The Heroic Age
Abstract: The evidence for occupation at the Roman fort site of Cramond between the fourth and tenth centuries A.D. is assessed using a variety of sources of evidence including artefacts, place-names, documents and the location of later structures. It is argued that this evidence suggests both British and Anglo-Saxon occupation. Although its exact nature is unclear, a religious element is likely.
This article was amended by the author on February 20, 2001.
The nature of the Anglo-Saxon impact in southern Scotland has traditionally been dominated by documentary sources and to a lesser extent place-name studies. More recently archaeology has begun to influence how we understand this phenomenon. Whilst much of the impetus for this has come from recent excavations such as Whithorn (Hill 1997), Dunbar (Holdsworth 1993), and the Mote of Mark (Laing 1973, 1975; Longley 1982), a considerable amount of information can be gained from older excavations. Although the evidence from older excavations is problematical it can be of significance when critically reappraised. Until relatively recently Roman fort sites were generally excavated in a manner unlikely to aid the understanding of post-Roman phases of occupation. We are therefore forced to rely upon stray finds of artefacts and other sources such as place-names, documentary sources, and the location of later Medieval structures to attempt to understand any post-Roman occupation of such sites.
The Roman fort at Cramond, Lothian [NT 189768], on the south side of the Firth of Forth next to the river Almond, was an important military base during the Antonine occupation of Scotland (c. A.D. 140-190) and the early third century Severan campaigns. The fort was probably garrisoned during the Severan occupation from A.D. 208 to 210, and is likely to have been evacuated in A.D. 211 or 212. Its importance did not end at this time, and there is a considerable body of evidence for post-Severan activity at Cramond that probably continued until the end of the first millennium A.D. The exact nature of this occupation is unclear but can be partially reconstructed.
Occupation of the fort during the later third and fourth centuries has long been recognised as a possibility (MacDonald 1918:214, 216). In the praetentura area of the fort a stone building, designated B4, was constructed re-using red sandstone ashlar from the Severan principia and sealed a coin of Julia Domna (A.D. 196-211) (Rae and Rae 1974:184-86). The exact date of construction is unclear, but it probably belongs to the first half of the third century. The building was crudely made of stone bonded with clay with a floor consisting of a hard tread of stones, cobbles, and dried mud. It appears to have possessed glass windows. The unskilled nature of the construction and the crudity of its floor were interpreted as suggesting that this structure did not indicate a further period of Roman military occupation, but it was seen as sophisticated enough that it could not be dismissed as squatter occupation. Other post-Severan structural activity has been discovered at the bath-house to the north of the fort, where furnaces were inserted and there was some late third-century pottery (Frere 1977:368-70; Grew 1980:354). This phase at the bathhouse is comparable to the activity at the praetentura, as it also has crudely built stone walls bonded with clay. The bathhouse was eventually deliberately dismantled, suggesting that the site was still occupied after the bathhouse went out of use. It is now thought that the evidence from the bathhouse and praetentura may relate to a short phase of small-scale Roman military occupation immediately postdating the general abandonment of the fort in A.D. 211/212.
Both building B4 and the bathhouse produced late third-century Roman pottery indicating continuing contacts with the Roman province to the south. There is also a general scatter of pottery of this date from other parts of the fort and the nearby vicus (Goodburn 1978:418; Rae and Rae 1974:217-18).Such contacts are also indicated by a number of Roman coins. The site has produced coins of Geta (A.D. 211-12), Caracalla (A.D. 211-17), Tetricus II (A.D. 270-73/4), Probus (A.D. 280-81), Diocletian (A.D. 284-305), Galerius (A.D. 305-11), Constantine I (A.D. 306-37) and Constantine II (A.D. 337-61) (Bateson 1989:167; Robertson 1983: table 2).These coins are all stray finds but do appear to indicate continued activity throughout the third and first half of the fourth centuries.
