Geophysical Survey at Aggersborg, North Jutland more

Geophysical survey at Aggersborg, North Jutland Helen Goodchild Hannah Brown & Søren M. Sindbæk University of York 1 © Helen Goodchild, Hannah Brown and Søren M. Sindbæk 2011 Permission is granted to copy and distribute this document in unaltered form only, for the purposes of teaching, research, private study, criticism and review, provided that the original authors are acknowledged. Cover image: Processed gradiometer survey image of the inner fortress area, Aggersborg © Authors and University of York Archaeology Department Expedition to Aggersborg. 2 Geophysical Survey at Aggersborg Circular Fortress, Denmark, April 2011 Summary of Results An integrated geophysical survey using fluxgate gradiometry and ground penetrating radar was conducted in April 2011 on the site of the Viking period settlement and ring fortress Aggersborg, Denmark. The 5.5 ha gradiometer survey identified concentrations of positive magnetic anomalies conforming to sunken-featured buildings adjacent to areas where such features are noted in excavations. The interpretation of these features as sunken-featured buildings is confirmed by the radar survey. A total of 89 potential sfb’s are identified, mostly in the eastern half of the fortress area, and in two rows west of the fortress area. A few anomalies show comparatively high magnetic values and have been interpreted as possibly being pits containing quantities of Iron Age pottery. Relating to the fortress the gradiometer survey identified preserved timber structures in the SW section of the rampart. Some large anomalies are found in locations, which may indicate a relation to the fortress: at a road corner, in the middle of a courtyard, or in the gable room of a house. In the areas of the north gate features were note, which are consistent with a defensive outwork, similar to traces observed in the fully excavated western gate. Despite strong modern disturbances, some anomalies hint at preserved structures in the area of the south gate. The distance between the traverses in the gradiometer survey imply that post build structures may easily have been overlooked. Thus no further post buildings are securely identified. It would be desirable to carry out further fieldwork, including surveys outside the fortress area, where more settlement remains and possibly a burial ground are to be expected, and to conduct tests with gradiometer surveys with closer spacing between the traverses, in order to determine if post buildings can be identified. Introduction Site location Aggersborg circular fortress is located on the north coast of the Limfjord, in Northern Jutland, Denmark. The site was once located in an area of strategic importance, near a crossing of the Limfjord, but its importance diminished after the silting up of the northern outlet to the Skagerrak and the North Sea. The site itself is centred at 56.995457 latitude, 9.254838 longitude, and lies only 10m above current sea level, and approximately 350m from the current coastline (Figure 1). The bedrock of chalk is overlain by glacial till, with prequaternary deposits near the coast (Møller 1986). 3 Fig. 1. Location of Aggersborg Circular Fortress (Natural Earth & Google Earth) Site description and history Aggersborg (Aggersborg parish, Vester Han herred, Northern Jutland) is the largest known fortress from Viking Age Scandinavia, and a site considered for inscription on UNESCO's World Heritage List. Settlement at Aggersborg had two main phases: the circular fortress itself, and the pre-fortress settlement. Some 20% of the site was excavated by the National Museum of Denmark in 1945-54 (Schultz 1948; Roesdahl 1976; Roesdahl 1986; Pedersen, Roesdahl, & Sindbæk in press). Further small-scale excavations were carried out in 1970 (unpublished) and 1990 (Ulriksen 1995). The pre-fortress settlement dates to c. 800-980 and consisted of post buildings together with sunken feature buildings. At least a dozen post buildings are noted, among which one particularly large building, located in the area just north of the western gate of the later fortress, is presumably the residence of a local magnate. Some 150 sunken featured buildings are noted, mostly in the eastern half of the excavated area. To the east of the fortress area and in the SE quadrant of the fortress area thick layers of soil, up to more than a meter in depth, had accumulated due to a combination of windblown deposits and cultivation. 4 Fig. 2 Reconstruction of Aggersborg fortress, as illustrated in the on-site exhibition by Lisbet Stevens. The circular fortress is one of the largest archaeological sites in Denmark with an internal diameter of c. 240 metres. It is believed to date to c. AD 980, but was only in operation for between 5-20 years. It consisted of a circular rampart, constructed of turf and spoil from the ditch, and strengthened by internal timber structures (fig. 2). Four entrance gates were oriented to the principal compass points. Two axial streets between the gateways divided the interior of the fortress into four quadrants, each of which are believed to have accommodated 12 large wooden buildings, each c. 31,5 m long, arranged in groups of four in strict, rectangular courtyards. The whole arrangement is laid out with meticulous geometric accuracy. The ramparts and axial streets were marked out on the ground by low earthworks in the 1990s. The site is located on relatively flat ground, and regularly maintained, however, it is prone to heavy wind, which occasionally caused issues with data capture. It has also led to differential depth of deposits across the site. The fortress and immediate surroundings is a scheduled monument, while the surrounding fields and the manor house with gardens to the south are in private ownership. The manor with surrounding fields is documented as a crown estate from the late Viking period until being sold in 1579 (Roesdahl in Roesdahl (ed.) 1986, 95ff). It has since formed a private manor in changing ownership (Nielsen 1986). The field in which the fortress is located belongs to the demesne of the manor, and has apparently never been built upon since the Viking period. Cadastral maps document the land use at regular intervals since 1803 (Nielsen 1986, fig. 10). 