Edinburgh Research Explorer

An Imperial Frontier of the Sasanian Empire: further fieldwork at the Great Wall of Gorgan Citation for published version: Sauer, E, Omrani Rekavadi, H, Wilkinson, T, Safari Tamak, E, Ainslie, R, Mahmoudi, M, Griffiths, S, Ershadi, M, Jansen Van Rensburg, J, Fattahi, M, Ratcliffe, J, Nokandeh, J, Nazifi, A, Thomas, R, Gale, R & Hoffmann, B 2007, 'An Imperial Frontier of the Sasanian Empire: further fieldwork at the Great Wall of Gorgan', Journal of the British Institute of Persian Studies, vol. 45, pp. 95-136.

Link: Link to publication record in Edinburgh Research Explorer

Document Version: Publisher's PDF, also known as Version of record

Published In: Journal of the British Institute of Persian Studies

Publisher Rights Statement: © Sauer, E., Omrani Rekavadi, H., Wilkinson, T., Safari Tamak, E., Ainslie, R., Mahmoudi, M., Griffiths, S., Ershadi, M., Jansen Van Rensburg, J., Fattahi, M., Ratcliffe, J., Nokandeh, J., Nazifi, A., Thomas, R., Gale, R., & Hoffmann, B. (2007). An Imperial Frontier of the Sasanian Empire: further fieldwork at the Great Wall of Gorgan. Iran : Journal of the British Institute of Persian Studies., 45, 95-136

General rights Copyright for the publications made accessible via the Edinburgh Research Explorer is retained by the author(s) and / or other copyright owners and it is a condition of accessing these publications that users recognise and abide by the legal requirements associated with these rights.

Take down policy The University of Edinburgh has made every reasonable effort to ensure that Edinburgh Research Explorer content complies with UK legislation. If you believe that the public display of this file breaches copyright please contact [email protected] providing details, and we will remove access to the work immediately and investigate your claim.

Download date: 04. Oct. 2021 British Institute of Persian Studies

AN IMPERIAL FRONTIER OF THE SASANIAN EMPIRE: FURTHER FIELDWORK AT THE GREAT WALL OF GORGAN Author(s): Hamid Omrani Rekavandi, Eberhard W. Sauer, Tony Wilkinson, Esmail Safari Tamak, Roger Ainslie, Majid Mahmoudi, Seren Griffiths, Mohammad Ershadi, Julian Jansen Van Rensburg, Morteza Fattahi, James Ratcliffe, Jebrael Nokandeh, Amin Nazifi, Richard Thomas, Rowena Gale and Birgitta Hoffmann Source: Iran, Vol. 45 (2007), pp. 95-136 Published by: British Institute of Persian Studies Stable URL: http://www.jstor.org/stable/25651414 . Accessed: 16/12/2013 07:29

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

.

British Institute of Persian Studies is collaborating with JSTOR to digitize, preserve and extend access to Iran.

http://www.jstor.org

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions AN IMPERIAL FRONTIER OF THE SASANIAN EMPIRE: FURTHER FIELDWORK AT THE GREAT WALL OF GORGAN

By Hamid Omrani Rekavandi (HO), EberhardW. Sauer (EWS), TonyWilkinson (TW), Esmail Safari Tamak (EST), Roger Ainslie (RA), Majid Mahmoudi (MM), Seren Griffiths(SG), Mohammad Ershadi (ME), Julian JansenVan Rensburg (JJ), Morteza Fattahi (MF), JamesRatcliffe (JR), JebraelNokandeh (JN),Amin Nazifi (AN), Richard Thomas (RT), Rowena Gale (RG) and BirgittaHoffmann (BH)

Abstract The 2006 season has yielded significantnew insightsinto the Great Wall ofGorgan's relationto landscape features and settlement,notably thedivision of theassociated complexwater supply system into sectors.The westernmost part of thewall is buried deeply beneath sedimentsfrom a past transgressionof theCaspian Sea. An unexpectedly high numberof brickkilns, of standardiseddesign, shed lighton themanner of thewall's construction.Geophysical survey,satellite images and excavations have established that some or all of the associated fortswere densely occupied with buildings, thoughtto be barracks, suggestinga strongmilitary garrison.

Keywords Forts; landscapeArchaeology; linearbarriers; Sasanians; water supply.

I. INTRODUCTION II.THE GORGANWALL IN ITS LANDSCAPE (HO, EWS, TW & EST) (TW&HO)

The second season of the joint project, between the One of the more significant results of the 2005 field Iranian Cultural Heritage and Tourism Organisation season was that a system of water supply had been (ICHTO) and the Universities of Edinburgh and incorporated into aspects of the wall design and con Durham, to explore the linear barriers of northern Iran, struction. Consequently, one of the main aims of the near modern Gorgan and Gonbad-e Kavus, took place 2006 field work by T.J. Wilkinson and H. Omrani in September 2006. Our aims included obtaining Rekavandi was to find out more about related hydraulic further dating evidence for the construction of theGreat installations as well as to understand how the Gorgan Wall of Gorgan and the duration of occupation of the Wall related to both the natural and cultural landscapes associated forts.We equally hoped to gain further of the region. These studies,which elaborate on the con insights into the extent, purpose and design of the siderable amount of earlier work by Jebrael Nokandeh associated water supply system, the organisation of and Hamid Omrani Rekavandi, covered most of the brick for the as as production wall, well the location of length of thewall from near Fort 2 in the east, to beyond the wall's terminals in the west and east. The 2005 Fort 33 near theCaspian Sea in thewest. season had yielded little tangible evidence for the nature of the occupation of the forts, and the size of their garrisons and substantial furtherprogress here in III.WATER SUPPLY 2006 was a major desideratum. Unlike 2005, all our (TW&HO) efforts in the short 2006 season were concentrated on theGreat Wall of and we carried out no Gorgan further The 2005 field season had recorded threemajor canals work on theWall of Tammishe. that led water from theGorgan River, and in 2006 two

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions 96 JOURNAL OF PERSIAN STUDIES

20 io is17 ]U}-<^y jV4 Gunbadr?\ ym\ \ \ ^6^\_A^ i-Qabus\ 2^^X^^^^ - 23/ \ / \ GorganWall ""1 \ CASPIAN\ ?^ * 3"v>^--^__^2^ 26 - , ^ V ?f \ Gumishan j ^Gumishan -\

N "^"^^^^ / -^^TammisheWail : J Morgan 0_^ , ^W^S ' - -~L-JS*? - ^TammisheWall_J_I

Fig. I. The Gorgan Wall. Suggested sectors of the wall are as follows: east, from Fort 1-2; east central, from Fort 2-4; main, from Fort 5 west. TheDigje canal is betweenForts 16 and 17, and a possible canal is located byFort 4. more were noted, to make a total of five cross (or previously discovered by Hamid Omrani and his team). as was feeder) canals which apparently conducted water from This feature, referred to the Digje canal, the Gorgan River to the ditch on the north side of the confirmed and mapped between Forts 16 and 17. This c. m. a Gorgan Wall (Fig. 1). The western-most feeder canal 12 wide canal forms well-defined channel that now a arm was recorded during the 2006 field season (having been led water fromwhat is relict of theGorgan

2.Area the , Fig. ofembanking along /;^ -v^?^^BSH^^^^^^^^^^^HHB ^?"^1^%^^ wall tothe (thealigment parallel top \ j m;^^^^M^h||^^^^^^^^^^BH|K, ^jT ^',>^R:*K*^ withFort the .% f border) 15, rectangular i ^^^^H^^^^^^^^^^^^^lH^^^^ '':*< ' f< ^ * '' enclosurenear the left margin, *i" , ? , V^^S^^I^^^^^^^hRw^^ % . abuttingthe wall. Four parallel 4'>A] J'^-mWI^^^^^^^^^BhR^F ' ?- If are w //ze at # ^ stripes interior, lllio '^BoH^^^^^Eff \ to *7zevva/7. CORONA - ^ ' ' ng/rt0?g/es sB*iRHl^^^^B^ 3 imagebycourtesy ofUS Geological ^ ?^Jrmwjt vi^R^^^^E^^^' ,|? > Survey,with permission. ^^m^l^J^W^tSKBL l^'^JML ?g W^K^f^W V*

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions FURTHER FIELDWORK AT THE GREAT WALL OF GORGAN 97

River, which has shifted its course about 1 km. to the may have been associated with flowing water. This bank on south, to theGorgan Wall ditch to the north. There was follows the line of thewall and ditch, and is visible no evidence of an earthen dam associated with this Corona images as a broad pale soil feature (Fig. 2), and canal, presumably because, like most of the others, it on themap of Kiani as a faint feature along the south had been removed by the dynamic and erosive Gorgan side of thewall (Kiani 1982: fig. 3). It thereforeappears River. that to the east of Fort 15, thewall gives way to a silt Less certain is a channel-like feature immediately bank thatmaintains the gradient of the terrainfrom east west of Fort 4 (Figs 11, 24). This feature joins thewell to west where a shallow valley would otherwise have defined ditch west of Fort 4, and appears to have required thewall to have followed down into the valley gathered itswater from theGorgan River c. 1.5 km. to and up the other side. Although these gradients are very the south. Unfortunately, the junction of this putative subtle, this highlights how theGorgan Wall differs from feeder canal with the river has been expunged by a many other frontierwalls: it does not follow the rises massive gully system which has subsequently eroded and falls of the terrain, but rather, in themanner of a back along the canal course south of Fort 4. canal, it attempts tomaintain a very gentle gradient (i.e. Other possible feeder canals were observed a hydraulic grade), in this case from east to west. immediately east of Fort 14 and east of Fort 18. Although the actual presence of a ditch or canal is Unfortunately, neither was sufficiently clear, nor had difficult to recognise along this stretch of wall (and sufficientdiagnostic features to be classified as a canal. indeed requires further investigation), a hydraulic In the central part of the wall, immediately to the function to thewall/bank is supported by thepresence of west of Fort 15, the line of thewall was taken by a broad thefreshwater bivalves on the surface. silt bank which became relatively higher to the west, Of the threewell-defined feeder canals recorded in apparently as the wall entered a broad, very shallow 2005, the centralAghabad canal (between Forts 8 and 9), south to north flowing valley (Fig. 2). Thus at was sampled by Morteza Fattahi forOSL dating.1 This 37?15.657 north and 55?02.292' east to the east, the section, which was cut back and cleaned through the bank was only 16 m. wide and 1-1.5 m. in height, eastern bank of the canal (Figs 3-4), clearly demonstrat whereas at 37?15.652' north and 55?01.590' east, 1 km. ed thatduring the original excavation of the canal consid furtherwest (a short distance from Fort 15), ithad risen erable quantities of pale brown silt loam had been cast up to some 3.00 m. high and 36 m. wide. The presence of over the original land surface. This buried soil, a slightly occasional bivalves on the surface suggests the bank reddish brown loam with well developed calcium

- 10m 14 m

West

>^ || || Buriedsoil

3. Sketch section across canal Fig. Aghabad showing buried soil and upcast bank deposits.

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions 98 JOURNAL OF PERSIAN STUDIES

" 4.Aghabad canal: the bank 'ji^B^^^^^^^BBPP^^^^ ?^ ^ - Vi** ^fe- S13BBBBI^MR5r^ >^^:^v^'-...; deposits overlyingburied soil

carbonate concretions, represents the ancient soil that Nevertheless, itdoes appear that therewere substantial developed on the loess plain over many millennia. The irrigation systems in existence before theGorgan Wall presence of a clear bank of upcast silt over a pre-existing was built. This is particularly evident immediately soil surface demonstrates that the feature was indeed a adjacent to Fort 23 where a linear feature originally canal, although it is not clear whether the siltswere from mapped by Kiani (1982: map 6) was demonstrated to the original excavation of the canal, from subsequent be a large linear bank of spoil from a major canal. The cleanings or a combination of the two. up-cast deposits that resulted from the initial Despite the presence of themagnificent Garkaz dam, excavation of the canal (now filled with sediment) as the Sadd-i Garkaz (Nokandeh et al 2006: 140-41), no well as its subsequent cleaning out were clearly recog other dams were found in association with the cross nisable in a recently cut section through the bank. canals. The probable association between the Chai Significantly, Kiani's map 6 indicates that this feature Ghushan Kuchek canal and Garkaz dam was confirmed had been cut by theGorgan Wall thereby demonstrat by high-resolution satellite imagery (Corona) which ing that the canal existed prior to the construction of clearly demonstrates how the dam is on precisely the the wall. This is supported by the observation (by same orientation as the canal, which received water Hamid Omrani) that a large site along the line of this from the dam. On the other hand to thewest near Fort 9, canal spoil bank is of Parthian or Sasanian date. It the Sarli Makhtoom canal was associated with a would therefore have been necessary for the Sasanian complex series of possible inlet channels in the vicinity builders of the Gorgan Wall to take such pre-existing of the Gorgan River, but therewas no evidence of a features into consideration during the construction of dam. The lack of dams in association with the feeder thewall. Interestingly, this type of canal is different in canals is probably a result of the very dynamic and its conception from the feeder canals of the Gorgan erosive regime of theGorgan River, which has removed Wall, because the heavy silt load evident from the all traces of them, an interpretation that requires testing clods of silt/clay found in the bank deposits is entirely by the use of satellite imagery and (if available) aerial different from the deposits of the upcast mounds of, for photography. example theAghabad canal (discussed above). It has Although a compelling case can be made for the already been suggested that the latter feeder canals presence of a canal system associated with theGorgan were dug but never cleaned out, presumably because Wall, it is not yet clear whether such systems were they derived their water from the upper layers of simply for the supply of water for themanufacture of reservoirs that were empounded behind cross-river baked bricks, or were also for the irrigation of fields. dams (Nokandeh et al 2006: 141).

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions FURTHER FIELDWORK AT THE GREAT WALL OF GORGAN 99

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^H Fig.The ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^H bya scatter ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^Hthe the ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^Hforeground, ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^H valleytothe to ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^Hthe Elburz HIHHHHHHHII^IHI^I^Hl^Hllli^HHHHi^llHi^^^^li^H for

IV.THE GORGANWALL'S WESTERN AND EASTERN TERMINALS (HO, EWS, JR& SG)

The location of the terminals of the Great Wall of Gorgan has been the subject of much debate. We know that the Wall of Tammishe runs from the Elburz Mountains to the coastal plain of theCaspian Sea, where the wall extended to below the present Caspian level because the Sea was lower at the time of its construc tion. Similarly, one might assume that the contemporary Great Wall of Gorgan reached the ancient sea shore in thewest (and might conceivably even have joined the Wall of Tammishe), as a defensive wall only makes sense, if therewas no easy way to bypass it (see below "X. The Gorgan and Tammishe Walls pp. 112-3"). At the easternmost end we followed the wall, which is easily recognisable by fragments of robbed bricks extending to 37?29.587' north and 55?45.096' east, to just a few metres from a vertical rock face (Figs 5-6). Further fieldwork is required to testwhether itcontinues on the other side of the rock face into themountainous landscape of theGolestan National Park, or even, with or without major gaps, much furthereast as some have suggested (Adle 1992: 195 fig. 2, 204^5). The westernmost known section of theGorgan Wall ismarked for some distance by a distinct robber trench, up to 4.5 km. west of Fort 33. In order to test a stretching F/g. 6. 77ze overhanging rock, marking possible terminal of or not wall was whether the continued and detectable, thewall (see Fig. 22), would haveformed a substantial we carried out a magnetometer survey (Fig. 7) in direct obstacle.

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions 100 JOURNAL OF PERSIAN STUDIES

E 54*07,498' 'Iti k fa >vTdW>

''h'W ,l..'''(.'i" ?> ? ]j V vM i''* ?*'?! ' pitinto kiln A %i ? ..j ;MV

c. F/g. 7.Magnetometer survey of the Gorgan Wall and two associated kilns, 4.8 km. west of Fort 33, with the location of Trench I, byJR, SG andHO. The linear anomaly is likelyto representthe Gorgan Wall robber trenchand associated ditch, the two are two rectangular anomalies south of it kilns. continuation of the robber trench some 300 m. west of case of the survey of Fort 30 on the protected south side itsvisible terminal. The survey revealed a distinct linear of the wall, we found two positive anomalies, whose anomaly, c. 18m. wide, likely to have been caused by a rectangular shape, size and strength suggested two combination of scattered brick fragments from the furtherbrick kilns, c. 86 m. apart. The postulated contin robbed-out wall, any remaining foundations of thewall uation of the ditch could have provided the water and, probably, an associated ditch. Behind it, as in the necessary for brick making.

