Chapter 1 Introduction

In: Greek Inscriptions on the East Bank
Maria Nilsson
Search for other papers by Maria Nilsson in
Current site
Google Scholar
Adrienn Almásy-Martin
Search for other papers by Adrienn Almásy-Martin in
Current site
Google Scholar
, and
John Ward
Search for other papers by John Ward in
Current site
Google Scholar
Open Access

1 Gebel el-Silsila—the Site

The archaeological site of Gebel el-Silsila is situated in Upper Egypt, between the temple areas of Kom Ombo and Edfu, some 65 km north of the modern city of Aswan and 130 km south of Luxor (Fig. 1). There, it stretches out on both sides of the Nile, where the river reaches its narrowest point.1 Including its northern neighbouring sites of Nag el-Hammam and Shatt el-Rigal, the current concession encompasses an area of 30 square kilometres.2 The west bank borders the agricultural plain of Fatira in the south, and extends through the quarries of Gebel el-Silsila West and the village of Nag el-Hammam to a small wadi just north of the more renowned Wadi Shatt el-Rigal. The east bank, to which the material presented herein is limited, borders agricultural land to the north and south, the Nubian village of Kalabsha to the east, and the Nile to its west. The massif and quarryscape that form Gebel el-Silsila—the Arabic ‘Mountain of the Chain’—was known to the ancient Egyptians as “Khenu/Kheny”,3 and was ancient Egypt’s principal source of Nubian sandstone, a fine- to medium-grained beige-grey stone that was exploited at least from the Middle Kingdom.4 The extracted stone was destined for sanctuaries throughout Upper Egypt, and was used in the vast majority of temples between Dendera in the north and Elephantine in the south.


Figure 1

Map of Egypt with inserted overview of Gebel el-Silsila

Courtesy of Google Earth

2 Brief Historical Outline

Archaeologically, human activity on the site is documented from the Late Palaeolithic5 (based on newly discovered lithic material) and epigraphically since the ‘Epipalaeolithic’ period, with continuous attestations throughout the subsequent periods.6 The site gained in importance during the 18th Dynasty when quarrying expeditions were sent out by some of the more renowned New Kingdom rulers, including Hatshepsut, Thutmosis III, Amenhotep III and IV/Akhenaten, Seti I, Ramses II and III.7 In addition to the formal exploitation of the sandstone, the Pharaonic presence at Gebel el-Silsila is preserved in the remains of several architectural monuments—shrines, stelae, the Temple of Sobek, and a speos that was dedicated to the Nile gods by Pharaoh Horemheb, but already constructed during the early Thutmosid period.8 Led by the high priest of Memphis, religious festivals were celebrated biannually, and high officials wished for their souls to return to the site in the Afterlife.9 However, after the demise of the New Kingdom, ancient Kheny falls into almost complete oblivion; these dark ages would last until the end of the Ptolemaic period, perhaps even up to the early Roman period under Emperor Augustus.10 Marking the transition was the destruction of the Temple of Sobek, along with the eradication of all crocodile images, but also the demise of the town, the closure of its cemeteries, and the discontinuation of both primary (official) and secondary (private) epigraphy. From this obscure period, only a couple of official monuments can affirm any activity at Gebel el-Silsila: the royal stela of Shoshenk I (22nd Dynasty) on the west bank, and the cartouches of Apries (26th Dynasty) on the east.11 While the stela of Shoshenk speaks about an official quarrying expedition, there is no evidence from any quarry that can testify to this. This, however, does not mean that such expeditions were absent, but rather that the Romans reused their quarries.

Considering the amount of restoration work and new temple structures that were built in sandstone during the Ptolemaic period, it is surprising—if not bizarre—to find nothing other than a few beer jugs and two stelae of potential contemporaneity. Naturally, the Romans may have usurped the Ptolemaic quarries, and the material could be buried beneath Roman archaeological material, but Gebel el-Silsila as it stands presents very limited evidence of this period. Beer jars or jugs have been documented amongst the ceramic finds at the (then destroyed) Temple of Sobek, and in the Main Quarry (Q34) of the east bank, as well as by ‘Pottery Hill’ and ‘Black Rock Camp’ on the west bank.