5 Survey objectives Since 2008 a team consisting of Dr. Else Roesdahl, University of Aarhus, Anne Pedersen, curator at the National Museum of Denmark, and Søren M. Sindbæk, University of York, has worked towards presenting the first comprehensive analysis of the archive and finds from the site. These were archived when the site excavator, C.G. Schultz, died unexpectedly in 1958. The analysis of the excavations, now in press, has highlighted some disturbing gaps in our knowledge of the site. It was judged that some of these problems could be solved by geophysical investigations. Aggersborg was considered to be a highly promising target for geophysical investigations because: a) the site formation – cut features filled with organic fill in chalk and chalk-rich glacial till or drift sand – should give good contrast for magnetometry and GPR; b) earlier excavations outlines a clear interpretational frame for the survey, while leaving 80% of the site untouched; c) earlier excavations also provide essential research questions, which could be tested and solved by the prospection. The key questions outlined for the survey were: a) How far does the concentration of sunken-featured buildings extend? Excavations in the eastern part of the site show a dense concentration of these. In the western zone they seem to disappear, while another cluster may be indicated by test trenches further to the east. Some field notes suggest, however, that the fortress area may have had leveling layers, which were only recognized as such in a late phase of the excavations; it is possible, therefore, that the concentration of SFB’s is in fact continuous. b) If some parts of the fortress area had leveling layers, as suggested by field notes, can the extent of these be defined? c) Are there any further longhouses from the pre-fortress settlement? The excavated parts of the site appear to comprise four farms, the individual histories of which can be re-constructed in some detail. This interpretation rests on the assumption, however, that there are no further longhouses in the immediate vicinity of the excavated areas. Further substantiation of this condition would be highly valuable for strengthening interpretations. The longhouses could be expected to show up in the survey, since later accumulation of anthropogenic soils has generally preserved internal features including hearths, which could be expected to yield strong magnetic signals. d) How far do the areas with deep, anthropogenic soils extend? Deep soils, providing exceptional conditions of preservations, were noted in some parts of the site, but their extent was not accurately mapped. Identifying their extent will provide an essential background for interpreting the distribution of artifacts and features in the excavated areas. e) Can further elements of the fortress plan be identified? While the general structure of the fortress can be established with reasonable clarity from the excavations, it would be valuable to see if the surveys can identify the general pattern in areas which are not excavated, or point to deviations from the plan. In particular it would be interesting to see if features could be observed in the areas of the north and south gateways and in the spaces between buildings, which have not been targeted by excavations. 6 f) Is there evidence of a conflagration? The excavators noted accumulations of “ash” in several areas, which they took to indicate a violent destruction of first the village, then the fortress; but there is some uncertainty whether these may in fact be simply decomposed sods. The effects of a major blaze might be detectable in the ground of building plots and sunken features. g) Are there areas with wet preservation? No excavated areas have provided material for dendrochronological dating. Such material might provide an invaluable key to the historical interpretation of the fortress. There might be a hope that good, wet conditions exist in the unexcavated, southern section of the fortress rampart. GPR, being highly sensitive to soil humidity, may help to define the most likely areas for furthe r exploration. Fig. 3 Hannah Brown (left) and Frida Norstein conducting GPR survey SW of Aggersborg fortress. In the background the church and the reconstructed rampart are seen. Seen towards NE. Photo: S. Sindbæk. 7 Methods Fieldwork took place across five days, from the 11th to the 15th April 2011. The survey was carried out by Hannah J. Brown, Helen Goodchild and Søren M. Sindbæk, The University of York Department of Archaeology, assisted by student volunteers Frida E. Norstein and Peter Aherne. The investigations were undertaken in collaboration with the regional museum, Museumscenter Aars, from which curator Bjarne Nielsen negotiated agreements with land-owners, provided additional equipment, and, together with Simon Kjær Nielsen, personal assistance. Further assistance was generously provided by Aalborg Museum, from which Jan Sloth-Carlssen conducted GPS surveying of the grid. Access to the lands surrounded the scheduled fortress was kindly granted by land owner Regitze Johnsen. Two geophysical methods were used: fluxgate gradiometry and ground penetrating radar (fig. 3 and 4). The instruments used were a dual sensor Bartington Grad 601, and a Noggin ground penetrating radar system utilising both 500 and 1000 MHz antennae. Fluxgate gradiometry was chosen as the primary technique due to the nature of the site. The archaeology was likely to consist of cut features such as ditches or postholes, for which this method is well-suited; this survey method also offers the benefit of rapid coverage. The radar was utilised to target potential buildings discovered via the gradiometry survey. The survey grids were laid out with a GPS receiver by Jan Sloth-Carlssen, Aalborg Museum. The area was divided into 30m x 30m grids, oriented on the standard Danish National Grid (UTM sys 32ed50), and located with the corners of each square at even, full 10 m points in this grid. Fig. 4 Helen Goodchild conducting gradiometer survey in the Aggersborg fortress area. In the background the church and the reconstructed rampart are seen. Seen towards N. Photo: S. Sindbæk. 