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions FURTHER FIELDWORK AT THE GREAT WALL OF GORGAN 101

-0.04 ^?000_^_ " ? " i -0.08 *j ! 1001 i |II^J)02^^^-

l_ -> ! Pale brownfine sand ^C^^_ j

^-.--v?-' -" \ 1.003 j

^^^C^V-^v 1005 / - ^ j ,_l

I 1-006\ q

! 5". I 5"

??. ? ? --^?

0 50 cm Section of brickkiln (Trench I) ^ 20Q6 5 section Trench I, ^^.^ F^ North-facing of Drawn byM.Hussein Zaden drawn byMoryam Hussein-Zadeh.

V. A BRICK KILN FLOODED BY THE CASPIAN them still found in pairs, suggesting that the soft tissue SEA (HO,TW, EWS, MM, SG & JR) of thesemarine molluscs had rotted in situ. There is thus no doubt that the kiln was flooded by theCaspian Sea. In order to test the hypothesis that the eastern anomaly Subsequently c. 1m. of weakly bedded mottled clay reasons indeed represents a kiln and to explore the why built up, with thin lenses of pale brown fine sand no brick fragments or other traces of the postulated kiln (deposits 1.001, 1.002, 1.003 and 1.008) of a type were on we and robber trench visible the surface, typically deposited either in lagoons, mudflats or a x a excavated a 2 1m.-large testpit (Trench I) through shallow embayment of the sea (Fig. 8). The accumula very low shell-covered north-south ridge. According to tion of this transgression deposit was followed by an some figure 8b of Kiani, thewall may have extended 2 episode of drying and soil formation followed by either km. furtherto thewest of Trench I, afterwhich it either a subsequent regression of the sea, a second transgres or terminated disappeared from view. We encountered sion (represented by a thin accumulation of sands and m. the top of the kiln 0.98-1.07 below the surface (Fig. the marine bivalve Cardium: cockle), or a localised 8). Substantial parts of two of the typical arches survived episode of aeolian accumulation. The elevation of this within our trench and a small of a third part (Figs 8-9). part of thewall, c. -22 m. in relation to global sea level, A in situwas selected for brick OSL dating. is some 6 m. above present Caspian Sea level. was a The abandoned kiln immediately overlain by According to the studies of the Russian scientist G.I. layer of marine bivalve shells (deposit 1.004), some of Rychagov (1997), themost likelyevent(s) thatwould

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions 102 JOURNAL OF PERSIAN STUDIES

N " ??" T^""'^ \ - \ Brickfor / ^ ^\ ps&&* - - \OSL ^ ^ ? r<\ datingy

cm O^^^^^^^^^^^O Plan Of Trench I The Great Wall of Gorgan Project 2006 Drawn by M.Hussein Zadeh

Fig. 9. Plan of TrenchI, drawn byMaryam Hussein-Zadeh.

correspond to this inundationwere a series of transgres In order to recover additional evidence for themarine sions that reached c. -22 m. between c. 500 years ago transgression at Trench I, the distribution of shells was and the present day (Fig. 10). The shells from context traced along the line of the Gorgan Wall to the east 1.004 have now yielded a radiocarbon date of A.D. between Trench I and Fort 33. Here thewall was evident 1344-1460 at 95.4% confidence level2 for the initial as occasional fired bricks and robber pits, partly obscured transgression. Although this high sea stand corresponds by a veneer of shells of the same species as those exposed chronologically to thatof Rychagov, the elevation of the inTrench I. This shell scatterwas traced east of Trench I equivalent event along the Gorgan wall, at -22 m., is towithin 1km. west of Fort 33, atwhich point itfaded out some 3 m. higher than Rychagov's estimate. Further at an elevation of c. -22m. (in relation to global sea level). dates and field studies are therefore required from the Significantly the shells fade out in the vicinity of a faint Gorgan Wall Project in order to sort out some of these north-south feature mapped by Kiani and colleagues contradictions. (stippled on Kiani 1982: fig. 8b). This feature probably

-20 -

_ ? Elevation \ y} m below /VV\/ - / \*.*/ A.A sea "32 / level _/ p Wall 36 y Gorgan _i_i_i_i_'i iH ?_i 10,000 8000 6000 40002000 0

Years BP

Fig. 10. Caspian Sea level curvewith main phase ofGorgan Wall indicated(rectangle), based on Rychagov 1997:fig. 5.

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions FURTHER FIELDWORK AT THE GREAT WALL OF GORGAN 103

/; Hollow//

" c- 300m ^\Vox^K \ \ Vi,la9e ) Outer Canal?^,^*^^-^x" ? settlement \

Fig. 11. Sketchmap ofFort 4 and hollowways.

a represents raised shoreline of the late or post-medieval Huff 1981). Defensive logic dictates that theWall would Caspian transgression that inundated thewall and subse have continued until it either reached the ancient shore quently obscured it.To thewest, the low ridge at Trench of the Caspian Sea, or possibly the Tammishe Wall, on a I appears to be similar relict Caspian shoreline, but terrainnowadays under the sea level. whether these two Caspian shoreline features,which are some 3.5 km. apart, relate to the same transgressive event or differentones requires furtherinvestigation. VI. THE EASTERN PARTOF THEWALL The 1m. deep marine deposits at this point explain BETWEEN FORTS 2AND 4 (HO & TW) for the first timewhy ithas so far not been possible to follow theGorgan Wall to the Caspian Sea. Relatively The Gorgan Wall is a massive feature within the recent marine transgressions have buried the remains of landscape, and themaps made byM.Y. Kiani provide a thewall so deeply that neither features on the surface very informative record of the wall, forts and other nor course. artefact scatters reveal its As there are features within their landscape context. Now with the neither traces nor a seems of thewall, of robber trench, it availability of high resolution satellite images, it is that ithad been robbed out to prior being flooded. It is possible to recognise and interpretmore features than to note no interesting that,despite its late flooding, soil was the case before, and the following texthas benefited horizon had over built up the top of the kiln. This from the use of these, together with limited field was suggests that the top of the kiln above ground and checking in 2006. that the depth of marine deposits is probably greater As noted above, Fort 4 may have been associated with next to thekiln. While the marine so a cross deep deposits have canal that led water from the Gorgan River. far a successful location of thewesternmost a prevented Moreover, high resolutionDigital Globe image indicates stretch of thewall, it seems most improbable that the thatat the north-west corner of the forta distinct "hollow western terminal of the known section corresponds with way" can be traced leading northwards across the now the real which has been terminal, postulated before (e.g. cultivated steppe (Fig. 11). In addition, slightly less distinct

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions 104 JOURNAL OF PERSIAN STUDIES

1 CiTs Site \ ;

/ Site\ %'\ fo /

: Site \ ' Hollow \ / way \ y hoiiow: \ !\ i / ,;A\Ways \ ^^^^

c. iooo mVu / betweenForts 2 and 4 and - /-/ / s^tesw^ assoc^ated hollow //VI \ ?^ / / _// /_/_ ways.

features can be discerned c. 100 m. and several 100 m. can be seen associated with thewall as well as with three to the east. In the Near East, hollow ways generally associated archaeological sites to the north, that are appear as broad shallow valleys (or soil marks) which beyond the limits of thewall (Fig. 12). Of these, themost radiate from archaeological sites, and they can be distinct is the site to the north of the bend in thewall, interpretedas having been worn down by themovement which is associated with a distinct hollow way that trends of people and thousands of animals over the centuries south tomeet thewall at the ditch, fades out at thewall (Wilkinson 1993). Although a good case can be made itself,and then continues albeit with a more faded trace for the north-west hollow way being associated with the south of the wall (Fig. 12). A second hollow way fortand wall, those furtherto the east have more distinct bifurcates from the main feature and again continues traces to the north than to the south of thewall, and may south of thewall, whereas an additional fainthollow way thereforepre-date thewall. Although at Fort 4 a case can to thewest of the first,joins that feature to the south of be made for the presence of a crossing point associated thewall. Overall, the configuration of these hollow ways with the wall itself, the presence of the ditch as a near thebend in thewall appears to represent a system of continuous feature along this part of thewall suggests: settlement and communications that pre-date the wall, 1) that the ditch at this point was drywhen the trackwas and were then cut by it,although this suggestion needs to in use, 2) that itwas dry immediately to the east of the be tested by furtherfieldwork. Similarly, it is difficult to putative inlet canal, or 3) that thehollow way was inuse statewhether thehollow way feature at Fort 4 was inuse before or after thewall was actively used, but not at the at the same time as thewall, or represents a pre-wall (or same time. Significantly, none of the hollow ways is perhaps post-wall) feature. However, it is evident that oriented precisely upon what would have been gateways with futurefieldwork it should be possible to tease out positioned in the centres of the north or south walls of some significant relationships between thewall and the Fort 4. route systems of the region. Further to the east, near the sharp obtuse angle bend Additional investigation of the eastern sector of the in thewall between Forts 3 and 2, similar hollow ways wall suggested that Fort 3, as designated by Kiani,

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions FURTHER FIELDWORK AT THE GREAT WALL OF GORGAN 105

Village/ f\\ f\ \ \ ^)

" VH,a9e ( J FORT2A[ ^~\\

PRudkhane Sari Seyyid

>WP

FORT2 \ L?eSS ) ^7 /^Sy Plateau^^

Fig. 13. Sketch map of Fort 2 area and associated sites. does not now exist. In 2006, GPS navigation enabled gap in thewall, some 35 m. E-W, suggests that the Fort us to pinpoint the location of the fort, and field inves 3 was situated at a crossing point. tigations suggests that the small fortmapped by Kiani A short distance to the east of the bend in thewall, (1982: fig. 1) remains as only two groups of fired Kiani's map records two branches of wall. Field brick. One of these forms a small but prominent checking in 2006, as well as inspection of Digital Globe mound of fired brick on the wall itself,whereas the images, suggests that the northern arm is associated second consists of a low scatter of fired brick c. 170 m. with a broad ditch with associated bank/ wall to the to the south, adjacent and immediately north of the south, whereas the southern branch has the same modern road. The only other evidence for cultural narrow trace as the main wall ditch to the west, and activity within the area where the fort should have continues the line of the wall to the east on a direct been is a light scatter of pottery across the surface of alignment with Fort 2 (Fig. 13). On the northern limb of the fields. The distance between the two scatters of thewall the site of a possible fort had been recognised fired brick, c. 170 m, falls within the dimensions of the (by Hamid Omrani) (Fig. 13: 2A). This site occupies an Gorgan Wall forts and therefore Fort 3 may represent a important strategic location on the high western bluffs "ghost" of a fort thatwas visible on the air photographs of the Rudkhane Sari Seyyid (a river), where the employed by Kiani, but is invisible today. There is no Gorgan Wall apparently terminated before resuming its evidence that Fort 3 has been destroyed by agricultur course to the east at Fort 2. This previously un-recorded al activities, and it is therefore possible that either the site (it is not evident on themaps of Kiani) lacks the feature was never completed, or that itwas partly laid distinct square plan of the other forts, and is heavily out (hence its visibility from the air) but the garrison pock-marked by abundant robber pits. A brief survey of was then moved to an alternative, perhaps more this site (near the village of Tamar Gharaghuzi) showed effective location. Interestingly the presence of a broad that a linearmound along thewestern side, as well as an

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions 106 JOURNAL OF PERSIAN STUDIES

^^^^^^^^^^^^^^^^^^^^^^^^^^^^ ^

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^mFig. Feature ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^H /?/7cA;s' between Hil^Hii^HIHIIHlHHHHHHHHHHIHllHHIiHIHHIHHi location.)

apparent moat along the same side would support the (upon which the two figures are standing on Fig. 14), identification of this site as a previously unrecognised is oriented N-S at an oblique angle to the first. This fort, referred to here by the temporary number Fort 2A. second part of the structure points upstream and Fort 2A appears to have been built to have oversight of disappears below thewaters of the stream (to the left the strategically important and deeply entrenched Sari of the figures on Fig. 14). The first element which is 8 Seyyid River, and may have been constructed to replace m. long and 2.16 m. wide then continues into thewest Fort 3, which was not positioned at an important bank of the river as is evident from a deep cut into the strategic location within the landscape. More signifi hardened river sediments. Laid out at an angle of 137 cantly, the existence of two distinct branches of the degrees to this, the second fired brick structure forms a wall, with quite differentmorphologies and on different platform on the bed of the stream. Part of this is alignments, provides compelling evidence for the submerged beneath thewaters of the river, but is still existence of two phases of the wall at this point. In visible and measurable by wading in the water. terms of morphology and alignment, the first phase Although the 2.16 m. wide segment of wall could be a would be the southern feature and the second phase component of the Gorgan Wall, the platform built would be that to the north (Fig. 13, 1 and 2 respective obliquely to it is intriguing. Being apparently ly).Why the construction of two phases was deemed constructed within the bed of the river, this feature necessary is difficult to say, but it is possible that the could be reasonably described as "hydraulic", and its first, southern phase of the wall was destroyed or location a few hundred metres off the course of the compromised by the deep, rapidly eroding gullies that wall suggests that it is not strictly on the wall's line. were eroding back from the Sari Seyyid River, and that One possible interpretation of this structure is that it itwas necessary to circumvent themwith a later phase does indeed represent a short length of wall in the of wall further to the north. valley floor, and that the fired brick platform may In the floor of the valley of the Sari Seyyid River, represent an apron that was constructed to conduct an apparent part of theGorgan Wall complex has been water beneath the wall at a point where an arch was noted (by Hamid Omrani). This substantial feature of constructed to allow the water free passage. Without fired brick (Figs 13, "WP 438" and 14) is located in the such a reinforcing apron, the river could incise into its bed of the flowing river. It consists of two lengths of bed and scour out the foundations of the wall. wall or platform made of Gorgan Wall fired bricks. The Although tentative this interpretation gives an insight first oriented at 233 degrees magnetic is at right angles into the sophisticated engineering of the Sasanian to themain channel of the river,whereas the second architects.

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions FURTHER FIELDWORK AT THE GREAT WALL OF GORGAN 107

VII. THE GORGAN WALL IN ITS LANDSCAPE: this length.The presence of an embankment east of DISCUSSION AND CONCLUSIONS (TW& HO) Fort 15, reinforces the observation that thewall followed a hydraulic gradient, and itappears also The presence of five well-defined feeder canals suggests that theDigje canal, between Forts 16 and 17,may that rather than forming one continuous water supply have been part of the same water supply sector as system, the ditch associated with theGorgan Wall was theAghabad canal. The role of water supply in the probably sub-divided into separate water supply sectors main sector of thewall (i.e. west of Fort 5) is (Fig. 1). supported by the clear relationship on Digital Globe images between theGarkaz dam and the canal that 1. The easternmost sector, to the east of Fort 2, has led towards thewall, as well as the evidence for not been examined for evidence of water supply, upcast banks on theAghabad canal. but it is clear that itwould have been difficult to have brought water from this sector to the sector Although it is not possible to unambiguously west of Fort 2A. distinguish specific crossing points through the wall, 2. Between the Sari Seyyid River, and the tributary the presence of hollow ways at Fort 4 as well as to the that enters theGorgan River west of Fort 4. This east near the bend in the wall, suggests that such sectormay have been supplied with water from the crossing points can be recognised, and that in places the probable feeder canal adjacent to Fort 4, as well as wall may have actually been constructed over pre from an unknown source to the east. existing settlements and their associated local route 3. The sector to thewest of Fort 5 provides the best systems. The presence of two lines of thewall between evidence for a system of water supply linked into Forts 2 and 2A, suggests that allowance must be made the construction of thewall. It is clear that this for two phases of wall construction, at least at the east system cannot have received water from sector 2 end of thewall. On the other hand at thewest end, near (between Forts 2A and 4) because the deep valley theCaspian Sea, thewall has been obscured by deposits between Forts 4 and 5 shows no sign of having from a high late and post-medieval stand of theCaspian been equipped with the complex engineering Sea, but whether the wall terminates to the north of systems required to transmit thewater from the east Gumishan (presumably at the edge of theCaspian Sea) to thewest banks. Overall, the entire stretchwest of or turned south to eventually link up with theWall of Fort 5 appears to have constituted one hydraulic Tammishe (see below "X. The Gorgan and Tammishe ... system,with the ditch being evident along much of Walls pp. 112-3"), remains an open question.

