Le fait urbain en Asie centrale préislamique : approche diachronique, approche synchronique, III : la crise urbaine et la réurbanisation (^IIIe-VIe s.), un processus général ? (continued) (8)

It's fair to say that the pace of urban decline differed between Sogdiana and the southern regions. In Sogdiana, urban decline was in fact the culmination of a long process that began after the end of Greek domination. Within Sogdiana itself, there were staggered rhythms: on the south-western edge, Erkurgan (Nakhshāb) was more important than Samarkand. This is a very interesting site, whose excavation has only been partially published [19] and should be revisited. There is still some doubt as to the chronology of the ramparts: for some, the huge 4.5 km outer wall is that inherited from the Achaemenid period (where its construction is attested by the archives of the satrap of Bactria) and the inner wall would testify to a later refocusing; for others, the opposite is true. In any case, even if the city was refocused, it was over a larger area than that occupied by Samarkand at the time. It remained prosperous until the ^6th century, with great monuments. It probably benefited from its geographical position between the Kushan empire and the Kangju confederation. Its decline was the result of an unfavorable position on the front line between the Hephtalites and Turks, then the Turks and Sassanids. Bactria experienced the same problem.

Once we've considered the crisis, we need to look for signs of a way out. In Sogdiana, from the ^3rd and 4th^centuries onwards, we can detect a first dynamic of military construction, but not yet truly urban, without being able to be more precise about the coincidences with the political upheavals (setting up defences before the Chionite invasion, or on the contrary establishing new masters of the land after this invasion?) In any case, a fairly sophisticated common military architecture appeared simultaneously throughout the "Kangju" political area, of three types: small square forts with round corner towers; larger four-lobed fortresses; and round forts with an orthogonal inner plan (a plan that appears to be Chorasmian in origin) [20]. For each of the three, a cult function was assumed, an idea that is strongly challenged today. At Punjikent, the first square-plan, one-hectare fortified settlement appeared in the 4th^century ; an identical settlement had been built in the previous century at the Paykand citadel, in a very specific context, a Sassanid encroachment attested by coins from Shāpur ^I.

From the^5th century onwards, the dynamism of military construction was confirmed, particularly well studied in Sogdiana, where it presented two aspects:

1. The spread, including in the high valleys, of forts of very similar appearance, continuing but modifying the first type of fortress of the previous period: square, 18 m square, with four square corner towers; thin walls; multiple archways.
2. A new phenomenon already mentioned is the planned foundation of new, medium-sized (8-10 hectares), fortified towns with orthogonal street networks, appearing at the foot of pre-existing citadels in the course of the^{5th century} . The best known are Pendjikent, Paykend, Rabinjan and Varakhsha; an identical layout is suspected in the ancient city of Bukhara. In Samarkand, the inner walls were built to confirm the new city limits, while the base of the citadel was raised by four meters (fig. 3). In this case, and in this case only, it is likely that this investment was linked to an identifiable political initiative: the installation of a branch of the Kidarites, documented by a seal mentioned above, followed by the northern branch of the Hephtalites. Elsewhere, these initiatives took place in the silence of texts. In Punjikent, there is archaeological evidence of a contribution by refugees from Tokharestān.

One factor that probably needs to be taken into account, although no source directly confirms it, is the enrichment of the Hephtalite state and society by Sassanid tribute. The impact of large-scale trade, which Sogdiana then definitively took under its control, seems to have a less direct causal link. However, a local commercial aspect could explain the highly professional, not very empirical, nature of the plans and architectural decorum: the chatelain, or his superior, chose the shell "from a catalog", while the interior spaces were improvised at the time of execution. The most sophisticated and best-preserved castles, a little later (^{6th-7thcenturies} ), are in Ustrushana: Chil'khudzhra (fig. 4), Urta-kurgan where burnt beams make it possible to restore a "lantern" ceiling very similar to those of the great houses of Punjikent and stone imitations in the caves of Bāmiyān. Behind all this we can guess a very solvent state or aristocratic demand, which would have been met by itinerant professionals, architects, or let's rather say site managers.

[16] M. Le Berre, Monuments pré-islamiques de l'Hindukush central, Paris, Éditions Recherche sur les civilisations, 1987.

[17] See the classic book by A. M. Belenitskij, I. B. Bentovich and O. G. Bol'shakov, Rannesrednevekovyj gorod Srednej Azii, Leningrad, 1973.

[18] The Khaldunian model of 100-120-year cycles does not yet seem to have been tested in Central Asia: disarmament of indigenous populations under fiscal pressure, delegation of military power to tribal groups alien to the founders of empires and which eventually rise to power, "squatting" of urban spaces at the end of the cycle, and so on. Yet there is no shortage of material for many periods, both pre-Islamic and Islamic.

[19] R. Sulejmanov, Drevnij Nakhshab, Samarkand, Tashkent, 2000.

[20] F. Grenet, "A view from Samarkand: the Chionite and Kidarite periods in the archaeology of Sogdiana", in M. Alram, D. Klimburg-Salter and M. Pfisterer (eds.), Coins, art and chronology, t. II, Vienna, Verlag der Österreichischen Akademie der Wissenschaften, 2010, pp. 267-281.