Stray finds of post-Severan coins from Scotland are a problematical source of evidence (Casey 1984; Robertson 1983:429-30). The evidence from Traprain Law clearly shows that the Votadini continued to have access to Roman coins throughout the third and fourth centuries (Sekulla 1982), so there is no a priori reason to discount the coins from Cramond. Casey suggests that coins from western mints, particularly Trier and Lyons, are more likely to be genuine, but this can not be taken as an absolute criteria (Casey 1984; see also Bateson 1989). Some of the coins from Cramond are of particular interest: the coin of Tetricus came from the spoil heap associated with the bathhouse excavations and was minted at Cologne; as it comes from a western mint it may well be genuine (Robertson 1983:408). Much of the bathhouse was robbed to foundation level in the seventeenth century and subsequently large quantities of soil and associated debris were dumped on the site. Much of this material may have come from some distance away, so that some of the later third- and fourth-century artefacts may not be associated with the bathhouse at all. Even if this is accepted, the material must still have come from somewhere on the Roman fort or the civilian settlement. Coins of Diocletian, Galerius, and Constantine I were all found in close proximity during the digging of a garden at 15 Glebe Road and are described as having an "emphatic" provenance (Robertson 1971:115-16), although only the coin of Constantine, from Arles, is from a western mint. This collection--which comes mainly from eastern mints, does not include any Severan or earlier coins, and was found in association with a later Byzantine coin (see Later Artefacts below)--probably represents a modern loss.
Excavations in the 1970s uncovered some poorly preserved structural remains indicating a short-lived immediately post-Severan phase of activity on the site; they also produced some pottery indicating post-Severan activity (pers. comm., Nicholas Holmes 1998). The excavator interpreted this as indicating that there was some native re-occupation of parts of the site and that Roman patrols probably continued to visit the site throughout the third century, but there was no evidence for activity after the late third century. >From the patchy evidence that has been recovered, it seems that activity continued at Cramond throughout the third century and possibly into the first half of the fourth century. The nature of this occupation is unclear and difficult to interpret, but it indicates that Cramond was not completely abandoned during the third and fourth centuries.
Although no structural remains of the Early Historic period (A.D. 400-1100) have been found at Cramond, a number of artefacts of this period have been discovered. There is a bronze coin of the Byzantine emperor Justinian (A.D. 527-65) (Robertson 1971:115-16), which was discovered in the same location as some post-Severan coins discussed previously. Although most Byzantine coins from the British Isles are dismissed as modern losses (Casey 1984:295), there is evidence for trade between the eastern Mediterranean and western Britain and Ireland c. A.D. 475-550 that penetrated as far north as central Scotland (Alcock and Alcock 1990; Fulford 1989). There is therefore no reason to automatically discount the coin of Justinian from Cramond, although the coins that it was discovered with suggest that it may well be a modern loss.
Another discovery was an eighth- or ninth-century enamelled bronze circular mount with equal-armed cross decoration incorporating millefiori and yellow champleve enamel discovered in the churchyard at Cramond, which overlies the fort (Bourke and Close-Brooks 1989:230-32). This was probably originally part of a larger composite Insular object, probably ecclesiastical. A plain bronze finger ring discovered at Cramond in 1870 whilst digging a grave near the church bears an Anglo-Saxon runic inscription "[.]ewor[.]el[.]u."[5 ]The letters "wor" are best interpreted as part of the Old English word worthe, "made," and the rest of the inscription probably consisted of one or two Anglo-Saxon personal names. Although it has been suggested that this item is ninth century, it is not closely dateable and the only thing that can be said with certainty is that it probably dates to between the second half of the seventh century and the tenth century (Laing 1973:18; Page 1973:36). A blue glass bead found near Cramond House (PSAS 1970: no. 10) is not closely dateable, but such items are typical of Early Historic Celtic sites.
These various stray finds from in and around the fort suggest that it continued to be a focus for activity until at least the eighth or ninth centuries. The enamelled mount in particular suggests a site of some wealth and importance. The lack of structural evidence hinders any interpretation. It is unclear from the archaeological evidence if the site was continuously occupied and whether it was a secular or religious site.
The place name Cramond is not recorded prior to the twelfth and thirteenth centuries, when it is found in a number of forms such as Caramonde/Caramonth (A.D. 1178-79), Karramunt (A.D. 1166-1214), and Karamunde (A.D. 1293) (Nicolaisen 1976:162; Watson 1926:369). The name contains the distinctive Brythonic or P-Celtic element cair/caer, which means "fort," plus the river name Amon, whose exact linguistic roots are unknown (Nicolaisen 1976:160-62, 178). Cramond therefore means "fort on the river Almond," and the name must have originated prior to the conquest of this area by Anglo-Saxon Northumbria in the middle of the seventh century.
