The Shape and the Name

Theodosiopolis changed hands with a regularity that must have exhausted its stonemasons: Byzantines, Arabs, Byzantines again, Georgians. As we have seen, David III Kuropalates took Manzikert and, with Gagik I of Armenia, destroyed Mamlan the Rawwadid—though not at Manzikert, as it happens, but near Archesh, in the district of Apahunikʿ. The coincidence with 1071 would have been very tidy indeed, had the sources been willing to cooperate.

Seventy-odd years later, at the same Manzikert, the rest was decided. In 1071 Alp Arslan, second sultan of the Great Seljuks, broke a Byzantine army so comprehensively that Anatolia ceased to be Byzantine by default. Theodosiopolis passed to one of his commanders, Saltuk, and became the seat of the Saltukids, the first Turkmen beylik in Anatolia.

The Seljuks were Oghuz Turks from the lower Syr Darya and the Aral steppe. On the way west they passed through the Iranian world and emerged with a chancery, a court vocabulary, a scholarly apparatus and—our concern here—a repertoire of architectural forms. In Erzurum these set into kümbets.

The word repays attention, because it contains the whole argument. Turkish kümbet comes from Persian گنبد gonbad, Middle Persian gumbad, “dome, vault.” Beyond that the trail forks. Standard etymological dictionaries treat it as inherited Iranian, from a root meaning “swelling, curve.” Sevan Nişanyan derives Persian gunbad from Aramaic-Syriac qubbtā, “dome”—the same source behind Arabic qubba and Turkish kubbe. The second reading belongs to one lexicographer and has not carried the field.

Two Roofs, One Building

Nor is a kümbet simply a tomb. Turkish usage distinguishes türbe, a domed mausoleum, from kümbet, which stands on a burial crypt, has a polygonal or cylindrical body, and is roofed with a dome inside and a cone or pyramid outside. A building, that is, with two entirely different roofs depending on which side of it one is standing.

Where the cone came from is a real dispute, and, as these disputes tend to be, a national one. The Persian pedigree is plain: the tomb towers of Iran, above all Gonbad-e Qābus of 1006/07. Turkish scholarship—the TDV İslâm Ansiklopedisi among others—prefers a different line: the kümbet appears in the Islamic world with the Turks, from Eastern Turkestan to Anatolia, and the cone remembers the tent. The Emir Saltuk kümbet obliges: the niches of its drum carry a bull, a snake, a bat and an eagle, figures that look very like the animal cycle of the Central Asian calendar. The word, meanwhile, remains obstinately Persian.

The Grandfather Who Took the Tomb

The Three Kümbets are what survives of a large Muslim cemetery that has otherwise disappeared. The largest is called the kümbet of Emir Saltuk. It bears no foundation inscription and is dated to the second half of the twelfth century, sometimes later—which raises a small difficulty, since Saltuk the founder died in 1102 or 1103 and can hardly be interred in a building put up two generations after him. Local scholarship assigns it instead to İzzeddin Saltuk II, who died in 1168 and had the misfortune of sharing a name with his grandfather.

The name, in other words, behaves exactly as the form does. Both men were İzzeddin, both were Saltuks, and a sixteenth-century Ottoman survey enters the tomb as that of “Sultan Melik Saltuk Gazi,” a title through which neither of them is now visible. We have met this before, in a different corner of the same century: one byname, Monomachos, doing duty for two men who never met. The difference is that Constantine and Vladimir at least kept their separate histories. The Saltuks did not: the grandfather drew his grandson’s monument to himself, by the same means that Persia drew the conical roof and the steppe drew the Persian word. Whoever is named loudest gets the credit. The two smaller kümbets escaped this only by never being named at all—fourteenth century, occupants unknown, described in the literature as statesmen, scholars or martyrs, which is the courteous way of saying that nobody has the faintest idea.

The best view of all this is from the museum. The old archaeological museum on Paşalar Caddesi was condemned as seismically unsound and demolished; the new one opened in August 2023 in the park beside the tombs themselves. One climbs the stairs past the Urartians, the Achaemenids and the Byzantines and emerges into a bright hall whose east windows look straight out at the Three Kümbets. The museum, in short, has moved house to be nearer its subject—a courtesy one wishes more institutions would extend.

Wall, Garden, Window

And nearer its vocabulary, too. The Persian layer in Turkish lies not in the higher literature but underfoot: duvar, “wall,” from dīvār; bahçe, “garden,” from bāġče, a little bāġ; çarşı, “market,” from čārsū, “four directions,” a crossroads; şehir, “city,” from šahr. It is much the arrangement English fell into after 1066, when the Normans handed over the vocabulary of masonry, vaults and towers and kept the buildings as well. The park the museum stands in is called Millet Bahçesi, half of which is Persian.

So, for that matter, is the window one sees the tombs through. Turkish pencere is Persian, but with reservations. The internet explains it confidently as penç, “five,” plus re, “way”: the fifth way out of a room with four walls. It is elegant, memorable and wrong. Nişanyan links pencere to Persian pancare and cautiously proposes bādgīr, “air-hole,” from bād, “wind,” and gīr, “catching”—cautiously being the operative word. The Persian origin is secure; the particular source is not; and the most popular explanation is the one nobody had until somebody needed one.

Through the wind-catcher, then, three domes—which arrived, as words and buildings often do, in the same consignment.