8 Fig. 5. Location of the gradiometry survey Gradiometry survey The gradiometry survey covered an area of approximately 5.5 hectares (fig. 5). Initially data was collected along traverses separated by 1m with 12.5cm sample intervals, but this was found to provide insufficient detail, and so the resolution was increased to 50cm traverse separation. Measurements were recorded at the 0.1 nanotesla (nT) instrument range. The survey was carried out in zigzag configuration with traverses walked east-west. Methods of data processing/variables used The gradiometry data were processed using Geoplot software, then exported to Surfer for presentation (fig. 6). It was necessary to process the different resolution surveys separately (the area outside the fortress at 1m x 0.125m, and inside the fortress at 0.50 x 0.125m). Data has been subjected to minimal correction processes: due to the zigzag collection strategy, it was necessary to apply a zero mean traverse tool; the data was clipped at 3 standard deviations around the mean; it was necessary to ‘destagger’ some grids. For the processed, composite image a zero mean grid tool has also been applied (fig. 7). 9 Fig. 6. Raw gradiometry data 10 Figure 7. Zero mean grid, zero mean traverse, clipped to three standard deviations and destaggered 20 nT 15 nT 10 nT 5 nT 0 nT -5 nT -10 nT -15 nT -20 nT 11 Fig. 8. Interpretation of gradiometry data. 12 Fig. 9. Simplified interpretation of gradiometry data: possible archaeological features. 13 Figure 10. Simplified interpretation of gradiometry data: possible archaeological features, combined with plan of fortress and excavation trenches. Results Description /interpretation There are many cut features detected by the gradiometry survey. These are, however, very difficult to distinguish – particularly in the north-western and central areas due to the effects of glacial scouring of the chalk bedrock. Striations oriented northwest-southeast in the results which likely represent the fill of these scours with other material such as clay and silt (fig. 8, light blue features). A number of anomalies can be related to known modern interventions: fences and wires, excavation trenches, and landscaping in the garden area in the southern area nearest to the manor house (fig. 8, light green features). Several weak, linear features may represent ditches. Two straight features, which line the course of the southern third of the NS axial street of the fortress, presumably relate to the fortress and may represent drainage ditches. Further linear anomalies observed in all four quadrants may also possibly relate to the fortress, as drains leading from the houses were observed in the excavated sections. However, these features may also relate to the pre-fortress settlement, or to medieval or modern field systems. Some strong, straight features oriented NE-SW in area of the fortress’ western gate and in the NW quadrant of the fortress presumably relates to a road on this orientation seen here on early modern maps, and noted during the excavations. A linear feature running NW-SE across the northern grid represents a road, which is seen on 19th century maps, and whose continuation was also noted during the excavations. A cluster of linear features in the NW quadrant of the fortress area appears to define rectangular enclosures. These could possibly relate to the large house (presumably a manor house) found in this area. However, a good part of the features in question are located within excavated areas, and no corresponding archaeological features were noted in the respective areas. The apparent features may therefore represent a coincidental combination of geological features and possibly modern drainage ditches, dug since the excavations in the 1950s. From the fortress period of activity, the main road axes are clear, though this is likely to reflect the modern reconstruction rather than the original road (fig. 10). The partially reconstructed external ditch also appears on the western side of the rampart and to the east of the northern gate. An anomaly noted in the corner of the outermost grid surveyed to the SE of the fortress area is also likely to reflect the ditch. The other clear fortress-related features are the structures within the ramparts. In the south west corner where the ramparts were low enough to survey, there appear radial structures which probably represent timber supports. Faint traces of similar structures may possibly be seen in the southernmost, central grids; however, these are difficult to affirm, as the area surveyed close to the farm suffers from a large quantity of ferrous material near the surface. Among the remaining features, there are a large number of medium-strong approximately circular magnetic anomalies (fig. 9). These may represent a variety of features including pits, hearths or sunken floors. Based on a comparison with the excavated areas, however, many of the large features, measuring 24 metres in diameter, may represent sunken feature buildings (sbf’s) belonging to the pre-fortress settlement. With the exception of weak, dubious features, large circular anomalies cluster in three areas. A first group is found in the eastern half of the fortress area. This cluster extends from a large excavated area where a concentration of sfb’s, a fact which strengthens the interpretation of these features as relating to further similar features. Some 57 possible sfb’s can be counted in this cluster (fig. 11) Another group of anomalies are noted in the northern half of the NW quadrant of the fortress area. However, most of these appear to be too small to represent sfb’s. Interpretation of the gradiometer data in this area is rendered difficult by the presumably geological features noted above, and it is likely that some