The earliest part of Cramond parish church is the fifteenth century tower, although parts of the building incorporate fourteenth century masonry (RCAHMS 1929:27-28, no. 37). It is located over the praetenturna of the Roman fort, as are a number of other parish churches in Scotland including Carpow and Inveresk (Smith 1996:24). This might be dismissed as coincidence, but as the praetenturna was a focus for post-Severan activity and the churchyard produced an Early Historic enamelled mount, this suggests that the administrative centre of the Roman fort retained some form of administrative or religious importance during the Late Roman and Early Historic periods.
There are a number of possible documentary references to Cramond. Bede (EH I.8; Colgrave and Mynors 1991:40-41) refers to a site known as Urbs Giudi in the early eighth century, and the same site is referred to as Iudeu in the tenth century Historia Brittonium attributed to Nennius. Historia Brittonium describes events in the 650s, when king Oswy of Northumbria was defeated by an alliance of Penda of Mercia and some British kings at the urbem, "city," of Iudeu and forced to hand over the Atbret Iudeu, "treasure of Iudeu" (Morris 1980:38, 79). There is also a reference to the merin iodeo, "sea of Iodeo" or Firth of Forth, in the Gododdin poem (Jackson 1969:108; Jarman 1988: line 944), and Iodeo is probably the same as Iudeu. Guidi/Iudeu has sometimes been equated with Cramond (Hunter-Blair 1947:27-28; Rutherford 1976:443) but opinion now generally favours Stirling instead (Alcock 1981:175-76; Jackson 1981; Jarman 1988:147, n. 944). There are two important pieces of evidence against the identification of Guidi/Iudeu with Cramond. Firstly, as it possessed a significant Roman past, Bede would probably have termed it a civitas rather than an urbs (Campbell 1979). Secondly, as the defences had been slighted and the ditches filled, the site would not have been easily defensible. It is therefore unlikely that Oswy would have retreated there when pursued by his enemies.
Another documentarily attested site is Rathinveramon, where two kings died. Domnall, son of Alpin, rex Pictorum-king of Picts-was killed in A.D. 862; and Constantine, son of Culen, ri Alban-king of Scotland-was slain by Kenneth, son of Malcolm, in A.D. 995. The deaths of two different kings in the ninth and tenth centuries suggest that this was an important site. The first element in the name Rathinveramon, which means "fort at the mouth of the river Almond," is the Goidelic or Q-Celtic element rath, which means "fort." Rath is an uncommon place-name element for Scotland and refers to a site with a bank and ditch. O. G. S. Crawford (1949:60) believed that rath indicated a Roman fort and equated Rathinveramon with the Roman fort of Bertha in Tayside, and this has generally been accepted. There are, however, two river Almonds in Scotland: one in Tayside, beside which Bertha is located; and one in Lothian, which flows past Cramond. It is therefore conceivable that Rathinveramon is Cramond rather than Bertha, a suggestion that has been made previously (A. Anderson 1922, 1:518). Rathinveramon is in fact the Goidelic equivalent of the Brythonic name Caramonde. It is conceivable that the Scottish Goidelic-speaking incomers adapted and gaelicised the existing Brythonic place-name. The evidence for activity at Cramond at this time, in the form of the eighth- or ninth-century enamelled mount and possibly the ring with runic inscription (see Later Artefacts above), supports this suggestion. In contrast there is no evidence for any post-Antonine activity at Bertha (Adamson and Gallagher 1986:203; Callander 1919:145-52).