+003_

? "- _____C!Z FirP V*\ I )/- '-"1. -0.39

1 _ _ |^-138 ^__ J !

1m 0_

The GreatWall ofGorgan Project 2006 East section of brick kilnwith 11 crossbars (trench K) Drawn by M.Hussein Zadeh

Fig. 15. TrenchK: west-facingsection ofkiln cut by a recentdrainage ditch,drawn byMary amHussein-Zadeh.

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions 108 JOURNAL OF PERSIAN STUDIES

VIII. SURVEY OF A BATTERY OF BRICK KILNS WEST OF FORT 30 (HO, EWS, JR,SG & AN)

Approximately 1.3 km. west of Fort 30 one brick kiln had been cut through by a recent drainage ditch (Figs 15-16) and a series of shallow mounds in itsvicinity, all roughly equidistant from the robbed-out Gorgan Wall, suggested that therewas a whole series of kilns. In order to evaluate this hypothesis, James Ratcliffe, Seren Griffiths and Hamid Omrani carried out a magnetome ter survey of a 240 m. long area (Fig. 16).We hoped that this survey would not only provide valuable insights into the scale and organisation of brick production, but would also allow us to identifya well-preserved kiln for excavation and dating. The survey identified a remarkable concentration of kilns south of thewall: three to thewest and three to the east of the kiln cut by the drainage ditch, i.e. no less than seven within just 220 m. (centre to centre). Their distance to each other (on average 37 m.) and their orientation (some parallel or near parallel and some at a right angle to the wall) varied. However, as far as a survey allows us to tell, the centre points of all seem to be on the same alignment and all appear to be of similar dimensions. We cannot assume that a similar concentra tion of kilns existed at all sections of the wall. While suitable material was available at all sections, water may have been more of a problem, even though most sections seem to have been supplied by canals (Nokandeh et al. 2006). In the central section east of Gonbad, kilns have been identified in theGorgan river valley, which may have produced bricks for thewall as well. Nevertheless, the discovery of yet twomore kilns, spaced 86 m. apart, 4.8 km. west of Fort 33 (Fig. 7) at a point where no brick slag ormounds had suggested their presence (i.e. a genuinely random sample), may suggest that a comparatively narrow spacing of kilns was not atypical and that several thousand kilns may line the wall. The same, in terms of distance between kilns, may be true for the roughly contemporary Tammishe Wall and our previous assumption (Nokandeh et al. 2006: 153-55 with fig. 32) that the concentration in the surveyed area was exceptional may have been mistaken. The high number of kilns is somewhat surprising, as one would have thought that building fewer kilns and reusing them more often would have been more even if itwould have resulted in a Fig. 16.Magnetometer surveyof theGorgan Wall and seven economic, slight increase in the distance between and associated kilns, c. 1.3 km. west of Fort 30, with the location average production of TrenchesG and K, byJR, SG andHO. construction site. It seems hard to explain why those

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions FURTHER FIELDWORK AT THE GREAT WALL OF GORGAN 109

-014_^__-Q09 -0.20 j -1.20 i N i 4

I I -1.21

1- 1 -L?_.?nm~' / -ao \ -008 J ? lHill I" f _3H1

-0.05 -?*?1A II

t JDED BM \6

TUD 8 \ fs her >

-o.i7 \ y^jJAJ??j 10 \ / 10 1- \ruTiLis---i \? -0.24Uiii^^ j V_J"|

-140 [ j

'-1.35 -1.27 rtoc -038_ -0.31 1?- ?"-L--j -0.35 0_1m Plan of brick kiln Tu ? ? v ' (trench G) ... AA. TheGreat Wall ofGorgan Project 2006 with 11 crossbars n kllu . Drawnby M.Hussein Zadeh

Fig. 17.Plan ofTrench G, drawn byMaryam Hussein-Zadeh.

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions 110 JOURNAL OF PERSIAN STUDIES

OSL brick 3.57W sample 1.33 N / 0.20W

\ OSL J 5 ^r?r-//WA20 /

V _ ? ? --? x-? ? ? ? -140_.

?m fiSi2!!__5i_5__iiiia South section of brick kiln (trench G) The Great Wall ofGorgan Project 2006 = NOS Correspond tO Context nOS, e.g. 1 G.001 Drawn by M Hussein Zadeh

Fig. 18. South-facingside of thekiln inTrench G, drawn byMary amHussein-Zadeh.

thewall did not reuse a building smaller number of kilns total. The width of the kiln is similar to its partially more often. Possible reasons for the narrow spacing of excavated counterpart on the Tammishe Wall kilns include thewall being built under great pressure of (Nokandeh et al 2006: 154 fig. 33), a possible a more time and using larger and skilled workforce than indication for the attribution of these two kilns and, by a previously thought, and/or that it reached greater implication, the two walls to the same building height. programme. Unlike theTrench A kiln and thekiln on the Wall of Tammishe (Trench F), the arches of the Trench G kiln did not survive to their full height, but had IX. EXCAVATIONOF A BRICK KILN collapsed (Fig. 19). (HO, EWS,MM & SG) The main purposes of the excavations of the kiln were to gain insights into its architecture and to obtain In addition to the one kiln cut through by a recent further scientific dates for its firing and thus, almost drainage ditch,whose section was cleaned and recorded certainly, for the construction of this section of the in Trench K (Fig. 15), another kiln (Trench G) was Gorgan Wall. It seemed important to date thewestern excavated (Figs 16-20) under the direction of Hamid section of the wall, as all 2005 dates were based on Omrani and Majid Mahmoudi. Both kilns had been cut samples from the central and eastern sections into thenatural soil. Like thekiln inTrench A, theyboth (Nokandeh et al 2006: 158-63); the dating evidence comprised eleven crossbars. The dimensions of the presently available does not allow us to decide Trench G kiln (6.60 x 3.60-3.80 m.) were smaller than whether the Tammishe Wall and the sections of the that of the kiln excavated in 2005 in Trench A Gorgan Wall west and east of the Pishkamar Rocks (Nokandeh et al 2006: 130-35) and that excavated by were all built simultaneously or successively and Kiani (1982: 17, 18 fig. 11) at Fort 13, as were the whether construction works were completed within a associated square bricks with a side length of just 0.37 few years or decades (as seems likely) or may have m. In all three cases the upper width of the kiln lasted for anything up to one and a half centuries. corresponds roughly to ten bricks; the kilns appear to The southern and northern side of the kiln were have been designed for ten stacks of bricks sideways excavated, but virtually none of the interior,except for a and 17 or 18 stacks lengthwise, i.e. 170 or 180 stacks in small sondage in the south (Figs 17 and 20). In the light

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions FURTHER FIELDWORK AT THE GREAT WALL OF GORGAN 111

0.46W

0 cm _50 North Section of brick kiln (trench G) Si^^^HiS The GreatWall ofGorgan Project2006 Drawnby M.Hussein Zadeh

Fig. 19.North-facing side of thekiln inTrench G with collapsed arch, drawn byMary amHussein-Zadeh.

of the lack of excavation of the interior,our understand arch and thus not exposed directly to the fire. Contexts ing of the history of the kiln is limited and what we can G.020, G.021 and G.023 also represent fire-reddened deduce from the sections shall thusbe summarised rather deposits. Brick slag (G.005), deposited on the south concisely. There is no doubt that thekiln was used. A thin facing side of the kiln is further evidence for the high well-defined band of dark brownish black silty clay temperatures reached during firing.G.015, G.016, G.017 (context G.001, Fig. 18) is best explained as a wooden and G.018 were rich in charcoal. Unlike the north-facing plank for the revetment of the kiln charred in situ, as first section (Fig. 19), therewere no signs that the arches had suggested by Seren Griffiths. The intense heat even led collapsed into the interior in the southern section?and in to the fire-reddening of soil (context G.002) behind the the case of the northern section this appears to have

20. Thekiln in Trench G seen ^^^^^^^^^^^^^H Fig. thesouth. Each the ^^^^^^^^^^^^^H from segmentof IBHHHIHHHIH scalescorresponds to0.50 m.

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions 112 JOURNAL OF PERSIAN STUDIES

happened at a timewhen much of the interiorof the kiln As noted above, the bricks from the excavated kiln west had already silted up. Too littleof the post-abandonment of Fort 30 had a side length of just 0.37 m. Interestingly, deposits in the kiln have been excavated to allow a bricks from furtherwest, which we saw reused in a certain reconstruction of events. There seems littlepoint cemetery on Gomish tepe or the so-called Island of in speculating what an isolated horizontal brick Abskun, at Gumishan, had identical dimensions. It is overlying G.015, which is overlain by a dark brown silty also worth remembering that the kilns from the clay deposit (G.014) or a later re-cut,filled by a brownish Tammishe Wall are on average smaller than those from red sandy clay (G.004) with a lens of orangey red sandy the central and eastern section of theGorgan Wall. It is clay (G.006) might reveal about lateruse. It is clear that not clear whether we are dealing with differentworking none of these deposits or theirvicinity were exposed to parties or chronological differences; theTammishe Wall great heat. They indicate that the kiln may have been and the western section of the Gorgan Wall may, for reused for some other function (e.g. makeshift shelter or example, have been constructed first.While the similar storage), but that itwas no longer used as a kiln. As with brick size of thewestern section of theGorgan Wall and the other kilns excavated, this example shows no sign of the Tammishe Wall and the similar architectural type having been reused in its original function at a laterdate. and spatial arrangement of kilns along thewalls need The OSL sample taken from a brick and context G.002 not indicate more than that their architects shared a thus almost certainly relate to theproduction of bricks for common tradition,neither can we exclude that the two theWall. Except for narrow stems, probably of the walls were jointly planned and executed. The Chenopodiaceae family, none of the charcoal from the observation that the ditch of eithermonument is on the fill of the kiln was identifiable (see below "XIX.2. same side (i.e. facing away from theGorgan Plain) may Charcoal Analysis" pp. 132-3). Whether or not the offer further circumstantial evidence for the walls scarcity of identifiable charcoal may be the result of the belonging to a singlemonument, and proves in any case use of dung fuel inwhat is, nowadays at least, a treeless that the Tammishe Wall, with its ditch in thewest, was landscape is impossible to establish with certainty.Dung not primarily designed as a second line of defence fuel would be more likely to disintegrate into small behind theGorgan Wall. If the Bansaran Fort was part particles which are hard to identify (Rowena Gale, pers. of the original scheme, despite its unusual location comm.). The results of the radiocarbon dating of these outside the Tammishe Wall, then its earth platform is charcoal and theOSL samples are, unfortunately,not yet also a common featurewith theGorgan Wall forts. It is available. even possible that the two walls were part of a single now Little would be gained by a detailed scrutiny of the monument, themissing link being on land flooded remainder of the fill of the kiln. Successive deposits by the Caspian Sea (a theory first advanced by Jebrael vary in the proportion of brick fragments. They almost Nokandeh and Hamid Omrani in 2002). certainly represent the gradual build-up of material Itmay be significant that thewesternmost known part while parts of the superstructure crumbled and were of thewall seems to curve southwards, especially if the on around probably partially robbed, while natural soil and brick concentration ofGorgan Wall bricks from and debris were washed into the interior. The noticeable Gomish Tepe and Gumishan indicates that the wall reduction of brick fragments in the latest fill (G.008) passed their vicinity (Kiani 1982: general map; Talbert = Gomish suggests that the final silting up post-dates the 2000: map 96 plotting thewall and "Gumush" of that robbed bricks postulated robbing of any remaining upstanding arches. Tepe). While it is, course, possible were brought to these sites from some distance, the sheer number of Gorgan Wall bricks in old buildings in X. THE GORGAN AND TAMMISHEWALLS: TWO Gumishan and scattered in and around Gomish Tepe, SEPARATEMONUMENTS OR TWO PARTS now and already in the late nineteenth century (Yate OF A SINGLEWALL? (JN,HO & EWS) 1900: 272-73; Adamec 1981: 200-1), renders itmore likely that thewall was in close vicinity.Yate (1900: 273; his visit Scientific dating carried out in 2005/06 (Nokandeh et al cf. 226-27, 260-61; Adamec 1981: 201), during Wall and other 2006: 158-63) suggests strongly that the Tammishe in 1894 to various sections of theGorgan a were to be Wall is either contemporary with the Great Wall of sites in the area, heard of "report thatbricks a under thewater some distance out in the the Gorgan or that the twowalls are atmost century apart. found sea",

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions FURTHER FIELDWORK AT THE GREAT WALL OF GORGAN 113

coast at the time apparently being only "a good half the entire wall). No clear traces of interior buildings mile" west of themound of Gomish Tepe; this suggests emerged, but it is impossible to tellwhether therewere that theGorgan Wall, like the Tammishe Wall, ran into none or,more probably, they did not cause sufficiently terrain flooded by the Caspian Sea. Joining theGorgan strong anomalies to be detectable. Parts of the forthave and Tammishe Walls would have prevented enemies served as amodern graveyard and a large burial enclosure from bypassing the system using boats. There is no and tomb-markers, consisting of reused bricks,may have parallel, to our knowledge, in the ancient world of a long contributed to the lack of clarity of the survey. wall running along a seashore, other than to protect cities, the reasons being that many of Rome's and Persia's northern enemies had no, or extremely limited, XII. THE BUILDINGS INA SASANIAN FORT: experience in seafaring. More importantly, in most MAGNETOMETER SURVEY OF FORT 4 instances (e.g. Britain or the Peloponnese) the length of (EWS,HO, RA, EST, JJ,ME, SG & AN) vulnerable shores exceeded that of the defended narrow land corridors many times over. It would have been The 2005 geophysical surveys of Forts 1, 9 and 10 impossible to effectively protect the shores through (Nokandeh^a/. 2006: 125-29, 142, 151), and the 2006 barriers. By contrast, the gap between thewesternmost survey of Fort 5, had yielded rather inconclusive results known point of theGorgan Wall and theTammishe Wall on the nature of these forts' interior occupation. The was substantial at c. 32 km. as the crow flies from density of artefacts from fort interiors and an excavated Gomish Tepe to thenorthernmost traces of theTammishe section through the ditch at Fort 9, as well as more Wall or associated fort in theCaspian Sea, but limited in general considerations on defensive strategy,neverthe relation to the 200 km. combined length of the known less suggested strongly that the interiors of the forts on sections of the two walls. This gap could have been the Gorgan Wall housed sizeable garrisons, probably closed if therewas a desire to do so, but this does not accommodated in structuresmade of materials which prove, of course, that itwas, and it is equally possible that are hard to detect magnetically. were they indeed two separate walls and that both It thus came as a pleasant surprise that the 2006 mag at terminated the ancient shore of theCaspian Sea. netometer survey of Fort 4 (Figs 21-22), by Abingdon Should theGorgan and Tammishe Walls be part of Archaeological Geophysics and the ICHTO, under the same the wall, themissing link being partially flooded direction of Roger Ainslie, provided us with astonish the by Caspian Sea and partially buried beneath marine ingly clear results. On either side of a NW-SE-running sediments, then there were presumably further forts central road we found two parallel anomalies of c. 228 along this section. This would have implications for the m. length. The regular subdivisions leave no doubt that size of the wall's to be in garrison, discussed the these have to be interpreted as elongated buildings, concluding chapters. divided into two parallel rows of rooms. A third, virtually identical, anomaly, parallel to the other two was detected in the SW of the fort.At first sight, theplot XI. MAGNETOMETER SURVEY OF FORT 5 may suggest that there are six rather than three such (RA,HO, EWS, EST, SG,ME & JJ) buildings, each of the three long double rows of rooms divided in themiddle by a road, running from SW to Crucial for understanding the purpose and history of the NE. However, while furtherfieldwork will be necessary are Gorgan Wall the size and composition of itsgarrison, to prove or disprove this theory, this route,which is now and the it to defend the Plain. as a strategy employed Gorgan visible shallow hollow way, is probably a secondary In order to shed on these we light key issues, continued to feature. The linear anomalies, thought to be walls of the the interiorof explore fortsvia geophysical survey.Only three buildings, do not appear to come to an end on a of Fort 5 was For this and for the Fort 4 either part surveyed. side of this track, but to run across it, suggesting a Grad 601/2 with 1m. we are survey, Bartington gradiometer that dealing with three very long, rather than six line was for the other a Geoscan spacing used, surveys, shorter, buildings. The absence of any suggestion of a FM256 Gradiometer. Fluxgate The results confirmed that SW or NE gate in alignment with this track is a further theGreat Wall of where itforms thenorth of Gorgan, side argument for itbeing a secondary feature,perhaps added the was fort, made of firedbricks (as appears tobe truefor in a later phase to facilitate movement from the