of the anomalies in this area represent geological features. Only eight features can be noted, which are consistent with sfb’s in the western half of the fortress area. This is consistent with the observation that few sfb’s are noted in the excavated parts of this area. The third group of large magnetic anomalies are found in the survey west of the fortress area. Most of anomalies in this area are concentrated in two rows, converging on the area around the west gate of the fortress. The size and outline of 24 of these features are consistent with sfb’s. During excavations a number of sfb’s on the alignment of the observed rows were noted in trial trenches. In total, the survey would suggest that at least 89 sfb’s can be added the c. 150 encountered during excavations. The two rows may represent separate two phases, since excavated buildings in the western row, which appear to aim at the western gate of the fortress, contain some of the few finds which can be related to post-fortress (i.e. 11th century) activity. The eastern row continues towards a point just south of the western gate area where a group of sfb’s were found during excavations to have been cut by features relating to the fortress. 15 Fig. 11. Summary of interpretation: features identified as possible sunken featured buildings. At least three of the presumed sfb’s in the NE quadrant of the fortress area are represented here by a positive magnetic anomaly surrounded by negative magnetic readings; while this may represent a negative ‘halo’ associated with the predominant positive field anomaly, GPR data (explained further below) suggests that it relates to the existence of a low wall of subsoil material (chalk). Such walls were observed on several occasions in the excavated areas. There are also many smaller anomalies which could represent pit alignments or similar cut features, though again these are difficult to distinguish due to the background geological activity. It is probable that some of these relate to post buildings, although none can definitively be identified as such. As the great majority of post holes identified during the excavations were substantially less wide than the distance between the traverses (0.5 m), it can be expected that even under ideal conditions only less than half of all post holes would be noted. Hence we should not expect to see a full system of post holes relating to post buildings to be exposed. In general, the magnetic anomalies are characterised by very slightly raised positive values in the range of 1-5 nT, occasionally reaching values up to 20 nT. However, three features in the NE quadrant of the fortress area show high magnetic values of 59.13 nT, 22.64 nT and 47.61 nT respectively (fig. 12, reckoned from west to east). While they may be attributable to a ferrous source, it is worth noting that some pits in nearby excavated areas produced large volumes of pottery from the early Iron Age, the only evidence of a settlement from this period. A substantial quantity of fired pottery might explain the levels of magnetism observed in the three features in question. 16 20 nT 15 nT Fig. 12 Location of features with strong magnetic signal in the eastern part of the fortress area, possibly early Iron Age pits with 10 nT pottery. 5 nT In the western part of the surveyed area, in the southernmost part of the NW quadrant of the fortress area, two very strong positive 0 nT features (up to 99.66 nT) with a negative halo can be identified as 20 nT a set of principal posts from the largest excavated pre-fortress -5 (fig. building nT 13). The strong magneticnT signature is presumably 15 created by material dumped in the holes after excavations, -10 nT possibly ferrous material or stones10 nTburned clay from the and nearby central hearth of the building, which was demolished in the last stages of the excavation. 5 nT Figure 13. Location of features with strong magnetic signal in the western part of the fortress area: principle posts of a large post -20 nT 0 nT building, and an unknown negative disturbance. -15 nT In the same general area we find a fairly -5 nT strongly negative feature, large, surrounded by a dark halo of positive “shadow”. The nature of this feature is unknown, but since it occurs directly on the reconstructed -10 nT path of the E-W axial street it may be surmised to represent a modern disturbance. -15 nT A few features command particular interest when seen in relation to the presumed fortress plan (fig. 14). -20 nT At least three large positive anomalies are placed in locations which may indicate a relation to the fortress: at a road corner, in the middle of a courtyard, or in the gable room of a house. Fig. 14. Gradiometer survey of the central southern part of the fortress area with indication of excavation trenches (yellow lines) and presumed roads and buildings (red lines). Green squares indicate anomalies which are placed in conspicuous locations in relation to the fortress plan, and which may relate to the fortress. 17 Minor anomalies may relate to wall posts or principal posts, hearths or other features; but as explained above the spacing of the traverses implies that it is unlikely that such features can be traced consistently. GPR survey The ground-penetrating radar survey covered five separate grids of approximately 1750 square meters combined. The data was collected using a Noggin+ ground penetrating radar system utilising 500 MHz antennae, except for one grid (grid 4) which was surveyed using a 1000 MHz antennae. The GPR was mounted on a Noggin SmartCart. The distance from the base line was measured using the odometer of the cart. The single profile georadar data was processed using EKKO view and EKKO view deluxe software. Data were filtered using dewow filter and noise filter (3 points). In order to compensate for loss and dissipation of energy a SEC (Spreading and Exponential Compensation) gain was applied to the data (attenuation: 1.0; start value: 1.0; Gain Max: 300). Data was merged to a 3D data volume based on the average raw amplitude. This data volume was cut into horizontal slices, so called time-slices using EKKO Mapper software. The time-slices have been converted into depth-slices using a constant estimated conversion velocity for the radar signal in the subsurface of 9.8cm/ns, based on calibrations of hyperbolae in several profiles. Although this process has a considerable margin of error, it successfully images the relative depths of structures. Holes in the data are due to obstacles (trees, stones, electric fences). Figure 15: Location of GPR Grid 1 in relation to gradiometer survey. GPR Grid 1 The survey was located south west of the fortress outside the rampart, in an area where strongly magnetic anomalies in the gradiometer survey suggested the location of sunken features. The survey covers an area of 16 x 20 m and was located in relation to the gradiometer grid (fig. 15). 