It is from the Roman period that the modern designation of the site may have its roots: it is presumed that the place-name ‘Sil-sil’ or ‘Silsili’, which was mentioned in a Latin document from to c. 400 AD,12 refers to the site, and that it was a Roman word deriving from the late Egyptian ‘Khol-khol’, meaning of ‘barrier’ or ‘frontier’.13 No Greek or Coptic name is known, but from the Roman Silsili sprung the modern name ‘Gebel el-Silsila’: the ‘Mountain of the Chain’, believed to have derived from a local tale that describes how a chain was once tied between the two banks in order to stop passing ships for taxation reasons.14 Roman activity is noticeable almost everywhere in the form of the partial takeover of older quarries, and the abundance of pottery and graffiti, but also in structural form with several clusters of shelters, lookout stations, storage facilities, domestic, administrative and religious buildings (Fig. 2), around which hundreds of ostraca have been discovered. These ostraca generally contain lists of names or equipment, and will be treated in a separate publication. Since 2015 the Swedish mission has excavated a series of Roman areas on the east bank, including the ‘Stables of Tiberius’, a naos and blacksmith’s area, quarry shelters in the Main Quarry; and surveyed the ‘peak station’ and a housing complex situated above the New Kingdom cemetery; as well as comprehensively documented all preserved epigraphy. On the West, ‘Pottery Hill’ and ‘Black Rock Camp’ as well as several stations have been surveyed and partially excavated, and all epigraphy documented.


Figure 2

Google Earth map of the east bank marked with Roman sites

Courtesy of Google Earth

Among the more intriguing structures is the Stables of Tiberius, which was an administration building and stables used periodically during an eight-year period when stone was extracted from its quarry (Q24—see Chapter 4). Fifteen rooms divided over four levels have been excavated so far, revealing well over 30,000 ceramic sherds, 150 demotic ostraca, coins, seals, stamps, textile, jewellery, etc. Another intriguing Roman complex is the so called ‘naos area’ in quarry 37 (Chapter 6), also on the east bank, and currently under excavation. Based on the epigraphy, it was in use during the reigns of Augustus, Tiberius and Claudius (with no mentioning of Caligula), and was then abandoned, as were all other quarries of the east bank. It was not exploited again until the early 1900’s when parts of it were dynamited to produce stone for the Esna barrage.15 During the early surveying in 2012 and 2013, the team found several oil lamps, a few coins, and textiles amongst the surface material scattered in the area. Soon thereafter, the Mission was given the permit to clean the so-called naos, which is a shrine-like structure situated on top of the quarry. The undecorated structure had been reused by a local Sufi-group and by Esna quarrymen (1906–1909), but remained in a relatively good state of preservation. Finds included architectural remains of columns and an extension to the shrine at its front, but it was not until 2019 that the Mission began a more detailed investigation of the area. Amongst the more intriguing remains was a staircase and indications of castellation, which will be excavated more fully in coming seasons. Coins, tools, organic material, pottery and ostraca will together allow a reconstruction of this area.

Of Gebel el-Silsila’s 104 quarries (excluding Nag el-Hammam and Shatt el-Rigal), at least 36 can be confirmed as partially or fully exploited by the Romans (19 on the east, 17 on the west) (see Fig. 2). So far, the epigraphic survey has enabled the team to identify the source (the quarry) and destination (the specific temple for which the stone was intended) in several cases, including the temples of Edfu, Dendera and Esna, the gate of Tiberius at Medamoud, the Temple of Isis and Min at Koptos, etc.16 Often, the Romans reused New Kingdom galleries (from Amenhotep IIIIV) focusing their work on the exterior, open surfaces rather than the subterranean rooms. In some areas, the ceiling was intentionally collapsed and extracted together with the pillars that once lifted it. Two of the large Ramesside quarries in the central part of the east bank were equally reused in parts, as indicated by the combination of Roman and 19th Dynasty pottery as well as the quarrying techniques. On the west bank, the 18th Dynasty Main Quarry (Q11) received a limited extraction during the Roman period, and some of the quarry faces are preserved with hundreds of pictorial and textual engravings. Similarly, there are traces of Roman usurpation of the quarries of Tutankhamun.