The problem is that the documentary sources that mention Rathinveramon were generally written down several centuries after the events they describe. This allowed plenty of opportunity for confusion to arise concerning which of the two rivers Almond Rathinveramon was sited beside. Such confusion is best exemplified by the fourteenth-century Scottish historian John of Fordun, who places the events of A.D. 862 in Tayside (IV.15) and those of A.D. 995 in Lothian (IV.34); this was repeated by the twentieth-century historian Alan Orr Anderson (1922, 1:291 n. 5, 518 n. 7; 1922, 2:782) when he quotes Fordun. The source with the best claim to be contemporary is the Annals of Ulster, which do not state where Domnall and Constantine were killed (MacAirt and MacNiocaill 1983:318-19, 424-25). In contrast the sources that seem to provide the most geographical detail are those like the Prophecy of Brechan, which was not composed until the 1160s and whose reliability is open to question. The river Almond in Tayside is located in an area that was an important focus of Early Medieval royal activity with sites such as Scone, Forteviot, Meigle, and Dunkeld, which encouraged early authors to place Rathinveramon on the Almond in Tayside, perhaps erroneously. The documentary sources do not allow an unequivocal decision between the rivers in Lothian and Tayside, so it is at least possible that Rathinveramon should be equated with Cramond rather than Bertha.
Edinburgh--Oppidum Eden--and therefore presumably Cramond, was under Northumbrian control from the mid-seventh century until the reign of the Scottish king Indulf (A.D. 954-62), when it passed into Scottish hands apparently without a battle (A. Anderson 1922, 1:468). When Constantine was killed at Rathinveramon in A.D. 995, Cramond was under Scottish control and his presence there is clearly possible. In A.D. 862, when Domnall was killed, Cramond was still under Northumbrian control. The frontier between Northumbria and Scotland was, however, probably relatively close to Cramond and the Scottish kings were closely interested in this area. Domnall's presence at Cramond in A.D. 862 is therefore quite plausible.
The exact territorial boundaries of the Votadini are problematical, the best source being Ptolemy's Geography, but other Roman forts that the Votadini could have occupied in their heartland along the southern side of the Firth of Forth were Carriden and Inveresk. Carriden has not produced any evidence of post-Antonine occupation during the Roman period, and the next clearly attested occupation is during the thirteenth or fourteenth century (Bailey 1997). There is a post-Roman phase consisting of a gully, drain, and post-hole, which can not be precisely dated; it occurs some considerable time after the Roman abandonment but prior to the thirteenth- or fourteenth-century occupation (Bailey 1997:591). This evidence is extremely difficult to interpret and only further work will elucidate its nature and precise date. Excavations on both the fort and the vicus at Inveresk indicate that it was not occupied during the Severan campaigns. In fact there is no evidence for occupation between the Antonine occupation and the thirteenth or fourteenth century. Two coins of Gallienus (A.D. 260-68) and Constans (A.D. 341-46) are reported to have been found at Inveresk (Robertson 1983:408) but are not mentioned in the excavation report (Thomas 1988:171, fiche 2 A.12-B.1) and must be assumed to be erroneous. The place-names of Carriden and Inveresk do not retain any elements that preserve the memory of the presence of a Roman fort. The fort of Carriden at the eastern terminus of the Antonine Wall, which also falls within Votadinian territory, might appear to be derived from Caer Eidyn, but this is problematical. Kenneth Jackson argued that the early forms of the name such as Karreden and Karedene can not be derived from this source but David Dumville has challenged this (Dumville 1994). Even if Dumville is correct his argument suggests that the name Carriden may be relatively late as he suggests that this fort was originally named 'End of the Wall' [Penguaul, Cenail, Peneltun] a topographical descriptive term later applied to Kinneil and not necessarily indicating any occupation. The third fort within Votadinian territory was Inveresk. This name contains the Gaelic element inbhear, "in-bring", denoting the junction or confluence of a river (Watson 1926:148), and must therefore post-date the Scottish take-over of the area in the tenth century; in any case it makes no reference to the presence of the Roman fort.
There is, however, some evidence that the site of Inveresk still retained some importance. The location of the Medieval church at Inveresk (see The Medieval Church above) is suggestive, as are documentary sources which show that it may have been the "centre of a territorial arrangement" (Proudfoot and Aliaga-Kelly 1997:40-42). The evidence suggests that Carriden's position at the end of the Antonine Wall was no longer of strategic importance, but that Inveresk's position at the end of Dere Street may still have been significant, although the site itself was probably not occupied. The continued occupation at Cramond suggests that links by sea may have been of greater importance than land-based links.