65 profiles were collected from south to north with 25 cm spacing and 5 cm inline trace spacing. The survey was walked from east to west. The base line of the survey follows the eastern edge of the gradiometer grid, and the south-east corner of the survey area is located 2,0 m north of the south-east corner of the gradiometer grid. The horizontal slices show a clear depiction of plough furrows, proceeding N-S, at a depth of up to at least 40 cm (fig. 16). In the centre of the grid a strong linear feature, with the same orientation as the plough furrows, is noted at a depth of up to at least 50 cm at the 10 m mark of the E-W axis. This feature, presumably a narrow modern drainage ditch, is also seen in individual transects (fig. 17). At a time-depth of c. 10-15 ns, calibrated to a depth of c. 50-70 cm below the surface, several further linear features (modern drainage ditches?) is noted along with two more substantial features. The larger of these, in the SW corner of the grid, is roughly circular and has a diameter of c. 3 m and is consistent in size and shape with a sunken featured building. This feature is clearly visible in transects, and appears in transect to reach a depth of up 18 to c. 65 cm below the surface. A smaller feature seen in transects to extend downwards from the rim of this feature may possibly indicate a gable post. In the transects the bottom of the large circular feature appears to be sloping downwards towards east; however, this is due to the fact that the surface of the surveyed area slopes towards the west. The data has not been corrected for topography as a topographic survey was not conducted. In fact the bottom of the feature is approximately horizontal. No further features are noted at greater depth. Although the survey appears to give clear indication of archaeological features, and there are similar indications in the gradiometer survey, it is difficult to match the two surveys. The linear features noted in the GPR survey are presumably too slight to show up in the courser resolution of the gradiometer survey. The large circular feature may be identified with the westernmost one of the three strong anomalies indicated in the gradiometer survey over the same area. The difference in absolute positions is likely to be due to the horizontal displacement of the magnetic field, which varies depending other features around it. Fig. 16 GPR grid 1, horizontal slice for time-depth c. 8-12 ns, calibrated to a depth of c. 40-55 cm below ground (left) and for time-depth c. 10-15 ns, calibrated to a depth of c. 50-70 cm below ground (right). Fig. 17 GPR grid 1, transect 4, at 1.0 m on the N-S axis. 19 Grid 2 Grid 4 Grid 3 Grid 2 Grid 5 Fig. 18 Location of GPR grid 2 to 5 in relation to gradiometer survey. GPR Grid 2 The survey was located in the NW quadrant of the fortress area. The gradiometer survey suggested strongly magnetic anomalies in this area, together with a circular feature which might be interpreted as a cut feature. Both were located in what was expected from comparison with the reconstructed fortress plan to be a courtyard surrounded by a group of buildings. The strong magnetic anomaly corresponded to the location of a test trench from the excavations in 1990. The survey covers an area of 30 x 18 m and was located in relation to the gradiometer grid (fig. 18). 73 profiles were collected from north to south with 25 cm spacing and 5 cm inline trace spacing. The survey was walked from east to west. The base line of the survey follows the eastern edge of the gradiometer square, and the north-east corner of the survey area is located 2,0 m south of the north-east corner of the gradiometer grid. The cart was started with the centre of the radar above the base line. The horizontal slices (fig. 19) show a clear detection of a circular system of plough furrows, following the line of the reconstructed rampart, down to a time-depth of c. 7 ns, calibrated to a depth 0f c. 35-40 cm. At a very shallow depth, less than 20 cm below the surface, a circular feature stands out. This feature appears to correspond to the location of a ring or arc of mushrooms, a fairy ring, which could be noted on the surface (due to the season no mushrooms were visible, but the ring was apparent in the colour of the grass). Possibly the feature apparent in the GPR images is caused by water or minerals retained by the fungus’ mycelium. Apart from the ring feature, there are traces of a disturbance around x=21/y=16 from the surface down to a time-depth of c. 8 ns (c. 40 cm). This may represent ploughed-up material or an animal burrow. Below the plough soil, faint traces can be seen around y=5-12/x=10-15 of the former excavation trench at a time-depth of c. 10-12 ns (c. 40-50 cm). No further features are noted. This result, though disappointing, is in agreement with the results from the excavation and the reconstructed fortress plan, which did not indicate any archaeological features in this area. 20 21 Fig. 19 GPR grid 2, horizontal slices for time-depth c. 2-13 ns, calibrated to a depth of c. 8-60 cm below ground. North is down. 22 GPR Grid 3 The survey was located in the NE quadrant of the fortress area. The gradiometer survey suggested very few and weak magnetic anomalies in this area; however, based on the reconstructed fortress plan traces of buildings as well as roads might be expected. The survey covers an area of 22 x 12 m and was located in relation to the gradiometer grid (fig. 18). 