3 Research History

Despite a series of early publications with brief references to graffiti and rock art at Gebel el-Silsila, only a limited selection of the epigraphy had been published prior to the arrival of the Swedish Mission. As a rule none of the early scholars stayed long enough to offer more than a cursory and limited vision of Gebel el-Silsila’s ancient richness, and the earliest records, including the Description de l’Égypte,17 viewed the site from a technical perspective, more concerned with the geological and natural features than ancient epigraphy. Other visitors and scholars, such as Carl Richard Lepsius, laid their focus on pharaonic monuments and epigraphy, with merely a brief synopsis of the Graeco-Roman graffiti.18

The first publication that processed graffiti was F.C. Gau, Antiquitiés de la Nubie in 1822, which included 16 entries.19 Some examples were incorporated in Jean Antoine Letronne’s Recueil des inscriptions grecques et latines d’Égypte, which also included brief commentaries.20 Fifteen entries were published in the Corpus Inscriptionum Graecarum III in 1853, in which facsimiles produced by Gau were used.21 A limited number of texts were included in G. Deville’s, Inscriptions grecques d’Égypte recueillies en 1861 (…).22 Later, Sir Flinders Petrie provided a few more examples of textual and pictorial graffiti in ‘A season in Egypt 1887’, but for Gebel el-Silsila it was more of an account of quarry marks than its textual graffiti, and there was no further analytical approach to their significance.23 Other examples were given by Petrie’s travel and research companion, F. Ll. Griffith, who in his Notes on a tour in Upper Egypt included references to graffiti at Gebel el-Silsila.24 In the following years, Archibald H. Sayce published graffiti from the area, although focused on the northern west bank and Shatt el-Rigal; Gustave Lefebvre, Recueil des inscriptions grecques-chrétiennes d’Ègypte included some graffiti; and in Wilhelm Dittenberger, OGIS, as well as René Cagnat and Pierre Jouguet, IGRR, there were also reproductions of already published texts.25

The first substantial corpus was accumulated by George Legrain while in charge of the monuments of Upper Egypt. He made an inventory of monuments and preserved epigraphy at Gebel el-Silsila, and a portion of the then visible graffiti was documented in 1894 and 1896.26 He spent the necessary time on site and numbered a large proportion of the existing inscriptions and pictorial graffiti. However, he never completed the task of publishing his results in detail. Instead, it took several years before he finally handed over his notebooks and the responsibility to distribute the material in a published form to Wilhelm Spiegelberg in 1911. Spiegelberg, who translated the demotic texts, was joined by Friedrich Preisigke, who transcribed the Greek inscriptions, and their joint effort led to a corpus of 306 inscriptions published as Ägyptische und griechische Inschriften und Graffiti aus den Steinbruchen des Gebel Silsile (Oberägypten) in 1915. This publication has remained the standard reference work for Graeco-Roman graffiti at the site and has been reproduced in various formats.27 R.A. Caminos and his students produced 1:1 copies of accessible texts during the EES epigraphic survey (1955–1984), but despite their efforts a comprehensive corpus remains unpublished. Thus, the existing documentation, while still precious, is already outdated, the bibliography remains very limited for such a large and important site, and its archaeological and historical value is largely ignored or undervalued.

4 Main Objectives

The main aim of the Swedish Mission was to produce for the first time a comprehensive corpus of all the rupestrian material, including a re-documentation of epigraphical texts already published, in order to provide an analysis of the chronological, ethnographical, cultural, socio-political, economic, religious and professional aspects of life at Gebel el-Silsila during the Graeco-Roman period. All Greek names28 are placed in a larger database, listing all individuals attested at Gebel el-Silsila, Nag el-Hammam and Shatt el-Rigal, throughout all represented ancient periods. This will include texts in hieroglyphic, hieratic, demotic, Greek, Latin, and Coptic scripts. The database will be published online to serve as a corpus reference system, allowing comparison with sites elsewhere. The objective is to learn about the people of ancient Gebel el-Silsila.