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions 114 JOURNAL OF PERSIAN STUDIES

Fig. 21.Magnetometer surveyof the interiorof Fort 4 and selected extramuralareas byRA (AbingdonArchaeological Geophysics),HO (ICHTO), EST, JJ,ME, SG & AN. For scale and northarrow see Fig. 22. buildings to the south-western and north-eastern the average NW-SE extent of a room, allows us to defences. estimate that each block contained c. 45^18 pairs of Unfortunately, a cemetery in the northern part of the rooms, or a total of 90-96. fort, comprising modern gravestones, circular and The regular layout and room size in the three blocks square burial enclosure ditches, mounds and suggests that they served as accommodation for the monuments built of reused fired brick, made it fort's garrison. It is also possible that some may have impossible to detect the precise extent of these ancient been used as stables or storage facilities or formultiple buildings. While we cannot be absolutely certain, there purposes, or that their function changed over time. The is nothing to suggest that the three buildings varied sig standardised design suggests in any case that all rooms nificantly in size or room divisions. The remarkable were either intended for the same purpose or designed to regularity of the fort's layout offers support for the be adaptable for a range of functions. There are no dis assumption that theywere intended to be identical. If so, tinctively larger rooms (or at least not in those parts of it may be permissible to reconstruct one block's the plot which are sufficiently clear), as one might have obscured parts on the basis of detected sections of expected for officers' quarters. Neither are there any another. Counting the rooms in clear sections of the plot rooms or recognisable separate buildings for special and dividing the approximate length of the buildings by functions, be it storage, administration, caring for the

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions FURTHER FIELDWORK AT THE GREAT WALL OF GORGAN 115

A: N 37?27.210'| | W. i E55?25.196' | ! t*\ B: N 37?27.122' 1 III; \Y I~"E 55?25.152'""\ I I I I I C: N 37?27.080' lil E 55?25.284' I i I I D:-N 37?27.t68'.?h-h-.~j-r""?t" E 55?25.328' [ !iI A\X .t. [Modern.t. .../. ; /D Modern/burial Modern ditched r\ \brick-built \ / enclosures \ brick-built \ \.\ lk iburial i I burial \ I l\ !enclosure j mound ^ / i _./__ / "-..i^ / iw\_ ^i2^_|_._,_|_I

Modern| I \ r i\ y burialA |\l V /\ /\ f?f?eOf enclosureGate I ''''^ ' ' V - --. X > | , . ' / ^ \ // ^ \-1-1-L. . ModernI \^\/f^4/ / />"--., I - A. i ^ a*, ! tomb- \ "'V,V\ ,

\ ? /-^/V ^. ^ building?'",, /

^ ~ Modern ^ ^ ! .. j brick-built c^^-J^^\ \ j ""f-.. X//>^0^ j "'X^(/Xyj'/ Gatej burial \ *"*\ I"-- | I monument \ \ ii^Xv^/S4j^>\ 11' ^/^/>y / \^ /^>v-<: I

! i i1 !: ^x ^-%^^<)v7>^ G? / / I : ! ! i ^\ ! ^<7>f\ / / I ! !

i | ! : ' j

22. Fig. Interpretativeplot ofFort 4 with the location of trenchesH and J. Each grid corresponds to 30 x 30 m. The projecting towers are not as cannot be on plotted they pinpointed the magnetometer survey Only some of the modern burial enclosures and mounds have not been plotted. They have been surveyed and their extent is only approximate.

sick or or or wounded repair production of equipment. 91-92) hypothesis that the "Barracks Forts" were Furthermore, the absence of parallels for non-standard designed to train recruits, which, as he concedes, is rooms or in the buildings other, admittedly few and "undeniably speculative", is unlikely to be the correct "Barracks Forts" below largely unexplored, (see "XVI. explanation forwhat appears to be emerging now as the Sasanian Fort renders itmore Design" pp. 127-8) likely main (or at least one of themain) fort type(s) in the that this fort contained exclusively buildings of similar border zones of the Sasanian Empire. for all design, catering required functions and for Instead it isworth pointing out thatmany lateRoman soldiers and officers alike. ordinary David Oates' (2005: forts also seem to contain far fewer, if any, specialised

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions 116 journal of persian studies

fiH^^^^^I^^U^^^^^^^^^^^^^^^^^^^^^^H Fzg. 25. ^ modern ^HB^^Hfj^HHHj^H^^ 6wna/ enclosuremade SBSHHHbHBo/*reused Gorgan Wall 6ncAs ffjjH (seeF;g. aw^/f/ie HSRH^H magnetometer BBHSgflHBM survey inprogress.

no buildings and overtly high-status accommodation. away in theNW, probably a length similar to thebarrack Gundolf Precht (1987), for example, postulates that in the blocks (Figs 21-22). There is a suggestion of a possible late Roman fort at Cologne-Deutz, the central four enclosure of similar width around the barrack block in barracks, which differedfrom the remaining 12 by a short the SW (Figs 21-22). Although it is possible thatwe are as portico only (as far the small-scale excavations allow to dealing with a road grid, it is tempting to assume that were tell), reserved for officers and administrative these postulated enclosures served specialised functions functions. It appears that inboth Sasanian and lateRoman (e.g. corralling horses, beasts of burden or livestock). In forts precious space and resources were normally no the centre of the complex in the NE there is (like the on longer expanded unnecessarily lavish accommodation three parallel blocks) an elongated mound today, for higher-ranking army members. The similarities suggesting it equally contained a collapsed mud-brick between Sasanian and lateRoman military architecture in building (and was not an empty enclosure). In common the interior layout of forts and their defences (e.g. the with the littlewe know from the other "Barracks Forts", widespread use of projecting towers, see below "AT//. itmay have been a fourth barrack block, which, for Topographical Survey of Fort 4 ..." pp. 118-20) or in unknown reasons, was not detected by the survey. Faint are no linearbarriers unsurprising.There is reason to think traces of a NW-SE running linear anomaly along its a of this as one-way flow of influences,whether from east axis of symmetry, and possible traces of linear to or west fromwest to east; inmost periods of history,e.g. anomalies at a rightangle to it,may offer support for this the recentCold War, competing "superpowers" have used latterhypothesis. similar military strategies and technology, copying each The magnetometer survey also yielded interesting on other's innovations, improving themand adapting them insights into the defences of the fort.The higher concen to local conditions. The frequent comparisons between trationof high readings on all four sides of the fort (Fig. Sasanian and Late Roman monuments in this article are 21) suggests strongly thatfired brick was employed not thusnecessary to identifycommon and diverging trends in just on theNW side, where the fortabuts theGreat Wall, thedefensive architectureof themajor empires of the time, but on the three other edges as well. Further research is and to reveal the sophistication and significance of needed to establish whether this is true for all or a Sasanian architecture on a global scale. proportion of the forts or whether Fort 4, as one of the A fourth complex was detected in theNE. It forms largest, if not the largest, fort on theWall, was more an enclosure of c. 35 m. width and, though its traces fade heavily defended than some of its counterparts. The

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions FURTHER FIELDWORK AT THE GREAT WALL OF GORGAN 117

survey of Fort 1 had yielded no positive indications of military might and organisational skills probably would fired bricks being used at the sides other than the one have suitably impressed any members of nomadic abutting theGreat Wall (Nokandeh et al. 2006: 126-27 groups admitted into territorysouth of thewall, whether or with fig. 3), but furtherresearch is required to decide they were involved in trade political negotiations. whether this reflects architectural differences between Even if no more than the odd embassy were allowed Forts 1 and 4 or differences in the thoroughness of access, the impressive scale and regularity of the archi modern brick robbing. The extensive reuse of fired brick tecture of the fort would have sent out a deterrent a a inmodem funerarymonuments inFort 4 (Figs 22-23) is message to any neighbours contemplating raid into symptomatic of brick robbingwhich affected, to a greater state with such a highly organised military force. or lesser extent, all fortsand all sections of thewall. Without furtherfieldwork we cannot be certainwhether or The two high anomalies on either end of the central Fort 4 was special inhaving a gate on the enemy side road are particularly interesting. The south-eastern is whether most other forts on thewall were designed for marked by a dense scatter of smaller brick fragments on military or civilian trafficwith territories beyond the the ground, while none are visible on the surface in the wall as well. area of the north-western anomaly. The location of the According toKiani (1982: 15 fig. 9), Fort 4 was the anomalies suggests strongly thatwe are dealing with the largest military establishment on the entire wall. remains of gates. The south-eastern of these appears to However, Kiani's estimates for the dimension of forts have been robbed out, a theory strengthened by the are rounded, and may have been measured, on the basis presence of a robber trench just inside the fort and the of aerial photographs, from the bottom of the fort brick scatter continuing, in alignment with the road, platforms and inner edges of the surrounding ditch, down the slope of theplatform. By contrast, parts of the rather than the edge of the defended interior.Thus for foundations of the north-western one are possibly still Fort 4 his figures are 240 x 300 m. (= 7.2 ha.), while the preserved in situ. The width of the north-western interior is only c. 200 x 268-78 m. (= 5.5 ha.), i.e. a c. anomaly corresponds to that of the road and its 32% overestimation. For Fort 1,Kiani's dimensions had rectangular shape suggests thatwe may be dealing with equally been too high (by c. 23% [Nokandeh et al. 2006: deeper foundations than those of the adjacent sections of 128]). Without independently measuring the size of all thewall, perhaps not reached when thewall was robbed forts,we thus cannot be absolutely certain whether, in out. Normally, however, the gate opening does not relative terms,Fort 4 is indeed the largest of all forts on require deep foundations, if any foundations at all, as it theWall. However, Kiani's very accurate maps (Kiani carries no weight. Alternatively, it is thus possible that 1982: figs 1?8b) equally suggest thatno other forton the we may be dealing with the later blocking of the road wall exceeded it in size. Being the largest, or one of the leading out the fort, but this would only have made largest, forts it is possible that itheld a stronger garrison sense, had there been a gate before. The magnetometer and a higher-ranking commander than other forts. This survey thus suggests strongly that Fort 4 had a gate on need not mean that itwas necessarily the command the enemy side.Whether thiswas intended mainly for centre, as there are even larger and, quite possibly con military sorties or also to enable and control traffic temporary fortified sites in the hinterland (Kiani 1982: between the territories divided by the wall cannot be 39-61, figs 30-31; see "XVIII. The Strength of the decided on the basis of the evidence presently available. Sasanian Army ..." pp. 129-30). Nevertheless, the fort's The observation that none of the hollow ways leads unusual size and its location between two vulnerable towards the gate (see "VI. The Eastern Part of theWall points, where the wall had to negotiate deep river was ..." pp. 103-6, with Fig. 11) suggests that there only valleys (Kiani 1982: fig. 1; see "VI. The Eastern Part of limited trafficpassing through this gate. The hollow the Wall Between Forts 2 and 4" pp. 103-6 and "VII. a ways could, of course, date to potential post-military The Gorgan Wall in its Landscape: Discussion and phase (see "XIV Daily Life ..." pp. 120-6) and deliber Conclusions" pp. 107-8), allow for the possibility that ately bypass the obstacle of the ruined fort. Any the fort's defences and internal layout reflect a special potential visitors from the north using the route through function rather than being symptomatic for all forts on the fort would have had to walk past two military the wall. There are several architectural features buildings of over 200 m. length on either side of the observed in Fort 4 that were not noted at Fort 1 a central road. Such symbolic demonstration of Persian (Nokandeh et al 2006: 125-29), including:

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions 118 JOURNAL OF PERSIAN STUDIES

1. strongermagnetic anomalies at the upper edges of (Kiani 1982: figs 1-7 passim) appear to feature north the fortplatform suggesting thatfired bricks were south-orientated strips in the interiorof a range of forts, on also used the three sides of the fort,which do which may well represent earth mounds formed by not abut theGreat Wall, collapsed barrack blocks. This suggests thatmany, ifnot 2. the anomaly thought to represent a gate or a all, fortshad such barrack blocks in the interior.It seems blocked gate on the enemy side and likely that all larger forts on the Gorgan Wall were 3. the barrack blocks. provided with barrack blocks, possibly four each. Obviously, as the forts differed in their dimensions, so Further fieldwork is required to testwhether 1. and 2. are would the barracks. The number of soldiers in a larger unique to Fort 4 or, more probably, may have been fort is likely to have exceeded that in a smaller fort,even represented in a proportion or all of the forts.There may, thenumber of barracks or other long buildings may have however, already be sufficient evidence to explore been the same. The evidence available to date suggests possible parallels for feature 3. Fort 1, for example, that a series of forts on the Gorgan Wall were densely seems to have some features in common with Fort 4: the filled with barracks. The absence of clear traces of long central road and the postulated gate at the side opposite mounds or barracks in some forts is more likely the theGreat Wall. There are also traces of linear anomalies result of their remains near the surface having been parallel to the road, two in thewest and at least one in the levelled and destroyed by the plough?rather than of a east, whose position suggests that theymight represent random selection of forts having been left empty. the central walls of barrack blocks; faint traces of Nonetheless, on theCorona images, a high proportion of possible rows of rooms on either side of the easternmost forts (including Fort 1) show clear traces of parallel long linear anomaly may offer support for this hypothesis mounds, likely to represent barrack blocks. (Nokandeh et al 2006: 126 fig. 3). While the tentative extrapolation of the existence and location of barracks in Fort 1, from patterns observed in Fort 4, is far from XIII. TOPOGRAPHICAL SURVEY OF FORT 4 proven, itmight be hazardous to read toomuch into the AND ITS PROJECTINGTOWERS discrepancy between Forts 1 and 4 in quality and (HO,EWS, EST, JJ& SG) quantity of evidence for buildings. The quality of the survey of Fort 1was lower than that of Fort 4, partially Whilst scarcely visible on themagnetometer survey (Fig. because Fort 1 (unlike Fort 4) has been ploughed 21), it seems probable that Fort 4 was provided with a recently (all themore so, as the orientation of modern series of projecting towers. That the remains of such unsur plough lines in Fort 1, like any potential buildings, are towers do not form strongmagnetic anomalies is or parallel or at a right-angle to the fort's sides). The quality prising: heavy erosion at the edge of the fort brick was also partially lower for technical reasons (e.g. the robbing will have severely damaged or removed their instrumentbeing held closer to the ground inFort 4 after foundations, not tomention that the uneven ground made near or the systematic removal of obstructive vegetation). This survey the platform edge difficult impossible. On could conceivably account for the non-discovery of the SW, SE and NE sides of the upper edge of the fort are possible structures in Fort 1. The buildings in Fort 4 platform small hillocks visible, where the sloping sides as caused anomalies of only approximately 2.5 nT above of the fortbulges out, first observed by Julian Jansen theNE and SW the background level, suggesting that if themagnetome Van Rensburg and Seren Griffiths. On are m. ter had been held higher from the ground their signal sides thehillocks roughly 31 to 32 apart.Distances would not have been detected. on the SE side seem tobe identical, except that the SE gate at half the Yet, even if the possible indications for the presence seems to be flanked by two such features about of barracks in Fort 1 are dismissed as being too distance. There is littledoubt that these hillocks represent have been uncertain, satellite images demonstrate that Fort 4 was eroded towers. Their slight elevation may as a of human and animal traffic not unique amongst thewall forts in having been filled thrown into relief result on with barrack blocks. The Google Earth and Corona bypassing these obstacles, thus accelerating erosion such hillocks on the images of Fort 15 (Fig. 2) shows four distinct parallel either side of them.We counted eight and one each north-south orientated longmounds, two each on either SE side (including the two flanking the gate on SW and side of the central road, filling most of the interior. at the south and east corners) and nine each the ones east Kiani's accurate maps based on his aerial photographs NE side (again including the at the south and

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions FURTHER FIELDWORK AT THE GREAT WALL OF GORGAN 119

corners), i.e. 24 in total.We did not observe any traces of A projecting mud-brick tower on the west side of towers on theNW side of the fortabutting theGreat Wall, Fort 13was excavated by Kiani (1982: 19, fig. 1). Such but the systematic robbing of the fired bricks makes it projecting towers are typical for Sasanian fortsand town impossible to decide (short of excavation) whether this is walls (Boucharlat and Lecomte 1987; Huff 1987: 333; the result of theirabsence or systematic demolition. Naumann 1977: 3^38; Whitehouse 1972: 68-71), as The topographical survey at Fort 4 (Fig. 24), carried well as for contemporary late Roman fortifications. out on behalf of the Gorgan Wall Base (ICHTO), While we had not observed towers on Fort 1, the impressively confirmed the observations made on the question arises whether such hillocks could not have even an on ground, if eroded hillock the SE side is less been removed by ploughing. An inspection of Fort 30 a clear. There is suggestion that theremight have been confirmed that this compound was provided with on ten, rather than nine, projecting towers theNE side. projecting towers as well. On the best-preserved side, a The survey also created precise record of the i.e. the east side, hillocks were clearly recognisable and dimensions of the fort and the associated ditch system. theirdistance to each otherwas similar to that observed

' f / " . / / /~ ::: -{4., r : :

' t: J I ^ I ; * , 3 . ( i . i r I i 1

F/g. 24. 77*etopographical survey of Fort 4 on behalf of theGorgan Wall Base (ICHTO) byMr Hosseini andMr Hamidi, checked byHamid Omrani. The verticalarrow symbolsaround theedge of thefort platform mark theposition of thepreserved x remainsof possible projecting towers.The smallgrids of30 30 m. are identical to thoseon Figs 21 and 22; see Fig. 22for scale and north arrow.