49 profiles were collected from north to south with 25 cm spacing and 5 cm inline trace spacing. The survey was walked from east to west, starting in the north-east corner. As an experiment the 1000 MHz antenna was used for this grid. The base line of the survey follows the eastern edge of the gradiometer grid, and spans over two adjacent squares. The north-east corner of the survey area is located 4,0 m north of the south-east corner of the northernmost of the two gradiometer grid. The cart was started with the centre of the radar above the base line. The horizontal slices (fig. 20) show a clear image of a circular system of plough furrows, following the line of the reconstructed rampart, down to a time-depth of c. 7 ns, calibrated to a depth 0f c. 35-40 cm. At a shallow depth, less than 20 cm below the surface, a group of linear features stand out. These are almost certainly mole tunnels, as mole hills were observed in the surveyed area. A strong anomaly is noted around x=7.5/y=4.0. As the signal occurs only in a single line it is likely to be due to electromagnetic interference from an external source, such as a mobile phone. No clear features were observed below the plough soil. As with grid 2, this result is in agreement with the results from the gradiometer survey and the reconstructed fortress plan, which only indicates faint archaeological features in this area. 23 Fig. 20 GPR grid 3, horizontal slices for time-depth c. 0-15 ns, calibrated to a depth of c. 0-70 cm below ground. North is down. 24 GPR Grid 4 The survey was located in the NE quadrant of the fortress area, just west of grid 3. The gradiometer survey suggested the existence of several large, albeit rather weak features in this area, together with one moderately strong feature. Based on the reconstructed fortress plan the latter could possibly relate to the hearth of a building. The survey covers an area of 22 x 12 m and was located in relation to the gradiometer grid (fig. 18). 49 profiles were collected from north to south with 25 cm spacing and 5 cm inline trace spacing. The survey was walked from east to west, starting in the north-east corner. The base line of the survey follows the eastern edge of the gradiometer grid, and spans over two adjacent squares. The north-east corner of the survey area is located 4,0 m north of the south-east corner of the northernmost of the two gradiometer grid. The cart was started with the centre of the radar above the base line. The horizontal slices (fig. 21) show a clear image of a circular system of plough furrows, the same as seen in grid 3, down to a time-depth of c. 8 ns, calibrated to a depth of c. 40 cm. Below this, at a time-depth of c. 10 nm (c. 45 cm) there are traces of an older field system with furrows running N-S, i.e. the system used until the partial reconstruction of the fortress in 1990. At lower depth several linear features are noted, which may relate either to Viking period features or to later drainage ditches. There are no clear indications corresponding to the anomalies noted in the gradiometer survey. It is possible that some very slight difference noted in areas at time-depth 10-18 ns may relate to the presumed features, but the evidence is inconclusive. Given the excellent penetration indicated by the reflection of the two plough systems, any cuttings in this area are likely to be very shallow or to differ only slightly in their fill from the surrounding natural. 25 Fig. 21 GPR grid 4, horizontal slices for time-depth c. 0-25 ns, calibrated to a depth of c. 0-100 cm below ground. North is down. 26 GPR Grid 5 The survey was located in the NE quadrant of the fortress area, a good 30 m south of grid 3. The gradiometer survey suggested the existence of two particularly clear circular features in this area, presumably two sunken-featured buildings. The survey covers an area of 22 x 12 m and was located in relation to the gradiometer grid (fig. 18). 49 profiles were collected from north to south with 25 cm spacing and 5 cm inline trace spacing. The survey was walked from west to east, starting in the north-west corner. The base line of the survey follows the western edge of the gradiometer grid, and the north-west corner of the survey area is located 2,0 m south of the north-west corner of the gradiometer grid. The cart was started with the centre of the radar above the base line. Due to the direction of the survey the slices are mirrored N-S. The horizontal slices (fig. 22) show the familiar image of a circular system of plough furrows down to a time-depth of c. 8 ns, calibrated to a depth 0f c. 40 cm. at a time-depth of c. 7-8 ns, calibrated to c. 35-40 cm below the surface, two ring-shaped features are noted; the larger of the two correspond to the location of the most clearly marked anomaly in the gradiometer survey. At a slightly lower time-depth, c. 10 ns (c. 50 cm) the same features stand out, together with at least four further ones, as areas with a reflection which deviates from the background. These observations can be noted with further clarity in enlarged images for selected time-depths (fig. 23). Comparing these results with the gradiometer survey a close correspondence is noted between the major features in both surveys (fig. 24). However, the radar survey indicates that there is a horizontal displacement of the positive magnetic anomalies in the gradiometer survey. The anomalies noted in the gradiometer survey relate to features which are located c. 2 m to the north of their apparent position in the grids. The two most prominent features in the gradiometer survey are surrounded by a ring of negative magnetic readings. These features correspond to the ring-shaped features noted in the GPR survey just below the plough soil. The GPR transects (fig. 25) suggests that this relates to the existence of a low wall surrounding the sunken featured. Such low walls of subsoil material (chalk) were observed in the excavated areas. 