The archaeological, place-name, and documentary evidence suggests that Cramond was an important site in the post-Severan period, although its exact nature is unclear. In general Roman forts do not seem to have been particularly favoured as locations for native settlements in Scotland, although the evidence is probably not as negative as suggested by Dark (1992:111, n. 7), who focused solely on the fifth and sixth centuries. Apart from Cramond there is evidence for a native fort at Inchtuthil, Grampian, which re-used Roman masonry (Abercromby et al. 1902). Perhaps less significant are a ninth- or tenth-century silver-gilt penannular brooch, iron sickle, and sword found at Carronbridge, Dumfries and Galloway, as their location at a Roman fort has been described as "almost certainly coincidental" (Owen and Welander 1995). The other Roman fort apart from Cramond with substantial evidence of Severan occupation is Carpow. While it lacks archaeological evidence for occupation after the early third century (Birley 1963; Wright 1974), it is mentioned in an early documentary source as Ceirfuill. The source refers to a lapidem juxta Ceirfull, "stone beside Ceirfuill" (Chadwick 1949:9-10), which may relate to a rather later fragment of a cross-slab from Carpow House (Cessford 1996). Although this does not demonstrate that Carpow was occupied, it indicates its continuing importance as a focal point in the landscape. The evidence from Ruberslaw, Borders (Alcock 1979:134; Curle 1905), is rather different, as the re-use of Roman masonry here involves a hilltop site and the stone probably came from a signal station tower rather than a fort.
Most known high-status sites in Scotland of the post-Roman period are small- to medium-sized hillforts, whereas Roman forts, which were larger, were probably too large for the requirements of the period. They were also not located in naturally strong defensive sites. There are, however, a number of important undefended sites of this period, the most notable being Forteviot, Tayside (Alcock and Alcock 1992:218-41). The continued activity at Cramond and other Roman forts may be of a similar nature to Forteviot. Roman forts were located at nodal communication points on the Roman road network, which probably continued in use during the early Medieval period, and in the case of Cramond the site was also well situated with regard to communications by sea. Such forts also acted as quarries of raw materials, notably for stone but also nails, glass, coins, and other artefacts that were re-used during the early medieval period in Scotland. Their history and a lingering sense of romanitas may also have meant that controlling Roman fort sites provided a source of authority and legitimacy to native dynasties.
Dark (1992; see also Snyder 1996:47-48) has suggested that the evidence of fifth- and sixth-century refurbishment of defences, timber halls, inscribed tombstones, and Germanic artefacts at forts on Hadrian's Wall indicates that they were reused as part of a still-functioning military frontier with high-status British occupation and the use of Germanic mercenaries. Whilst there are certain similarities between the sub-Roman evidence from Hadrian's Wall and that from Cramond, Dark's model appears unsuitable for Cramond. There are no indications that the defences at Cramond were still functional after the Severan period or that there was a military presence at the site, so a martial interpretation is difficult to substantiate. The enamelled bronze circular mount and the location of Cramond parish church both suggest that there was a religious component to the occupation at Cramond. The most likely scenario is that ownership of the fort at Cramond passed into the hands of the native dynasty of the Votadini after its abandonment by Roman forces. There is no evidence that occupation of the site was continuous and what the discoveries may instead reflect is a continuity of royal ownership. Cramond was probably controlled by a succession of royal owners in a manner similar to that suggested for some Roman forts in England, where churches occur from the seventh century onward (Bidwell 1997:108-9). The site continued in use after the Anglo-Saxon take-over of the region in the early seventh century and may have remained in use throughout the Anglo-Saxon occupation until the tenth century. It is even possible that the Scots subsequently occupied it.
The evidential basis for assessing the Early Historic occupation at Cramond is fragmentary and open to a range of interpretations. Nevertheless, given the focus upon substantial Roman structures during previous excavations at the site combined with the frequently insubstantial nature of building remains and the general lack of dateable artefacts during the Early Historic period in Scotland, the body of evidence that does survive is perhaps as much as might reasonably be anticipated. Cramond appears to be a potentially important site for understanding the complex interactions and transitions between different political and ethnic groups in Northern Britain during the second half of the first millennium A.D. If its identification with Rathinveramon is correct, it is a site where events of considerable historical significance took place. Recent work on Hadrian's Wall has shown the potential for careful excavation of Roman fort sites to uncover the less obvious evidence for later periods of occupation and it is to be hoped that this will eventually be repeated at Cramond.