5 Methodology

As part of the initial topographic documentation, the site was divided into 104 individual quarries, 52 on each side of the Nile, running from north to south. Following this classification, each quarry was subdivided into partitions (in case of larger quarries), followed by individual quarry faces (the vertical cliff walls resulting from extraction), and epigraphic material (i.e. inscriptions and quarry marks) recorded from the top down. Thus, when creating the comprehensive corpus of the marks, they are divided in accordance with the following successive classification: Quarry (abbreviated ‘Q’), Partition (using alphabetic letters), Quarry Face (using Roman numerals) and Inscriptions. The epigraphic material is catalogued as ‘Inscr.’ (Inscription: textual graffito) and ‘Pict.’ (Pictograph: pictorial/quarry mark), and followed by recorded inventory number. For example, the southern section of Quarry 34, Partition F, consists of 17 separate quarry faces (Q34.F1–17); its second quarry face (F2) displays 110 textual graffiti labelled as Q34.F2.In.1–110, and 100 quarry marks classified as Q34.F2.P1–100. This classification system enables the addition of marks and inscriptions found after the initial survey without any disorganisation of the original records and without interfering with a consecutive numerical system for the entire quarry. It is accordingly more suitable and flexible than Legrain’s system that cluster-labels epigraphic features when one or several examples were noted after his initial numeral system was completed.29 Many of Legrain’s original chalk numbers are still visible on the quarry faces and show this numerical system, to which any graffiti found later were appended to the nearest recorded number with the abbreviation ‘ADD’, and which created a rather messy structure. Throughout the book, however, references will be made to these unpublished reference numbers, as they are part of the current context.

The material has been studied and processed through the production of 1:1 analogue acetate copies in combination with digital layer-drawing techniques (Adobe Photoshop and Illustrator). Texts situated on quarry faces not accessible from the ground level have been photographed with various digital cameras and lenses, including aerial (drone), macros (60 and 100 mm) and micros (150–300 and 400–500 mm) in various lights, seasons and times of the day, and angled from below, full front, and above. For texts situated at higher levels, approximate measures will be listed based on their context, primarily block size/superimposed tool marks and spatial relation with surrounding epigraphic documents.30 After photographing, each image was digitally enhanced, and manipulated in DStretch© algorithms, as well as inverted to a negative in order to clarify all outlines and separate the text from the irregular surface marked by tool grooves from the block extraction. Photogrammetry and laser scanning were applied in areas where traditional photography was prevented due to poor lighting and/or difficult locations, and yielded clearly defined reproductions.

Each catalogue entry includes information about field number (inv. no.), dimensions (L.—length; W.—width), approximate height above the current ground level, state of preservation (conditions: ‘well preserved’—intact and legible; ‘poorly preserved’—fragmentary, but still legible; ‘illegible’—too poorly preserved to interpret), bibliography and, if available, date. Also displayed is a facsimile, transcription into majuscules, articulated text, translation and short commentary.

6 Sigla

Due to their position within an open landscape and exposure to the physical elements, the texts of Gebel el-Silsila are naturally eroded and some are also damaged by breakage. In addition to texts fragmented by external force, several were left unfinished (intentionally or by accident), or composed erroneously by the producer. For the editorial reconstruction of the inscriptions the Leiden conventions (Leiden +) has been applied,31 using the following sigla:

Square brackets with dashes [---] are used in cases where the text is too damaged to estimate the original proportions and to indicate lacuna of uncertain length (e.g. no. 51: [---]μιος). Square brackets with individual points [.] are used to indicate illegible individual letters (e.g. no. 10: Α[.]ι[..]η). Superscript v is used for spaces left empty by the producer either due to imperfect surfaces or to separate names. Each listed vacat (v) represents one empty space (e.g. no. 120: ΨενσουτένσιοςvvvΟσοτυχιοῦς). Subscript dots (α̣ β̣) are used for doubtful readings of partially preserved letters (e.g. no. 49: Πρεξῖνο̣ς̣ for ΠΡΕΞΙΝϹΙ based on a poorly written epsilon used erroneously for an alpha). Angular brackets ⟨ ⟩ are used to indicate letters erroneously omitted by the producer (e.g. no. 6: Ἀσκλη⟨…⟩). Parentheses ( ) are used to indicate abbreviations (e.g. no. 171: Ἁρ(βέσχινις)). Parentheses with points (…) are used to indicate an unfinished name or word (e.g. no. 69: τὸ πρ(οσκύνημα) Ἀννοῦφει). Brace brackets { } are used to indicate editorial suppressions of erroneous repetitions or inserted letters (e.g. no. 68: προσκύ{ει}νημα⟩). Omission brackets (raised diagonals) ⸌ ⸍ are used for texts in linear suspension, to indicate the correct position of the superscripted letter(s) (e.g. no. 169: ⸌πιφάνες)




text omitted by the scribe




text written by the scribe in error

⸌ ⸍

text inserted or added above a line


vacat of one character


vacat of characters (plural)


imperfectly preserved characters


Based on a combination of satellite imagery and documentation, the current minimum width is approximately 320 m, which can be compared with the 395 m measured in 1901 and 1964; see Petermanns (1901: 9); Butzer & Hansen (1968: 17).