27 Fig. 22 GPR grid 5, horizontal slices for time-depth c. 0-14 ns, calibrated to a depth of c. 0-65 cm below ground, together with simplified interpretation. Due to the direction of the survey the slices are mirrored N-S. North is down. 28 Fig. 23 GPR grid 5, enlarged horizontal slices for selected time-depths, mirrored and rotated to natural orientation. Above time-depth c. 6.7 to 7.8 ns, calibrated to a depth of c. 35-40 cm below ground; below time-depth c. 6.7 to 8.5 ns, calibrated to a depth of c. 35-45 cm below ground. North is up. 29 Fig. 24 Comparison between the interpretation of GPR grid 5 and the gradiometry survey. Note the close correspondence between the major features noted in both surveys, as well as the horizontal displacement of the positive magnetic anomalies in the gradiometer survey. Fig. 25 GPR grid 5, transect 24, at 6.25 m on the N-S axis. Below a simplified interpretation. 30 Figure 26: Location of GPR Grid 6 in relation to gradiometer survey. GPR Grid 6 The survey was located in the Southern part of the fortress area. The southern face of the fortress was not accessible for excavations during previous campaigns, being located in the garden of the existing manor house, Aggersborggård (fig. 27). The symmetry of the fortress design calls for a fourth gate in this area, the South Gate. The survey was complicated by uneven terrain and by the presence of trees and shrubs, which had to be by-passed, causing blank areas in the surveyed grids. As the gradiometer survey show, part of the garden area was also marked by strong magnetic disturbances, which could in some be related visibly to modern disturbances or soil-improving materials. The areas where traces of the rampart were expected, however, show relatively few magnetic disturbances. The survey covers an area of 30 x 20 m and was located in relation to the gradiometer grid (fig. 18). 60 profiles were collected with 50 cm spacing and 5 cm inline trace spacing. The survey was walked from north to south, starting in the north-east corner. The base line of the survey follows the northern edge of the gradiometer grid, and the north-east corner of the survey area coincides with the north-east corner of the gradiometer grid. The cart was started with the centre of the radar above the base line. The horizontal slices (fig. 28) show a number of disturbances in the upper layers, which presumably relate to landscaping and old flowerbeds and paths in the garden. Beginning at a time-depth of c. 18 ns (c. 80 cm) some features can be noted, which may relate to preserved structures along the line of the inner face of the rampart, in an area running E-W along a band approximately following x=10.0 m. In addition, a pair of features is noted at time-depth c. 20 ns in the area x=5.0/y-16.0-20.0, which may correspond to deep-set posts surrounding the gateway area in the excavated East and West gates. 31 Fig 27. The area of the southern front of the fortress rampart, now the manor-house garden, seen towards NW. The supposed location of the south gate has been identified according to GPSmeasurements and is marked out by four ranging rods. Note the slope immediately in front of the supposed outer line of the rampart. Photo S. Sindbæk. 32 33 34 Fig. 28 GPR grid 6, horizontal slices for time-depth c. 0-40 ns, calibrated to a depth of c. 0-200 cm below ground. Due to the direction of the survey the slices are mirrored E-W. North is to the left. Conclusions The following conclusions have been reached with regards to the survey objectives: a) Concentrations of large, positive magnetic anomalies of a size and shape conforming to sunkenfeatured buildings can be identified in the gradiometer survey adjacent to the areas where they are noted in excavations. The interpretation of these features as shallow, round cuttings with a diameter of 2-4 m and cut to a depth of less than 50 cm into the natural is confirmed by the radar survey, in particular GPR grid 4. The radar survey also corroborates the absence of sfb’s in areas where they are not indicated in the gradiometry survey, in particular GPR grid 2 and 3. Further, the radar survey identifies the occasional presence of low walls of subsoil material surrounding some sfb’s – a feature previously noted in excavated areas. The features identified as potential sfb’s occur only in some areas, and thus a clearer picture of the settlement structure is achieved. With few exceptions these features cluster in the central eastern half of the fortress area, and in two rows west of the fortress area. The concentration of sfb’s is not, therefore, continuous. A total of 89 potential sfb’s are identified. Among these three in the NE quadrant of the fortress area show comparatively high magnetic values; based on comparison with features in nearby excavated areas these have been interpreted as possibly being pits containing quantities of Iron Age pottery. Three or more potential sfb’s are located in positions which suggest a relation to the fortress. Based on the integrated survey is may be proposed that c. 80-85 sfb’s exist within the surveyed area, in addition to those previously excavated. b) The presence of leveling layers is suggested by zones with dense clusters of moderately high magnetic anomalies in a band just west of the centre of the fortress area and in a smaller area just 35 c) d) e) f) g) to the NE of the centre of the fortress. However, there are no indications that these layers – if that is what they are – cover deeper lying sfb’s or other large features. No further longhouses from the pre-fortress settlement are identified. However, given the distance between the traverses in the gradiometer survey post build structures may easily have been overlooked. A concentration of possible sfb’s in the SE quadrant of the fortress area, together with linear features in the same area, may suggest the existence of a farm in this area, in addition to those identified within the excavated parts of the settlement. The extent of areas with