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions 120 JOURNAL OF PERSIAN STUDIES

- -0.38 i-,-.-_?,, 0.39

I i f Artificialstep j -290 pOJ -2.58 / ^ 33? Northernroom ^ o -332 ^?

-, Section v^..-1 , j , throughi i 1 ^ (j 1 -245 /-^ , ?L , i >/^- m-10 Internal wall -oea internal i o.' o i wall ! "?'-? ! ?: ?lj-,-,- _

; Southern room i LamP -3.51j -3.80 .351

i j --!_-*j*jt\ % Artificialstep -f\^ ^ -1-07 Stone !1 i_

-1.35 I -1.28 -0.26 ?31 1_L_1 1m 0_

Plan of trench H The Great Wall ofGorgan Project 2006 Drawn by M.Hussein Zadeh

Fig. 25. Plan of Trench H, drawn byMaryam Hussein-Zadeh.

at Fort 4. The lesswell-preserved south side was equally the duration and intensity of their occupation, whether were made of fired bricks or sun-dried provided with projecting towers, while the west and they mud-bricks, north sides were too heavily damaged for detailed and their architecture in three dimensions. we confined observation. No fort has as yet yielded positive Due to considerable pressure of time, and indications for towers on the side facing theGreat Wall. ourselves to excavating two small trenches (H J) It is to be feared that at many forts the traces of (Fig. 22) under the direction of Hamid Omrani, Eberhard Sauer and Esmail Safari. Trench H could projecting towers, if still preserved, will disappear Only within the 2006 season. We first before they can be recorded, as a result of ploughing and be fully explored marked out a 1 x 3.5 test trench a well the use of agricultural machinery. m.-large through defined anomaly, thought to be a room division of the building NE of the central road, to establish the depth of we were XIV. DAILY LIFE INA SASANIAN FORT: the stratigraphy. It soon became clear that THE EXCAVATIONS IN FORT 4 dealing with a mud-brick wall reaching much deeper to (HO,EWS, EST & SG) thanwe had anticipated. There was not enough time excavate an entire room, but as the trench (H) had to be reasons of and to allow us to While the magnetometer survey provided us with a extended for safety understand thewider we it to 5 x 5 m. detailed plan of substantial buildings within Fort 4, only context, enlarged at and it covered an area of c. 4.10 x 3.25 m. at excavation had the potential to establish their function, the top,

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions FURTHER FIELDWORK AT THE GREAT WALL OF GORGAN 121

26) was almost imperceptible for the uppermost 0.50-0.56 m., but was distinct at deeper levels. The impression that the preserved wall extends right to the modern surface is, however, confirmed by two crack lines which formed when the soil dried out, and ran in direct alignment with either side of thewall (Fig. 26). In contrast to the lack of evidence for any surviving brick surfaces at a higher level, the bottom courses, notably in the north-east corner of the southern room, were well preserved (Figs 27-30). Between the rooms we found one arched window (or door?) opening (Fig. 27), first observed by Esmail Safari, which could not be fully explored for safety reasons, because of the substantial weight of the overhanging wall. We are thus unsure whether the stones (Fig. 27) mark the bottom of a c. 0.40 m. high window or whether it was an up to 1.40 m. high doorway, blocked later. During the latest phase of occupation (see below), all of this would have been below ground. The initial connection of the two rooms (whether passable for a person or not) would have facilitated communication between the rooms. Practical advantages in passing messages or goods from one room to another may have been the sole reason for the door or window. It is, however, also conceivable that it might have been intended as a control mechanism, so as to make it more difficult for ordinary soldiers to discuss with each other. Fig. 26. East-facing section of TrenchH, showing clearly the secretively any grievances 1.20 m.-wide mud-brick wall. Above c. 1.50m. of occupation layers (c. 3.54?2.07 below the benchmark, to be discussed below), a collapse was m. the bottom. Trench H thus encompassed most of the layer observed (to 1.50 below the benchmark), windblown division wall of two adjacent rooms, a smaller part of followed by slower sedimentation of the interiorof both rooms and a part of the eastern wall particles, as observed by Tony Wilkinson. The depth of of thebuilding (Fig. 25). the collapse layer and the observation that all buildings The room division consisted of a substantial 1.20m. detected by magnetometer survey form elongated as two wide mud-brick wall (corresponding approximately to mounds, first observed by Roger Ainslie (the ones now a the diameter of three bricks). In section (Fig. 26) itwas central separated by the central road, deep us we were visible as a vertical band of slightly darker colour hollow way), made wonder whether dealing or constructions. The (greyish-brown clayey silt with lime flecks) than the with single-storey double-storey course of bricks was found at deposits which had built up on either side of it, foundation of the lowest m. benchmark. The bottom of thewall presumably as a result of organic components in the 3.60 below the ditch mud-brick; we observed traces of possible straw imprints foundations and themodern surface, outside the are on same in themud-bricks. The discoloration extended right to surrounding Fort 4, average at about the themodern surface, but no individual bricks were found height above sea-level (i.e. c. 95 OD). This suggests that were at level. The barracks were not preserved in the uppermost 1.50 m, probably due to they erected ground a natural processes, such as insect and root activity; in fact standing on purpose-built fort platform, but what as a 24 and is some plant roots were found in over 2 m. depth and appears platform today (Figs 31) mainly sometimes they grew in the joints between mud-bricks. the result of the collapse of these mud-brick buildings within The colour difference to the deposits which had built up and the defences and the build-up of deposits on either side of thewall in the east-facing profile (Fig. them before. The observation that the collapse of the

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions 122 JOURNAL OF PERSIAN STUDIES

Fig. 27. Northface of the internalwest-east wall with preserved mud-bricks and partially excavated arched window i (or door?), drawn byMary amHussein-Zadeh. I

" i 1 \ A . 311N \ - I | -151 I -0.98W -151\ j i i i

ben I ~ i - . _ /^Arched%N )|T / \ j i-?\ ,-j 1 (-" (-\ < ]/window (or ? _J 1_) 1_) t_I door?)I ' i ,_tN CZZD U^-' v \LX~JCZ] [ ~~](-?-s L^Stones f \ ?!

i _J L-.:!?2_ cm O^^^^^^O TheGreat Wall of Gorgan Project 2006 Drawnbv M.Hussein Zadeh

- -

-1 _^-- 28. South the internal ! Fig. face of j west-east wall with preserved mud- I bricksand location ofOSL sample, i I \ drawn byMaryam Hussein-Zadeh. j 122W J -.- 178N ' 1 -1.78 I

JOSL sample from j mud-brick wall and i 6 soil samples _ i ? L ^^fp3 t^z^^^Ej'

PCZD -?^XZD (^ZICZD CZZDCZD j .rrrrrL- \~lrZ --' 1-'2r-: *-x-? Bottom of themud-brick wall \ ; s -3.80 s__. cm Haesgs^sg^i TheGreat Wall of Gorgan Project 2006 Drawnby M.Hussein Zadeh

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions FURTHER FIELDWORK AT THE GREAT WALL OF GORGAN 123

building appears to have raised the ground significantly I ~?1 in the area of Trench H suggests that itwas a very tI substantial building, unlikely to have been only a single I storey high. A potential second storey could have doubled the available floor space, thus having significant ramifications for the building's capacity to store food stuff and to accommodate soldiers and possible non 1.54N I combatant family members. Two-storey barracks, or -J--0.85W-1.78 ! barracks with a stairway leading to a second storey or a roof terrace, are also known from late Roman forts (Poulter 2006: 36, 41; Groot et al. 2006: 182, fig. 5.1). The lowest artefact (in both cases a piece of pottery) was found at 3.54 m. below the benchmark in the r southern room and 3.18 m. in the northern. In both i rooms artefact loss or deposition appears to continue ( until a floor level of c. 2.0112.06 m. is reached. At levels above this, artefact concentration is much reduced. It thus appears thatoccupation deposits reached a depth of U at least 1.47 m, suggesting a longer occupation of the fort (whether continuous or discontinuous) and, possibly, a longer period when theGreat Wall of Gorgan was maintained as a military barrier. The occupation of parts of two rooms need not, of course, reflect the history of all rooms, all buildings and all forts. It seems

Fig. 29. Westface of theeastern north-southwall of the The GreatWall ofGorgan Project 2006 barrack blocks with preserved mud-bricks in the southern Drawn M.Hussein Zadeh by room, drawn byMaryam Hussein-Zadeh.

Fig.50. 77ze north-east ^^^^^^^^^^^^^^^^^^^^^^^^K^/k^^U/^^^^^^^^^^^^^^^^K^ cornerofthe southern ^^^^^^^^^^^^^^^^^^^^^^^^^^^HESj^^H^^^^^^^^^^^^^^^^^^^Hj room mwd-^^^^^^^^^^^^^^^^^^^^^^ ^^ ^^^^^^^^^^^^^^^^^^^^^^^H

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions 124 JOURNAL OF PERSIAN STUDIES

Fig. 31. Fort 4 (like theother forts on theGorgan Wall) forms a substantialplatform, here thesmall side seenfrom thesouth-east. Note thehuman scales (between thecows) inclose vicinityto thefort.

Fig. 32: 1. Lamp found in thesouthern room; 2. Lampfound in thenorthern room: 3. Fragment of a glass vessel (H.006), drawn byMohaddeseh Mansouri Razi. nonetheless improbable that two rooms selected at exclusively as stables. Needless to say we should not random would reveal an entirely atypical occupation exclude the possibility that other rooms in the same or pattern, since an adequate number of soldiers had to be one of the other buildings may have served as stables or maintained on all sections of the wall as long as the store rooms, but it is equally possible that all rooms system was meant to function effectively as a defensive housed soldiers, with or without their families. barrier. Interestingly, one lamp each was found in either The relative abundance of pottery,which we hope to room, both a similar level (3.23 m. below thebenchmark study inmore detail during the next season, suggests in the southern room and 3.14 m. below in the northern), strongly that the rooms in question were used from the each immediately next to thewall (Figs 25 and 32.1-2), start as accommodation and not, for example, perhaps as a result of each having dropped off some

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions FURTHER FIELDWORK AT THE GREAT WALL OF GORGAN 125

TABLE 1.Radicarbon dates from occupation levels in the northern room inTrench H.3

Small No. Material Depth below Uncalibrated Calibrated date BM findno. date (95.4% probability)

H.061 OxA-17015~ Bone ^2T3 l468?27BP jAJ). 550-644 H.020 OxA-17uT4~Bone ^Z67 T501?26BP AT). 444-636 FL027 OxA^170Tl Charcoal(willowr"T70 1489+29BP A.D. 471-644 or poplar) ~ H.096 OxA-17012~ Charcoal (alder) -2.96 1399127 BP A.D. 604-66

narrow ledge. The lamp from the northern room thatone ormore of the samples represent organic matter apparently had broken in situ, its fragmented nozzle, re-deposited long after the relevant organism's end of life with wick-hole, still being attached to the lamp when and should await further scientific dates from the found. While we still await the results of scientific uppermost occupation layers before postulating a analysis of thepossible traces of the lamp fuel,fragments specific date for the abandonment of the room with of charcoal from the occupation layers indicate that the absolute confidence. One is nevertheless inclined to fort occupants used beech (or possibly plane), alder and think that the absence of any positive evidence of activity willow or poplar as fire wood (see "XIX. 2. Charcoal beyond themid-seventh century amongst the samples is Analysis" pp. 132-3), shedding lighton the types of trees unlikely to be coincidental. Possible historical contexts growing in the surroundings at the time. for a potential abandonment between A.D. 604 and 644 Bones were only found from 2.91 m. upwards in the are provided by themajor wars the Sasanian Empire southern room and 2.87 m. in the northern, suggesting fought against theByzantine Empire fromA.D. 603-28 that the rooms were, initially at least, kept reasonably and against theArabs fromA.D. 636 onwards. tidy and thatno food waste was allowed to accumulate. By contrast, the best parallels for one rim sherd of a There is no evidence formuch organic matter having glass vessel (H.006: Fig. 32.3) 2.40 m. below the been allowed to rot in situ.Much of the occupation levels benchmark from the adjacent southern room (i.e. at a consist of re-deposited loess, perhaps representing one or lower level than thatof the storage vessel from the same more re-flooring(s) to raise theground. The lack of rotted room?and also lower than the bone sample of A.D. organic matter, except for the odd charcoal fleck,made it 550/604^44 from the northern room), are, according to impossible to visually differentiate between different Birgitta Hoffmann, as late as the ninth or tenth occupation levels. That there is at least a second one is centuries.4 This raises intriguingquestions: was thewall a suggested by partially preserved red storage vessel maintained as a military barrier much longer thanwe (H.025) with organic residue from itsbottom, apparently thought, or were perhaps some of the forts transformed interred in situ (as suggested by the perfect preservation intovillages? Are thepossible traces of settlement SE of and horizontal level of its bottom part) at the northern Fort 4 Sasanian or early Islamic or both? Of course, one room. was m. edge of the southern Its innerbottom 2.31 should not base the chronology of the later occupation below the benchmark and itshighest preserved part 2.07 of Fort 4, let alone other forts, on the typological dating m. The latterfigure, similar to the levelwhere finds peter of a single find from a single room. An attribution of the room out in either (and notably the southern room), may piece to the Sasanian period should not be excluded; indicate the approximate level of the floor surface during indeed, while more research is desirable, unpublished the last phase of the room's occupation. Storage vessels local parallels may well point to a Sasanian date. Unless have also been found interredat ground floor level in the in future other finds or scientific dates should barrack fifth-/sixth-centurytwo-storey blocks of Dichin corroborate late occupation, it seems more plausible to inRoman Moesia/ 2006: Bulgaria (Poulter 36, 41). assume that the rooms ceased to be occupied in the first A sequence of radiocarbon dates from occupation half of the seventh century. levels in thenorthern room was (Table 1) suggests that it We also found a yellow glass bead of 5mm. diameter until at leastA.D. 604. A bone from near occupied the top (H.055) 3.27 m. below the benchmark and just one has a terminus ante (H.061) yielded probable quern of object of copper alloy, a square plaque with rounded A.D. 644 for the formation of the highest occupation edges of 9 mm. diameter (H.092) 3.46 m. below the We of deposits. cannot, course, exclude the possibility benchmark; these three objects came all from the

