Shirin, F1931.34, circa 1400.