deep, anthropogenic soils could not be clearly identified. A number of further elements of the fortress plan were identified. The gradiometer survey identified preserved timber structures in the SW section of the rampart. Some large anomalies are found in locations, which may indicate a relation to the fortress: at a road corner, in the middle of a courtyard, or in the gable room of a house. Minor anomalies may relate to wall posts or principal posts, hearths or other features; but the spacing of the traverses implies that it is unlikely that such features can be traced consistently. In one GPR survey area (grid 4) a shallow linear feature may be identified as a trace of the wall line of the building presumably found in this area. In the areas of the north gate features were note, which are consistent with a defensive outwork, similar to traces observed in the fully excavated western gate. The area of the south gate was marked by strong magnetic disturbances, related to modern disturbances or soil-improving materials. Still, some features hint at there being preserved structures of a gateway in this area. Evidence of a conflagration could not be clearly identified. No raised magnetic values were noted in areas near the ramparts or adjacent to the presumed locations of buildings. Areas with wet preservation have not been clearly identified. However, linear features in the gradiometer survey suggests the existence of deeply cut drains adjacent to the N-S axial street in the southern part of the fortress area. Also, the gradiometer survey and radar survey both support the existence of preserved features in the area of the rampart surrounding the south gate of the fortress, now in the manor house garden. More might be learned if features were subjected to coring. The survey was very successful at locating pre-fortress activity. It is possible that the limited length of occupation for the fortress has meant it is relatively less prominent magnetically. It would be desirable to carry out further fieldwork in order to gauge the full extent of the pre-fortress settlement. In particular, the following objectives can be outlined: a) Tests should be made with gradiometer surveys with closer spacing between the traverses, in order to determine if post buildings can be identified in this way. The most promising area to investigate in this way would be the southern central part of the fortress area, where the existing survey indicates a “quiet” background with few magnetic disturbances. b) The areas outside the fortress, to a distance of at least 100 m from the rampart, should be subjected to gradiometer survey similar to that conducted within the fortress in order to identify the limits of the settlement structures. c) As an additional target a survey might also employ appropriate geophysical techniques to seek to identify traces of burials relating to the fortress. In Aggersborg’s “sister” fortresses Fyrkat and Trelleborg burial grounds were found close to the fortress. Areas NW and NE of Aggersborg have only seen cursory archaeological investigation, and could well include burials. 36 d) Further targeted GPR surveys would be valuable in order to define the nature of anomalies identified in the gradiometer survey. In particular, surveys in the SE quadrant of the fortress area might identify how observed features articulate with the deep anthropogenic soils noted in excavations in this area. The combination of pre-fortress settlement, fortress and subsequent accumulation of soil creates a potentially rich stratigraphy. Radar surveys might indicate areas with particularly interesting combinations, which should either see special protection or could be targeted by future research excavations. Acknowledgements The investigations were undertaken in collaboration between the University of York Department of Archaeology, Vesthimmerlands Museum, and with assistance from Aalborg Museum. The editors of this report would like to give their warmest thanks to the following:     Bjarne Nielsen and Simon Kjær Nielsen, Vesthimmerlands Museum, for negotiated agreements with land-owners, provided additional equipment, and personal assistance. Jan E. Slot-Carlsen, Aalborg Museum, who conducted GPS surveys of the grid. Regitze Johnsen, Aggersborggård, for generously allowing access to the private lands surrounded the fortress, including the manor house gardens and flowerbeds. Peter Aherne and Frida E. Norstein, both students at the University of York, who volunteered to assist in the survey. Adresses The University of York Department of Archaeology The King’s Manor York YO1 7EP Tel: +44 (0)1904 323903 Fax: +44 (0)1904 323902 United Kingdom Vesthimmerlands Museum Søndergade 44 9600 Aars Tel: +45 98623577 Hannah Brown: hannahjocelynbrown@talk21.com Helen Goodchild: helen.goodchild@york.ac.uk Søren M. Sindbæk: soren.sindbaek@york.ac.uk 37 The team: From left to right Peter Aherne, Hannah Brown, Søren M. Sindbæk, Helen Goodchild and Frida E. Norstein. References Møller, J.T. 1986: Han Herrederne – et gammelt ørige. In E. Roesdahl (ed) 1986: Aggersborg gennem 1000 år. Fra vikingeborg til slægtsgård. Herning: Poul Kristensens Forlag, 13-28. Nielsen, A. K. 1986: Herregården Aggersborggard. In E. Roesdahl (ed) 1986: Aggersborg gennem 1000 år. Fra vikingeborg til slægtsgård. Herning: Poul Kristensens Forlag, 103-145. Pedersen, A, Roesdahl, E and Sindbæk, SM in press: Aggersborg i Vikingetiden. Højbjerg: Aarhus University Press. Roesdahl, E 1976: Aggersborg. The Viking settlement and fortress. ”Chateau-Gaillard”. Etudes de castellologie medieval VIII, 269-278. Roesdahl, E 1981: Aggersborg in the Viking Age, Proceedings of the Eighth Viking Congress. (H. BekkerNielsen, O. Olsen & P. Foote eds.). Odense, 107-122. Roesdahl, E. (ed) 1986: Aggersborg gennem 1000 år. Fra vikingeborg til slægtsgård. Herning: Poul Kristensens Forlag. Schultz, C.G. 1949: Aggersborg. Vikingelejren ved Limfjorden. Fra Nationalmuseets Arbejdsmark 1949, 91108. Ulriksen, J. 1995: Aggersborgs forsvarsværker. Aarbøger for Nordisk Oldkyndighed og Historie 1993, 181203. 38