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions 126 JOURNAL OF PERSIAN STUDIES

southern room. To what extent the scarcity of metal The magnetometer survey by Abingdon the finds reflects a real objects amongst shortage of Archaeological Geophysics and the ICHTO (Fig. 21), mud-brick for not metal, architecture, example, requiring prominently displays recent plough lines, but yielded no iron and to nails, what extent the potential predominant evidence for any extramural settlement. This, however, use of items not or larger easily lost leftbehind, is hard proves merely that there are probably no remains of to decide. We do not know whether or not the garrison substantial structures,made of materials of different as comprised heavily armed soldiers, the typical helmets magnetic properties to the subsoil. A pottery scatter or coats of mail that would identify such soldiers (cf. around the fort, notably on its SE side together with Overlaet would not have or 1993) easily been lost left associated mounding typical for settlement sites,might behind or broken our into recognisable parts. While derive from a settlement outside the defended perimeter. too sample is small for definite conclusions, the The SE side was, of course, the point where the road led comparative infrequency of metal finds has also been out of the fort and thuswas themost obvious area for observed at Sasanian Merv (Simpson 2004: 235). settlement. Lightly-built structureswould probably not A second trench was (J) marked out to encompass be detectable and even less so on recently ploughed the south-western enclosure of the NE complex and fields.Without extensive excavations in the extramural some of its interior.Work on the trench was not areas itmay not be possible to prove the presence of season completed in the 2006 and we hope to do this in settlement outside the defended perimeter. Should future 2007. Only a brief interim survey will therefore be fieldwork fail to find traces, it may nevertheless be offered here. To our surprise, no featurewas found in the impossible to disprove the existence of extramural alignment of the south-western side of a large settlement, as we always have to allow for the possible rectangular anomaly, thought to be an enclosure. It presence of untraceable yurts (Stronach 2004) or mud seems improbable that magnetometer survey would brick buildings destroyed by ploughing. Fortified a have picked up deeply buried ditch. The bricks of Sasanian sites in theMughan Steppe in north-west Iran, substantialmud-brick walls inTrench H only showed up some of themwith a partially rectangular shape similar a clearly at deep level (and thewall itself only became to the Gorgan Wall forts and supplied with water by apparent as a soil discoloration, once the soil had dried canals, have all yielded evidence for extensive out for several days); thus it seems possible that there extramural settlement (Alizadeh and Ur 2007: 152-54). was a mud-brick wall, not recognised during our short Despite the inconclusive results, a few more general season. Full excavation of the trench will hopefully considerations on the use of extramural areas may not be yield an answer. Until such results are available we have out of place. It is perfectly conceivable that family to rely on themagnetometer survey and observations on members were admitted into the safety of the defended the ground, which suggest that the fourth complex compounds, rather than soldiers living in the barracks on contained a long building as well (see "XIL The their own. That the small finds assemblage fromTrench Buildings in a Sasanian Fort" pp. 113-8). H does not allow us to prove or disprove the presence of women and children is unsurprising. How difficult it is to establish whether theywere present or absent with any XV.MAGNETOMETER SURVEY OF THE degree of certainty is best demonstrated by the fact that, EXTRAMURAL AREA OF FORT 4 despite extensive fieldwork in numerous Roman forts (EWS,HO, RA, EST, TW,ME, SG & AN) and fortresses and some work on distribution of "gendered" artefacts (Allison et al 2005; cf.Groot et al Whether or not therewere contemporary undefended 2006: 161), notably a sizeable number of women's and settlements outside the forts is essential for assessing the children's shoes in barrack blocks at Vindolanda (Driel security situation on the Great Wall and any potential Murray 1997), opinion remains divided between those economic function of the wall forts. Did Fort 4, like who argue for the frequent presence of partners and many of its early Roman (though less often its contem children of rank and file soldiers (James 2001: 80, porary late Roman [Crow 2004: 109; Southern and 83-85; Niinnerich-Asmus 2006), and those who argue Dixon 1996: 136]) counterparts attract camp followers, that only the officers' wives would normally have lived e.g. merchants, not tomention the soldiers' families? If within the defended compounds (Crow 2004: 81-82, so, were the soldiers' wives and children permitted to 104, 109; Debrunner Hall 1994: 221; Sommer 1984: livewithin the fort or did they have to stay outside? 30-31, 52; Speidel 1997). One is inclined to think there

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions FURTHER FIELDWORK AT THE GREAT WALL OF GORGAN 127

is an increased probability of their intramuralpresence in territoriesceded by Rome to Sasanian Persia in thepeace forts without contemporary settlement outside (cf. treaty of A.D. 363. The information provided by the Southern and Dixon 1996: 85, 136). ancient sources is insufficientlyprecise to tellwhether Not a single coin has as yet been found at any of the theywere within Sasanian or Roman territoryafter A.D. Wall forts, suggesting that, contrary to modern views 363 (Ammianus Marcellinus 25,7,9; Zosimos 3,31,1; and possibly other units elsewhere (Puschnigg 2006: Boeft et al 2005: 233-39). With the possible exception 21-22), the soldiers were not paid in hard currency and of Tell Bati and Saibakh, whose location in the frontier did not normally participate in a monetary economy. zone neither excludes an attribution to thePersian nor to This might add credence to the fourth-centuryhistorian theRoman Empire, thereare, to our knowledge, no exact Ammianus Marcellinus's (23,6,83) claim that the parallels within theRoman Empire in itspost-A.D. 363 Persian infantry received neither pay nor gifts. Of boundaries. Since Tell Bati and the fort at Saibakh have course, such a generalised statement by a historian, never been excavated, the latternow being buried under whose work pre-dates the construction of the Gorgan a modem village and the former "completely defaced" Wall, should not imply that the soldiers serving there (Oates and Oates 1990: 229-30), their date and the necessarily consisted mainly or exclusively of infantry. cultural association of any artefacts cannot easily be The absence of coinage implies nonetheless that they verified. Oates (2005: 85, 145; cf.Oates and Oates 1990: probably received remuneration in kind. There is 230) postulated a Roman origin forAin Sinu I and a date nothing in the layout of the forts to suggest that they between the Severan campaigns of the lateA.D. 190s were involved in farming.While they are still likely to and Ardashir's invasion ofMesopotamia in c. A.D. 237. have participated in goods exchange with traders or While the occupation debris in the barracks and gate craftsmen providing specialised goods or services, we chambers were only a few centimetres thick and "almost do not know whether the latter settled next to the fort. barren of objects", sherds similar towares fromHatra of the firsthalf of the thirdcentury A.D. were found on the floor and embedded in thewalls (Oates 2005: 85). Oates XVI. SASANIAN FORT DESIGN (2005: 85) cited two coins, minted inA.D. 218/222 and (EWS & HO) A.D. 222/235 "found inside the camp in debris just above the floor or ground level" as furthersupport for an The phenomenon of fortsbeing occupied exclusively or occupation period within the postulated parameters. almost exclusively by barracks or other elongated St John Simpson (1996: 90-92) was the first to structures is not unique to the Great Wall of Gorgan. challenge this dating proposal and to argue that the Three forts inUpper Mesopotamia, Ain Sinu I (Oates "Barracks Forts" need not be Roman, but could be 2005: 81-92; Kennedy and Riley 1990: 168-69, Sasanian. He rightly stressed that the postulated third 213-15), Tell Bati (Poidebard 1934: 150, pi. CXXXIX) century dating restedmainly on residual and unstratified and a fortat Saibakh, 3.5 km. fromTell Brak (Poidebard material and argued that a reoccupation under Sasanian 1934: 144, pi. CXXII; Oates and Oates 1990: 229-30), rule was highly probable. The sherds "embedded in the share key features with Forts 4 and 15 on theGorgan mud-brick" (Oates 2005: 85) must indeed be "residual Wall (and probably the other forts on thewall): sherds re-used in the construction" as Simpson (1996: 92) observes; had the barracks been built at a virgin site, All theirbuilt-up interior space appears to be taken then few, ifany, sherds should have found theirway into up by such parallel buildings, consisting ofmud themud-brick. Should Oates nevertheless be correct in brick walls (which, after their collapse, have assigning the barracks ofAin Sinu I to the thirdcentury formed longmounds). (even ifhardly immediately after the Severan conquest), All have corners. square then it is still impossible to tellwhether we are dealing At some seem to now. least form distinct platforms with short-lived Roman or early Sasanian barracks; pottery chronology does not permit us to tellwhether the Kennedy and Riley (1990: 213-15) have appropriately small assemblage pre-dates or post-dates Ardashir's called them "Barracks Forts". Ain Sinu is in an area held invasion and temporary conquest of the area. As the Rome in the by temporarily late second and in the third excavation report does not reveal which precise sherds Tell are century. Bati and Saibakh in the immediate were found in the occupation layers "almost barren of zone frontier between the Roman Empire and the objects", there is no certain way of establishing whether

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions 128 JOURNAL OF PERSIAN STUDIES

the Ain Sinu I barracks date to a few decades or a imply a lack of functional versatility. If, for example, possibly a few centuries after the establishment of the each group of soldiers occupying a room (or two rooms earliest compound at the site. Simpson's preference for in a double-storey building?) kept their proportion of a later date may be further strengthened by Ain Sinu I food supplies within their living quarters, then there and Fort 4 on the Gorgan Wall, in addition to the was no need for central granaries. common features of the "Barracks Forts" listed, sharing two other characteristics: XVII. THE SIZE AND NATURE OF THE The shape and size (43 x 43 x 9 cm.) of the square GARRISON OF FORT 4 (EWS & HO) mud-bricks, used for the barracks and outer wall at Ain Sinu I (Oates 2005: 84), are remarkably similar The evidence, throughmagnetometer survey, thatFort 4 to those used for the Fort 4 barracks and the fired contained threebuildings with an estimated 90-96 rooms brick from theGreat Wall of Gorgan. each, not counting the fourth complex, eradicates any Some of theAin Sinu I barracks were located doubt that the fortwas densely occupied. It is likely that within a walled courtyard (Oates 2005: 83-84 with most, ifnot all, the rooms in these threevirtually identical fig. 6) and the geophysical surveymay suggest that buildings served as human accommodation as proven for the SW barrack in Fort 4, as well as the unidenti the two rooms explored in Trench H. In early Roman fied NE complex, were within similar enclosures barrack blocks each unit consisting of a double room (the (Figs 21-22). front room for storage and the back room for living) formed a contubernium (i.e. a sub-unit sharing a tent on Unfortunately, there is not enough evidence to decide campaign) of eight soldiers (Johnson,A. 1983: 166-67, whether all three "Barracks Forts" in Upper 171-76). In Late Antiquity, this type of double room was or Mesopotamia are Sasanian, whether they are all Roman reduced to a single unit,with without subdivisions, but or whether both empires built this type of fort in a these single units were not necessarily smaller than the fought-over contact zone. Neither can we be certain earlier double-rooms (e.g. Crow 2004: 100-4). The whether the "Barracks Forts" all date to Late Antiquity substantial variations in size amongst both early imperial or whether some are earlier.What is beyond doubt now, (Johnson,A. 1983: 166-76; Petrikovits 1975: 36-43) and however, is that the Sasanians (exclusively or non late imperial barrack rooms (Johnson, S. 1983; Gregory to a exclusively) constructed this type of military installa 1996; Groot et al 2006) make it impossible give tion in the fifth and/or sixth century A.D.. It is worth typical size, but it is fair to say that the Sasanian single rooms were within noting here that, while there are no exact parallels in Fort 4 (roughly 5x7 m.) perfectly furtherafield in the Roman world, there are also late the range of their lateRoman single-room and even their Roman installations, such as Cologne-Deutz (Precht early or high imperial double-room counterparts. It thus 1975; cf. Southern and Dixon 1996: 141), with square seems reasonable to assume that theyprovided shelter for to corners and no buildings others than barracks in the a similar number of occupants, i.e. up eight soldiers. fortress at El interior.Why therewere no recognisable administrative The Late Antique Roman legionary east Dead Sea inmodern Jordan contained buildings or more lavish accommodation for the higher Lejjun of the a rooms in the and ranks has to remain a matter of speculation. Of course, barracks with minimum of 256 early some of the control centres and administrative installa mid-fourth centuryA.D. and was slightly smaller: 4.6 ha. for its tions could have been located in the larger and as yet as opposed to Fort 4's 5.5 ha. Estimates garrison between and soldiers et al unexplored hinterland fortresses (see "XVIII. The vary 1,000 2,000 (Groot 2006: Parker 2000: Southern and Dixon 1996: Strength of the Sasanian Army ..." pp. 129-31). 183; 134; a below cannot be excluded Confining building in forts to basic accommodation 31-32), though figure 1,000 The estimate is based on the certainly helped to keep the perimeter to be defended (Tomlin 2000). higher were room small. The exclusive use of the same type of standard assumption that there still eight soldiers per "Barracks Fort" Ain Sinu I ised building would also have accelerated construction unit. The above-mentioned a to Fort 4. IfOates's reconstruc works. Avoiding unnecessary architectural complexity provides closer parallel tion is therewere 528 rooms of size or probably reflected a military strategy conscious of the correct, unequal two rooms each and 88 units of one room fact that speed could make the difference between 220 units of fort success and failure. Architectural uniformity need not each (not counting further chambers around the

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions FURTHER FIELDWORK AT THE GREAT WALL OF GORGAN 129

x m. at Deutz walls). Of these 308 room units Oates (2005: 83 fig. 6, barrack blocks of 11.5 57.4 each (Precht have a smaller 89-90) assumes that 28 served junior officers and that 1975; 1987)?should had garrison. between and is the remaining 280 were occupied by eight soldiers each, While any figure 1,000 2,000 possible or i.e. 2,240 soldiers in total (or still over 1,100 on the, as (and a figure slightly below theminimum above the Oates stresses himself, unproven and highly speculative maximum estimate not inconceivable), round figures, we or men four or assumption thathalf may have served as stables). If whether 1,000 2,000 (i.e. roughly eight per assumed that each individual room in theAin Sinu I fort room, leaving a few rooms for other functions),may be we about the was occupied by four soldiers (or double room by eight) easier to reconcile with the little know men on average, then the total number of soldiers would be c. organisation of the Sasanian army. One thousand a 2,000, ifby eight, c. 4,000. If the same average number would correspond to thepostulated size of Sasanian unit of soldiers shared a room in both forts, the garrison of called gund, and there is in any case circumstantial Ain Sinu Iwould have been about twice the size of that evidence to suggest that the organisation of the Sasanian was on men of El-Lejjun. In terms of area, Ain Sinu Iwas, with 10.6 army based units of 1,000 and multiples ha. (Kennedy and Riley 1990: 215), roughly twice the thereof (Widengren 1957: 162-63; 1976: 282; Huyse size of Fort 4, and both compounds were densely 1999: 133-34; cf. Shahbazi 1987:497-98; Gyselen2001: one occupied. It is thus possible that Fort 4, being of 20-22). Admittedly, the postulated division of either comparable size and number of rooms to El-Lejjun and or two units over threebarrack blocks seems odd (unless about half the size and room number as Ain Sinu I,was there is a fourth,as yet undetected, barrack block in the designed for 1,000 or 2,000 soldiers (which implies c. NE complex), but itought tobe admitted thatwe know so four or eight men per room, i.e. a similar occupation littleabout the organisation of the Sasanian army thatany density as inRoman barrack blocks). of the above assumptions are speculative. A comparison with earlier Roman forts in thewest As thewall was a threat frommounted armies, one adds furtherstrength to the assumption that 1,000-2,000 wonders whether it would not equally have been soldiers may be the right order of magnitude. Fort 4's defended by cavalry, or at least partially mounted, units. size (5.5 ha.) corresponds roughly to the average space If so, the question where the horses would have been needed for 1,000 horsemen or 2,000 foot-soldiers in the kept (e.g. in some of the rooms of the three barrack second or third century A.D., though it ought to be blocks or in the fourth compound or outdoors) is as yet conceded that there are considerable variations in the unresolved. Research on Roman fortshas demonstrated size of Roman forts for units of similar strength and that horses were often kept in the same room units as composition (Johnson,A. 1983: 292-93, cf. 22 tab. 1). horsemen and that the absence of separate stables by no Fort 4 is roughly a third to a fifthof the average size of means disproves the presence of substantial numbers of a high imperial legionary fortress for some 5,000-6,000 horses within a defended compound (Hodgson 2002). A men, an observation which might offer further support single equid bone from the small assemblage from the for a garrison of 1,000-2,000 men. Closer in time and Sasanian-period ditch fill at Fort 9 (see "XIX. 1 Animal layout than legionary fortresses of the principate is the Bones" pp. 131-2) does not resolve the question, early fourth-centurybridgehead fort at Cologne-Deutz, especially since we do not know to what extent this filled exclusively with barrack blocks, 16 in total. assemblage represents food or butchery waste, as (While the reconstruction of the interior is based on the opposed to disposal of the remains of mounts or beasts excavation of a series of small trenches, covering just a of burden. tiny fraction of the area, the regular size and spacing of inner buildings is indeed best explained with theDeutz being a kind of "Barracks Fort" as well.) Precht (1987: XVIII. THE STRENGTH OF THE SASANIAN 515) postulates that theremay have been a century each ARMY IN THE GORGAN PLAIN (EWS& HO) in 12 barracks blocks, the four central ones being set aside for the officers and administrative functions, Despite many as yet unresolved questions (e.g. the a c. men. adding up to total garrison of 1,000 There is no number of storeys per barrack block and the proportion reason good why Fort 4, whose interiorwas three times of floor-space reserved for storage or horses, or larger than that of Deutz (1.8 ha.)?even if the potentially occupied by civilians), itnow seems possible combined area occupied by itsbarracks was of a similar to venture at least rough estimates of the total size of the order of magnitude to that of the 16 narrowly spaced Sasanian army stationed at this frontier.According to