See Nilsson & Martinez (2017: 445).


E.g. Kucharek 2012, 1. Other variants include R-H̱n—‘the mouth of Khen’ (P.Brooklyn 351446, rto 21b); pꜣ mw wꜥb—‘the pure water’, see Caminos & James (1963: 9 with n. 1, 34 with nn. 2–3)).


E.g. Harrell (2016: 23); Klemm & Klemm (2008: 180); Nilsson et al. (2019).


Cf. Smith (1967).


The authors apply the term ‘epipalaeolithic’ in accordance with Huyge (2005). For epipalaeolithic rock art see Nilsson & Ward (2016: 172–173; 2020a: 236–237, 239–241), and for a general overview of the temporal distribution of epigraphy, see Nilsson (2018a; b); Nilsson & Ward (2016; 2019a; b; 2020a; c); Osing (2006). For reference to Kheny as an Old Kingdom funerary domain, see Jacquet-Gordon (1962: 59, 432, no. 6). For Middle Kingdom references to Kheny, see P. Kahun and Gurob, Text 69, letter 7, with pl. 28, iii, 2 col. 2; P. Ram. D, 187, publ. AEO, pls. 2 and 2a, lower; P. Brooklyn 351446, rto 64b.


E.g. Brand (2000: 176, 264, 359,362); Caminos (1987); Martinez (2009); Nilsson & Martinez (2017); Nilsson & Ward (2019c, d, 2020a, b, d, e).


Nilsson & Martinez (2017).


For the annual festivities and celebrations of Hapi, see LD III (pl. 200, d); Kitchen paresis (1975: 84) 15; Janssen (1987: 136).


For activity during the early Roman period see Nilsson et al. (2019); Nilsson & Almásy (2015).


Nilsson & Ward (2021b: 38).


Notatia Dignitatum Orientis xxviii, § 1,18.


Forbiger (1844: 802 n. 22b); Champollion (1814: 171); Weigall (1910: 360). For a discussion on its validity, see Caminos (1977: 444 n. 5).


E.g. Pococke (1743: 114); Norden (1798: pls 120–124); Perry (1743: 356, 360). On the Arabic name: J.M. Cowan (ed.) Arabic-English Dictionary. The Hans Wehr Dictionary of Modern Written Arabic. Wiesbaden, 1994, p. 492a: ‮سلسل‬‎


E.g. Maspero (1912: 198, 201, 224, 256, 286–287); Weigall (1910: 360). See also Caminos (1977: 444 n. 3) for reference to more modern quarrying on the west bank.


Nilsson (2015b); Nilsson & Almásy (2015); Nilsson et al. (2019).


La Description de l’Égypte: Antiquités I (1809: pl. 47).


LD IV: 98; VI: 24.


Gau (1822: Inscriptions, pl. X, nos 2–7, 9–18).


Letronne (1842: 430–433; 1848, 231–234).


CIG III 4843–4858, add. 1218.


Deville (1865: 457–492).


Petrie (1888: 15–16).


Griffith (1889: 229, 232–234).


Sayce (1891: 49–52, nos 1–39; 1908, 28–29); Lefebvre (1907: 102, xxii. 560); OGIS II no. 676; IGRR I nos 1276–1280.


Graff. Silsile 3.


E.g. SB III 6843–6919; SB V 8386–8387, 8651, 8829; Geraci (1971: 107–112, facs IIV); I. Thèbes à Syène 78–165, pls 42–63.


The Egyptian names written in Greek without Greek endings are not accentuated; otherwise we follow Willy Clarysse’s rules on Greek accentuation of Egyptian names (cf. Clarysse 1997).


Spiegelberg reorganised Legrain’s original number system for the 1915 publication (Graff. Silsile).


Block sizes are generally homogeneous at each quarry face and consist of a diagonal, parallel pattern of a series of two longer (19–29 cm) segmented grooves and a final intersection (5–12 cm), creating a block size of 52–58 cm high for the early Roman quarries.


  • Collapse
  • Expand