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions 130 JOURNAL OF PERSIAN STUDIES

Kiani 15 who (1982: fig. 9), slightly overestimates the not be discussed here, as no information on their date size of the two forts we have so surveyed far, the could be found. The square plans of the listed fortresses combined size of all 33 forts known to him was 15.7 leave littledoubt that theywere carefully planned instal times that of Fort 4 alone. If all fortswere as lations and that theirouter densely perimeter at leastmust belong as Fort and ifFort 4 housed a of c. to a occupied 4, garrison single phase. Four of the five compounds include a to men 1,000 2,000 (as suggested above), then the total citadel, invariably located in a corner. Further fieldwork the 33 fortswould have been between c. is garrison 15,700 required to testwhether these citadels are contempo and 31,400 soldiers. Ifwe assumed that therewere artificial eight rary mounds and lookouts, or whether they are soldiers in each room often to case (as thought be the in ancient tepes incorporated intonew fortresses. The latter Roman and that therewere 96 rooms each compounds) development is likely to account for Qaravol Tappeh in the three the estimate for the of Fort 4 blocks, garrison being situated in Fort 13 on the Gorgan Wall (Kiani could be raised to even a thirdmore was 2,304 (or if there 1982: 43-^4, figs 3, 32, pi. 4,3). fourthbarrack block in theNE and that for all complex) The observation that the rectangular installations forts to over on the other we 36,000. If, hand assumed (excluding rectangular tepes), at least as far as included that the forts were as as those on occupied densely inKiani's survey (Kiani 1982: 39-61, figs 30-31), all Hadrian's Wall (c. 9,500 men in 15 forts totalling 28.7 ha. seem to be south of the wall and seem to have been and Dobson 2000: [Breeze 54, 159-62]), then the occupied in the Partho-Sasanian period suggests that on the garrison Gorgan Wall would have been in the theymay have formed part of the same, defence-in order of the 30,000 men?assuming combined size of depth, system. Should this assumption be correct, then the fortswas 90 as toKiani's approximately ha., opposed the number of Sasanian soldiers stationed in theGorgan 15 of 113.1 ha. Even (1982: fig. 9) (over-estimate the Plain could have been substantially higher than lowest of the above estimates for the number of soldiers suggested above (even if already the lower estimate, stationed at one frontier that single implies the Sasanian excluding any hinterland garrisons, suggests a well one Empire may have had of the strongest and best remarkably strong army). It would be premature to armies of the ancient In to come organised world. order postulate that these hinterland compounds were to a more reliable estimate, it will be necessary to necessarily all mono-functional and all contained continue the of the interiorsof thewall forts investigation barrack blocks. Not all need have been occupied mainly and to establish whether or not Fort 4's or density, type and exclusively by military personnel; the polygonal 338 size of internal were on buildings typical for the forts the ha. Dasht Qaleh (Kiani 1982: 48-52, figs 30-31, pis wall. Circumstantial evidence, notably from aerial 17,2-22,1) has plausibly been interpreted as a city and photography and satellite imagery (as discussed above), an explanation as civilian settlements may also be that it was. suggests probably correct formany agglomerations of irregular shape or The estimate for the size of thewall's garrison may, long-lived tepes. The square compounds would, of be too however, substantially low, if square fortresses in course, have been more suitable for being filled with the hinterland are contemporary, had military garrisons rectangular barrack blocks and themilitary interpreta and a similar occupation density. Qaleh Kharabeh, for tion gains in strength, if the first or only peak in pottery a example, reached size of 56 ha. (i.e. c. ten times the deposition is reached in the Sasanian period. size of on the largest fort the wall) and the surface An aerial photograph of 56 ha.-large Qaleh was pottery suggests that it occupied in the Parthian Kharabeh (Kiani 1982: pi. 15,2, cf. fig. 7) shows the and/or Sasanian periods (Kiani 1982: 43, figs 7, 30-31, typical "stripy" pattern, four or five parallel stripes in the or pis 15,2-16,1). Other square, almost square, north-east quadrant being particularly clear; while any fortresses with Partho-Sasanian pottery include the 25 interpretationwithout inspection on the ground has to ha. largeQaleh Gug A (Kiani 1982:42, figs 30-31, pi. remain preliminary, it seems possible that these stripes 14,1), Qaleh Yasaqi of similar dimensions (Kiani 1982: represent a large number of barrack blocks. On the 53, figs 30-31, pi. 22,2), Qaleh Daland of 33 ha. size assumption of a similar dense occupation as Fort 4, its (Kiani 1982: 58-59, figs 30-31, pi 30,1) and Gabri garrison could easily have been up to ten times larger Qaleh of 42 ha. (Kiani 1982: 56-57, figs 30-31, 39, pis (i.e. 10,000-20,000 soldiers, temporarily or 26,2-27,2, 28). Four further compounds of square or permanently, in this one fortress?). This is particularly rectangular shape (Kiani 1982: figs 30-31) may astonishing as there were no longer any fortresses, conceivably belong to the same system as well, but shall anywhere near this size, in theWestern and Eastern

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions FURTHER FIELDWORK AT THE GREAT WALL OF GORGAN 131

Roman Empire in Late Antiquity. Kiani's aerial just on land bordering the Caspian Sea, but also we photographs do not show the same stripes in the other elsewhere. To the local garrisons must add the field men square compounds, but fieldwork will be required to test army, thought to have comprised 50,000 whether this is due to their original absence or due to (Widengren 1976: 296-97; Greatrex 1998: 57-58). The modern cultivation in the fertile Gorgan Plain having 2006 season has yielded fascinating new evidence to obliterated any traces on the ground. suggest that the Sasanian army in the fifth and sixth Should we be right in thinking that at least some of centurieswas substantially stronger, in terms of numbers the square compounds served as fortresses, thenwe will and organisation, than had previously been thought. Just have to reassess not only the strength of the Sasanian how little we had known until very recently is army, but also the military strategy employed in the exemplified by Greatrex's (1998: 52; cf. Puschnigg Gorgan Plain. While there are also smaller installations, 2006: 22) recent doubts as towhether or not the Sasanian notably the Sasanian fort of Tureng Tepe (Boucharlat Empire had a standing army at all. The barrack blocks and Lecomte 1987), several of the forts in the and the sheer scale of themilitary installations in the hinterland are substantially larger than those on thewall Gorgan Plain not only eradicate any such doubts, but and are often in a more advantageous strategic location. suggest that the Persian army was an equal to its late Thus they are more likely to have contained the Roman counterpart?and in some respects perhaps more command centres. Positioning the strongest military than an equal. Itwould be wrong to ascribe the ability of units in the hinterland would make perfect sense and is thismajor power tomaintain and expand itsvast Empire paralleled by the strategy employed by the Roman primarily to its rivals' (e.g. theEastern Roman Empire's) army. In case hostile forces threatened any particular weaknesses; the inner strength and dynamism of the section of theGreat Wall of Gorgan or had managed to Sasanian Empire, facing no fewer external threats along break through, troops in the hinterland could have more its extensive frontiers (such as those adjoining the easily and rapidly repelled such an attack thanwould Caspian Sea) than any of its rivals, ismore crucial for have been possible by assembling troops from other explaining its success. forts on thewall. More fieldwork will be required to test whether Qaleh Kharabeh indeed was densely filled with barrack XIX. APPENDIX blocks and, if so,whether or not this isparalleled in some or all of the other square compounds. Substantial barrack XIX.L Animal Bones (RT) blocks are unlikely to be designed for a temporary garrison, e.g. troops in transit,more likely to be accom This report provides an analysis5 of the animal bones modated in tentsor yurts. Even the evidence available at from ditch fills from Trench B near thenorth-west corner present suggests that the Sasanian Empire was able to of Fort 9 of theGorgan Wall, Iran. These bones all derive mobilise substantial manpower and that the permanent from the middle and lower levels of deposit B.006 garrison in theGorgan Plain alone numbered many tens (Nokandeh et al 2006: 143 fig. 22), sealed well beneath of thousands of men. It seems highly likely that an thewall tumble (B.013, B.004 and B.010) and from the Empire stretching from Mesopotamia into the Indian lower levels of B.003/005, and are likely to represent Subcontinent would have held strong local garrisons not rubbish deposition in the dried-up ditch by a latergarrison

TABLE 2. Representation of species

Sasanian ?Sasanian Late Sasanian/ Medieval Total earlymedieval

Cow (Bos taurus) 15 4 Sheep/goat (OvistCapra) 14 5 2 21 Sheep (Ovis aries) 1 1 Equid (Equussp.) 1 1 Unidentifiable 22 12 10 473 Tbtal 42 17 13 3 75

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions 132 JOURNAL OF PERSIAN STUDIES

of Fort 9.A radiocarbon date ofA.D. 433-610 (Nokandeh Despite the small size of the assemblage, and the et a al. 2006: 161-62) for sheep mandible from this poor preservation a clear emphasis on domestic species, cluster proves an attribution to the Sasanian period. particularly sheep/goat is apparent. The animal bones In total, twenty-six secure, non-modern, contexts deposited in the ditch fills seem likely to represent contained seventy-five fragments of hand-collected activities related to the preparation of animals for animal bone, of which only twenty-eightwere identifi consumption. A more detailed examination of the faunal able to species. The majority of the bones from the site remains from sites located along theGorgan Wall will were poorly preserved, exhibiting extensive cortical fill an important gap in our understanding of the nature degradation consistent with acidic and/or sandy free of past human-animal interactions in this region, partic draining soils.6 Analysis of the species represented ularly since previous research in this area has largely (Table 2), reveals that the identifiable specimens derived focused on earlier periods of occupation (see, for from only three domestic taxa, with sheep/goat example, Bocherens et al. 2000; Young 2007; Zeder and predominant (79%). Hesse 2000). Examination of the body parts represented for Sasanian and probably Sasanian contexts7 suggests the inclusion of both primary and secondary butcherywaste, XIX.2. Charcoal Analysis (RG) with an emphasis on the former?although taphonomic factors could have led to the relative over-representation Charcoal samples from the fill of the kiln in Trench G of teeth and the denser lower limb elements. Two (see "IX. Excavation of a Brick Kiln" pp. 110-2) and Sasanian sheep mandibles were sufficientlycomplete to from occupation deposits fromTrench H, in thebarracks provide an assessment of age and produced ages of 2-3 of Fort 4, were submitted for species identification prior and 3-4 years. Where the state of fusion could be to C14 dating. Charcoal was sparse in most of the recorded on post-cranial bones, all but one specimen (the samples. The charcoal fragments were mostly very epiphysis of a Sasanian equid metapodial) were found to small and frequently degraded. The samples were be frilly fused.8No butcherymarks or carnivore gnawing prepared using standard methods (Gale and Cutler was recorded; however, given the poor cortical preserva 2000). Anatomical structures were examined using tion this is perhaps unsurprising. incident light on a Nikon Labophot-2 compound

TABLE 3. Charcoal ofwell stratified samples recovered in 2006

SampleContext & description Taxa identified TrenchG

G.004:O002 fill of kiln Too degraded to identify G.012:G.003 fill of kiln Insufficient to identify G. 004G.010: fill of kiln, sample C14-dated ?Chenopodiaceae, narrow stems,very degraded G.016:O005 ofkihfill Insufficient to identify TrenchH

R027H.0057T46 m. E, 3.34 m. N, 2.70 m. below 3 x willow (Salix sp.) or poplar (Populus BM: occupation layerN room, sample CI4- sp.) dated H. 049H.005, 3.87 m. E, 3.73 m. N, 2.23 m. below 4 x beech (Fagus sp.) (or possibly plane, BM: occupation layerN room Platanus sp.) R096H.005, L47m7E7328^m. N, 2^%ni7below ~6x alder (Alnus spOJast-grown BM: occupation layerN room, sample C14- 2 x beech (Fagus sp.) (or possibly plane, dated Platanus sp.)

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions FURTHER FIELDWORK AT THE GREAT WALL OF GORGAN 133

microscope at magnifications up to x400 and matched Curtis, Dr Tom Higham, Houman Kordmahini and Dr St to prepared reference slides of modern wood. When John Simpson provided invaluable advice and support. possible, the maturity of the wood was assessed. The detailed comments and thought-provoking advice Fragments selected for dating were weighed. by Dr Cameron Petrie and two anonymous referees have The results are presented in Table 3. Beech (Fagus) improved many aspects of this paper, not least its and plane (Platanus), although unrelated, are anatomi structureand clarity.The staffof Edinburgh University's cally very similar. While the charcoal from samples Graphics and Multimedia Resource Centre provided H.049 and H.096 is almost certainly beech, it is not help in editing some of the images. possible to rule out plane from the diagnostic data Most importantly we are very grateful to all available. members of the expedition. In addition to the co-authors of this article, we are particularly grateful toMaryam Hussein-Zadeh, who accomplished the detailed plans Acknowledgements and section drawings included in this article, and Mohaddeseh Mansouri Razi, who drew the small finds. We are very grateful toDr Seyed Taha Hashemi, thevice We are also indebted toMr Abdolhussein Badpa, the director of the Iranian Cultural Heritage and Tourism driver of theGorgan Wall base, and to our other drivers, Organisation (ICHTO) and the head of the Research workmen and all other supporters of the project, space Department of the ICHTO, and toDr Hassan Fazeli, the does not allow to list. director of the Iranian Center ofArchaeological Research (ICAR), for theirpermission to carry out our joint project, for theirkind and essential help and formaking the 2006 Notes fieldwork season possible. We are indebted toDr Seyed 1 Mehdi Mousavi, the vice-director of the Research The OSL dates are not available yet. 2 Department of the ICHTO, for granting permission to OxA-17021, dl3C= 1.30: 933?26 BP, dated by Dr Tom process samples in theUK and toMr Mahmoud Rabi'ie, Higham (RLAHA, University of Oxford), using the the director of the Golestan ICHTO, for his continued INTCAL04 marine calibrationmodel with a deltaR value support and interestin ourwork. Ms Mojgan Seyedin, Mr set at 0?50 years. 3 Karim Alizadeh and Ms Leyla Safa'ie, members of the OxA-17011, dl3C= -27.45 (1.46 m. E, 3.34 m. N of SW International Relations Department of the ICAR, have cornerof Trench H), OxA-17012, dl3C= 24.95 (1.47m. E, offered invaluable help with our visa applications and 3.28 m. N), OxA-17014, dl3C= -11.90 (3.36 m. E, 3.87 m. have considerably facilitated our project.We would also N) and OxA-17015, dl3C= -11.59 (1.86m. E, 3.58 m. N), like to thank Mr Fereydoun Unagh, the director of dated by Dr Tom Higham (RLAHA, University of ICHTO at Gonbad-e Kavus, and members of the local Oxford). 4 ICHTO office for facilitating our research project inmany Glass by Dr Birgitta Hoffmann: This report, on a strati ways. Amin Nazifi's contributions to the project at every graphically importantpiece of glass, is based on thephoto stage, from initial preparations to post-excavation, have graphic record and a drawing (Fig. 32.3) rather than an been essential, and included translating this long report inspectionof thesherd. The piece was found in the southern from English intoFarsi. room of thebarrack block inTrench H at 2.40 m. below the Without the generous support by the AHRC, the benchmark. Itwas thus 1.20 m. above the foundations of the British Institute of Persian Studies, the ICHTO, the Iran internalwall and 1.14m. higher than the lowest find,but Heritage Foundation, the Stein Arnold Exploration Fund still0.33 m. lower thana storagevessel in situ. Itprovides of theBritish Academy, Edinburgh University's School a terminuspost quern fora late (butnot the latest)phase of of History and Classics and the Carnegie Trust for the occupation Universities of Scotland, none of this could have been While superficially similar to the bowls from Tell achieved. The British Institute of Persian Studies also Madraseh pitXI3 orW21 (Nishapur), thepiece has a very kindly provided facilities and essential equipment. clearly inturnedcollar-rim, while the ninth/tenthcentury Sally Ainslie, Professor Graeme Barker, Professor pieces fromNishapur (modernNeyshabur) have out-turned Douglas Cairns, Professor Robin Coningham, Professor rims (Kroger 1995: 44-45). Similarly,the in-turnedrims of James Crow, Professor Sir Barry Cunliffe, Dr Vesta the sixth centurydo not agree with the suggestedvessel

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions 134 JOURNAL OF PERSIAN STUDIES

shape,which is flatand ovoid, but probablyfrom a bowl or One pathological bone was recorded: a cow caudal vertebra a deep dish. In-turnedrims like the one shown on the from late Sasanian/early medieval context, at the interface photographs appear to be comparativelyrare on bowls or between B.006 and the wall tumble (B.004), which deep dishes and the closest parallels I can find are from exhibited two cavities in the cranial epiphyseal plate with Qa'alat Semaan in Syria (Dussart 2003: 172 fig. 3.1-lc), associated pitting.These lesions are probably the resultof which are dated to theAbbasid periodwith furtherparallels intervertebraldisk degeneration and the subsequent citedfrom theIsland of Bijan. Thiswould thussuggest a date herniationand penetrationof thenucleus pulposa (a jelly in theninth-twelfth, but most likelyninth-tenth centuries. likemass in the centreof the intervertebraldisk) into the 5 In lightof the small size of theassemblage an attemptwas epiphysealplate. Such changes typicallyoccur as partof the made to identifyall fragments.The zoning systemof Dobney normal process of ageing, and thus provide indirect andReilly (1988) was adopted,and all butcherymarks were evidence for thekeeping of old animals. 8 recorded following Lauwerier (1988). Measurements were TABLE 6. Post-cranial bones (all measurements are in takenusing themethods outlinedby von denDriesch (1976). tenthsof millimetres) The toothwear stages of Grant (1982) were used for the Specimen I GL I Bd I Dd recordingof cattle and Payne (1973; 1987) for sheep/goat. Sheep astragalus(Sasanian) 292 197 160 The preservationof thebone was recordedusing a four-point Equid metapodial (Sasanian) 397 299 scale (followingHarland et al 2003): 1.Majority of surfacefresh or even slightlyglossy; very localised or powdery patches; TABLE 7.Mandibular sheep teeth(all measurements are in 2. Bone surface lacks fresh appearance but solid; very tenthsof millimetres) localised flakyor powdery patches; M1W M2W M1/2W 3. Surface solid in places, but flakyor powdery on up to | | | 76 49% of specimen; Sasanian 4. Surface flakyor powdery over 50% of specimen. (RT) Sasanian 82 6 TABLE 4. Preservationof post-cranialelements (after Sasanian 73 Harland et al 2003) ?Sasanian 72 Score I MSP 1 0 TABLE 8. Toothwear stages for sheep 2 2 (afterPayne 1973; 1987) 3 7 P4 Ml M2 Ml/2 M3 [ | | | [ 4 5 Sasanian 8A 9A 4A Sasanian 15A Such taphonomic conditions account for the abundance of Sasanian 9A 2A teethwithin the assemblage (48%), since these have a Sasanian 9A 9G than bones, and it is to be higher density post-cranial Sasanian 9A thatanatomical zones of lower density and the expected ?Sasanian 9A unfused bones from young animals are under-represented in ?Sasanian 11A thisassemblage due topost-depositional degradation. (RT) 7 TABLE 5. Representationof skeletalelements (RT)

Cow Horse Sheep/goat Bibliography Cranium (skull,mandible and 17 teeth) Adamec, L.W. (ed.) 1981. Historical Gazeteer of Iran 2: Vertebrae 1 Meshed and Northeastern Iran. Graz. Forelimbs 1 (scapula, humerus, Adle, C. 1992. "Investigations archeologiques dans le Gorgan, radius) au pays Turcoman et aux confins Irano-Afghans (notes)", Pelvis 1 and In J.-L. Bacque-Grammont and R. Dor, with F. Hitzel Hindlimbs (Femur,tibia) A. Tibet (eds),Melanges offertsa Louis Bazin. Varia Lower limb 2 2 1 (Metapodials,carpals, Turcica XIX, Paris: 177-205 (referencekindly provided tarsals, phalanges) byDr Ali Mousavi).

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions FURTHER FIELDWORK AT THE GREAT WALL OF GORGAN 135

Alizadeh, K. and Ur, JA. 2007. "Formation and destruction of Groot, J.C., Jones, J.E. and Parker, S.T. 2006. "The Barracks at pastoral and irrigatedlandscapes on theMughan Steppe, el-Lejjun (AreasK, L, R, and B.6)'\ In S.T. Parker,The on north-western Iran", Antiquity 81, no. 311: 148-60. Roman Frontier in Central Jordan. Final Report the Allison, P.M., Fairbaim, A.S., Ellis, S.J.R. and Blackall, C.W. Limes Arabicus Project, 1980-1989, 1.Washington:

2005. "Extracting the social relevance of artefact distribu 161-85, figs 5.1-20, pis 5.1-14.

tion in Roman military forts", Internet Archaeology 17 Gyselen, R. 2001. The Four Generals of the Sasanian Empire:

fhttp://intarch. ac.uk/joumaFissue 17/allisontoc.htmO. Some Sigillographic Evidence. Rome.

Ammianus Marcellinus. Harland, J.E, Barrett, J.H., Carrott, J., Dobney, K. and Jaques,

Bocherens, H., Mashkour, M. and Billiou, D. 2000. "Isotopic D. 2003. "TheYork System:An integratedzooarchaeolog studyof Qazvin Plain (Iran) fauna from theNeolithic to ical database for research and teaching", Internet the Iron Age", Journal of Environmental Archaeology 5: Archaeology 13 1-19. rhttp://intarch.ac.uk/journal/issue 13/harland_index.htmD. Boeft, J. den, Drijvers, J.W., Hengst, D. den and Teitler, H.C. Hodgson, N. 2002. "'Where did theyput thehorses?' revisited: 2005. Philological and Historical Commentary on the recentdiscovery of cavalry barracks in theRoman Ammianus Marcellinus XXV. Leiden and Boston fortsat Wallsend and South Shields onHadrian's Wall". In

(referencekindly provided byDr Gavin Kelly). P. Freeman, J. Bennett, Z.T. Fiema and B. Hoffmann Boucharlat, R. and Lecomte, O. 1987. Fouilles de Tureng Tepe (eds), Limes XVIII. Proceedings of the XVIIIth sous la direction de Jean Des hay es 1. Les periodes InternationalCongress ofRoman Frontier Studies held in Sassanides et Islamiques. Paris. Amman,Jordan (September2000) II. BAR International Breeze, D.J. and Dobson, B. 2000. Hadrians Wall. 4th edn. Series 1084 (II). Oxford: 887-94.

London. Huff, D. 1981. "Zur Datierung des Alexanderwalls", Iranica Crow, J. 2004. Housesteads: A Fort and Garrison on Hadrian s Antiqua 16: 125-39. ? Wall. 2nd edn. Stroud. 1987. "Architecture iii. Sasanian", EIr 2. London and

Debrunner Hall, M. 1994. "Eine reine Mannerwelt? Frauen und New York: 329-34. das romische Heer". In M.H. Dettenhofer (ed.), Reine Huyse, P. 1999.Die dreisprachige InschriftSabuhrs I. an der Mannersache? Frauen inMdnnerdomdnen der antiken Ka 'ba-i Zardust (SKZ). Corpus Inscriptionum Iranicarum Welt. Cologne: 207-28. III.2. London. Dobney, K. and Reilly, K. 1988. "A method for recording James,S. 2001. "Soldiers and civilians: identityand interaction archaeological animal bones: the use of diagnostic zones", in Roman Britain". In S. James and M. Millett (eds), an Circaea 5(2): 79-96. Britons and Romans: advancing archaeological C. van Driel-Murray, 1997. "Women in forts?", Jahresbericht agenda. CBA Research Report 125. York: 77-89. der GesellschaftPro Vindonissa 1997: 55-61. Johnson,A. 1983. Roman Forts of the1st and 2nd centuries Driesch, A. von den 1976. A Guide to theMeasurement of A.D. in Britain and the German Provinces. London. Animal Bones from Archaeological Sites. Peabody Johnson, S. 1983. Late Roman Fortifications. London. Museum Bulletin 1. Mass. Cambridge Kennedy, D. and Riley, D. 1990. Rome's Desert Frontier from Dussart, O. 2003. Les verres islamiques de Qal'at Sem'an. the Air. London. Annates de 15e de MY. congres TAIHV. New York: 171-79. Kiani, 1982. Parthian Sites in Hyrcania. The Gurgan Gale, R. and Cutler, D. 2000. Plants inArchaeology. Otley and Plain. Archaologische Mitteilungen aus Iran London. Erganzungsband 9. Berlin. A. 1982. "The use of tooth wear as a to the of Grant, guide age Kroger, J. 1995. Nishapur: Glass from theEarly Islamic Period. domestic ungulates". In B. Wilson, C. Grigson and S. New York. Payne (eds), Ageing and Sexing Animal Bones from Lauwerier, R.C.G.M. 1988. Animals in Roman times in the Archaeological Sites. BAR British Series 109. Oxford: Dutch Eastern River Area. ROB. Amersfoort.

91-108. Naumann, R. 1977. Die Ruinen von Tacht-e Suleiman und G. 1998. Rome and Persia at 502-532. Leeds. Greatrex, War, Zendan-e Suleiman und Umgebung. Fiihrer zu an Gregory,S. 1996. "Was there easternorigin for thedesign of Archaologischen Platzen in Iran2. Berlin. lateRoman fortifications?Some problems forresearch on Nokandeh, J., Sauer, E., Omrani Rekavandi, H., Wilkinson, T., forts of Rome's eastern frontier". In D.L. Kennedy (ed.), Abbasi, G.A., Schwenninger, J.-L., Mahmoudi, M., The Roman in the East. Journal of Army Roman Parker, D., Fattahi, M., Usher-Wilson, L.S., Ershadi, M.,

Archaeology Suppl. 18.Ann Arbor: 169-209. Ratcliffe, J. and Gale, R. 2006. "Linear Barriers of

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions 136 JOURNAL OF PERSIAN STUDIES

Northern Iran:The GreatWall ofGorgan and theWall of Mesopotamia". In K. Bartl and S.R. Hauser (eds), Tammishe", Iran 44: 121-73. Continuityand Change inNorthern Mesopotamia from the Nunnerich-Asmus, A. 2006. "In guten wie in schlechten Zeiten: Hellenistic to theEarly Islamic Period. Berlin: 87-126. ? Frauen und das romische Militar?eine schwierige 2004. "Glass and small finds from Sasanian contexts at

Beziehung?",vtofe Welt 313: 31-34. the ancient city-site of Merv". In VP. Nikonorov (ed.), Oates, D. 2005. Studies in theAncient History ofNorthern Iraq. Central Asia from the Achaemenids to the Timurids: 2nd edn. London. Archaeology,History, Ethnology, Culture. Materials of an ? and Oates, J. 1990. "Aspects of Hellenistic and Roman International Scientific Conference dedicated to the Settlement in the Khabur Basin". In P. Matthiae, M. Van Centenary of Aleksandr Markovich Belenitsky, St. Loon and H. Weiss (eds), Resurrecting thePast. A Joint Petersburg, November 2-5, 2004. St. Petersburg: 232-38.

Tribute to Adnan Bounni. Istanbul and Leiden: 227^48. Sommer, C.S. 1984. The Military Vici in Roman Britain. BAR

Overlaet, B. 1993. "Organisation militaire et armement", in British Series 129.Oxford.

Splendeur des Sassanides: L 'empire perse entre Rome et Southern, P. and Dixon, K.R. 1996. The Late Roman Army. la Chine [224-642]. Bmssels: 89-94. London. Parker, S.T. 2000. "Roman Legionary Fortresses in the East". In Speidel, M.A. 1997. "Frauen und Kinder beim romischen

R.J. Brewer (ed.), Roman Fortresses and Their Legions. Heer", Jahresbericht der Gesellschaft Pro Vindonissa London and Cardiff: 121-38. 1997: 53-54. Payne, S. 1973. "Kill-off patterns in sheep and goats: the Stronach,D. 2004. "On theAntiquity of theYurt: Evidence mandibles fromAsvan Kale", AS 23: 281-303. fromArjan and Elsewhere", The Silk Road Newsletter 2.1 ? 1987. "Reference codes for wear stages in themandibular (http://www. silkroadfoundation.org/newsletter/2004vol2 cheek teethof sheep and goats",Journal ofArchaeological numl/yurt.htm). Science 14: 609-14. Talbert,R.J.A. (ed.) 2000. BarringtonAtlas of theGreek and Petrikovits, H. von 1975. Die Innenbauten romischer Roman World. Princeton and Oxford. R.S.0.2000. "The in the Late In R.J. Legionslager wdhrend der Prinzipatszeit. Opladen. Tomlin, Legions Empire". Poidebard,A. 1934.La Trace de Rome dans le desert de Syrie. Brewer (ed.), Roman Fortresses and Their Legions.

Le Limes de Trajan a la conquete Arabe. Recherches London and Cardiff: 159-81. at Interim aeriennes (1925-1932). Paris. Whitehouse, D. 1972. "Excavations Siraf: Fifth Poulter,A. 2006. "Dichin: Twilight of theGoths", Current Report", Iran 10: 63-87. sur WorldArchaeology 17: 34-41. Widengren, G. 1957. "Recherches le feodalisme iranien", Precht,G. 1975. "Die Ausgrabungen imBereich des Castellum Orientalia Suecana 5, for 1956: 79-182. ? Divitia; Vorbericht iiber die Kastellgrabungen",Kolner 1976. "Iran, der groBe Gegner Roms: Konigsgewalt, Jahrbuch 13, for 1972/1973: 120-28. Feudalismus, Militarwesen", Aufstieg und Niedergang ? New York: 219-306. 1987. "Koln-Deutz". In H.G. Horn (ed.), Die Romer in der romischen Welt 11,9,1. Berlin and T.J. 1993. "Linear hollows in the Nordrhein-Westfalen. Stuttgart: 513-16. Wilkinson, Jazira, Upper no. Puschnigg,G. 2006. Ceramics of theMerv Oasis. Recycling the Mesopotamia", Antiquity 67, 264: 548-62. C.E. 1900. Khurasan and Sistan. and London. City. Walnut Creek. Yate, Edinburgh of the 2007. Plain Animal Bone In H. Rychagov, G.I. 1997. "Holocene oscillations Caspian Young, R., "Qazvin Report". the Plain Sea, and forecasts based on Palaeogeographic Recon Fazel Nashli (ed.), The Archaeology of Qazvin sixth to the millenniumBC Tehran: 275-84. structions", Quaternary International 41/42: 167-72. from the first B. 2000. "The initialdomestication of Shahbazi, A.S. 1987. "Army i. Pre-Islamic Iran", EIr 2. London Zeder,M.A. and Hesse, andNew York: 489-99. goats (Copra hircus) in the Zagros mountains 10,000 287: 2254-57. Simpson, S.J. 1996. "FromTekrit to theJaghjagh: Sasanian sites, years ago", Science settlementpatterns and material culture in northern Zosimos.

This content downloaded from 129.215.19.197 on Mon, 16 Dec 2013 07:29:47 AM All use subject to JSTOR Terms and Conditions