Navigation sidebar:



  1. Aims of the model
    1. Purpose of the Strahlenwälle
    2. Theorizing the inner features of the oppidum
    3. Visibility
  2. Basis of reconstruction
    1. Surface model
    2. Ramparts
    3. Houses
  3. Reconstruction phases
    1. Software
    2. Digitizing
    3. Surface model in ArcView
    4. VRML and minimizing
    5. Reconstruction in 3D Studio Max and AutoCAD
      1. Ramparts
      2. Table: Height of ramparts
      3. Houses
      4. Springs and Roads
      5. Animations and stills
    6. Viewpoint Media
  4. Shortcomings and flaws of the model
    1. Ramparts and ditches
    2. Gates
    3. Water
    4. Houses
  5. Using the model
  6. Footnotes
This chapter will deal with the research questions, their relationship to the model, further resources which were necessary to create the model and the technical details of the reconstruction.

Aims of the model

The aims of the Dünsberg model lie in research purposes, even though the model might be later used to give visitors of the site a slight idea of what it would have looked liked in past times.

The questions which were tried to be answered with the model are the following:

Purpose of the Strahlenwälle

One of the initial research questions concerned the use of the Strahlenwälle. Several possible explanations for them exist:

Through walking along these features and seeing them in their assumed original height, I hoped to find out whether they were useful for defensive, traffic guiding or agricultural purposes.

The Strahlenwälle which are enclosing the Grinchesweiher and the Schulborn and thus making them part of the settlement area, can certainly be compared to similar features at other sites. For example the fortification on the Glauberg, displays a triangular annex which encloses a spring (Frey, 1997)1.

Theorizing the inner features of the oppidum

By placing houses on all the known platforms, I tried to illustrate the occupation of the site. The Dünsberg provides a great amount of platforms2, but it is not clear which ones were contemporary, or what kinds of buildings were built on them. The platforms vary largely in size, so it seems possible that they had different uses or contained different numbers of buildings, which might also have varied in size.

Some probable roads, which pass through the gates, have been marked by Reeh. I decided not to include them into the final model, so as not to distract and bias the viewer in his/her perception of the features in the landscape, thus helping to keep an open mind about different ways through and towards the oppidum.

Areas with different settlement density and the extent of the conglomeration should be assessed. Also interesting are the areas with only very little settlement features, even though their real use can only be determined by excavation, observations about the landscape and visibility can still be made. I tried to include the springs into the model, because they also structure the inside of the oppidum and were important features for daily life.

Most of this is a prerequisite for the visibility studies I wanted to undertake.


It was suggested that I would not only use my reconstructed model to answer these questions, but that I would also design a GIS to find out about the visibility from the site and about the obstruction of the line of sight by the fortifications. Unfortunately this could not be carried out, because the 40 m DEM3 of the region could not be obtained in time from the Landesvermessungsamt.

With the construction of a GIS analyses, like the visibility of contemporary sites in the region, could have been undertaken4. Also investigations in the relation between the Dünsberg, the surrounding burial mounds, and the Late La Tène graves would have been possible (see sections Mounds and La Tène). With the DEM in place more about the look of the oppidum in its surroundings could have been learned.

Otherwise the reconstruction is a useful tool to tackle the first two questions. The perception of the viewer approaching the site could give useful hints about the use of the Strahlenwälle, and about the general impact of the site on an observer. Also the view from inside towards other features, like the settlement areas, the gates, the springs or the Strahlenwälle could give hints about how the site might have been structured and whether intervisibility of the three concentric ramparts and other features was important.

Basis of reconstruction

The reconstruction is highly dependent on assumptions and correlates, as there are very little excavation data to support it (see chapter Dünsberg). In the following sections I will elucidate my assumptions and the supporting material for some of the features.

Surface model

The surface model is mainly based on the survey of the FH-Frankfurt in 1988/89 (figure FH). Several flaws have been discovered while digitizing the plans: In the northwest of the hill, south of the Kleinen Dünsberg a gap appears in the isolines. Apparently at the time of the survey a forest plantation had inhibited the area from being surveyed. Nevertheless it is impossible to join up the isolines north and south of this gap the way they were recorded. The way in which I transformed the isolines so that I could connect them followed the map by Dehn (plate 29) (figure Dehn). The banks have been left out, so as not to bias the model.

The platforms were incorporated into the surface model. As sources for the digitizing served the survey by the FH-Frankfurt and the mapping of platforms by Reeh (p. 137). The FH-Frankfurt map supplied height values for some of the platforms, but it was not entirely clear whether these points were only spotheights or an average for the entire platform. None of the platforms which were sketched by Reeh had any height value associated with it. For these platforms, as well as for the FH-Frankfurt platforms without height value, I assigned heights according to the surrounding isolines.

Problems were posed by some platforms north of the Hinteren Eulenkopf. As no isolines were remaining in this area, I had to guess the height values. In retrospect I believe that my heights were chosen too low, as the surface of the model is curving down too much in this area.

Height values for the springs had been desirable, but were unfortunately not available. This could have aided their modelling considerably.


The construction of the ramparts, especially the smaller, more accurate models of the bottommost wall and the Strahlenwall M5, were guided by the excavation findings (Rittershofer, 1999; Rittershofer, 2000; Nickel, 2001). The shape of the ditches has been digitized in AutoCAD R14 from the excavation plans. Calculating the volume from the fill in the ditch and the remaining bank, the height and breadth of the original rampart, including a ramp, were estimated. In the case of the bottommost rampart the reconstruction was further supported by the fact that three “anchors” had been found. The highest of these anchoring beams must have connected to the beam at the front of the wall at a height of 4.4 m6. I assumed this height to be the minimum height of the rampart at this point.

The reconstruction shows a section of the rampart with its three anchoring beams. According to the excavation findings the beams still had the beginnings of branches attached to them. This was also modelled (see Lowestwall_side.jpg and Lowestwall_SW.jpg). The ditch, which I added in the reconstruction, was not observed during the excavation. The justification for adding the ditch lies in the fact that most of the lower bank is accompanied by a ditch (as proven by the excavations in 1906-8, see chapter Dünsberg.) and the model was supposed to show rather a generalization instead of a detailed excavation account. Still the remaining structures were modelled according to the excavation findings, for example the thickness and the spacing of the beams (see Lowestwall_front.jpg).

There are no remains of the palisade on top of the rampart. In this case I followed the reconstructions offered by Motykova (fig. 6, fig. 7) and Drda (fig. 12, fig. 17) for Závist, Maier (fig. 23) for the Heidetränkoppidum and Endert (fig. 20) for Manching. The front of the rampart is faced with wood in my reconstruction. No actual evidence has been found for this, but no evidence for a stone facing was found either. The necessity for a stabilizing blend work is obvious, considering that the rampart consisted of rubble and sand. Only wattle or wood may have been used for this purpose, leaving almost no trace when they were decaying. Parallels for the use of wood as facing for the ramparts are mentioned by Leicht (p. 137-138).

The situation at the Strahlenwall was much more complicated. The excavation drawings were not very clear7 so different interpretations were possible. Rittershofer writes that the rampart was about 5 m wide. Calculating the volume of the infilled ditch and the remaining bank, and using this to estimate the size of the original rampart, I came to the result that a 5 m wide rampart would have been only 60 cm high (see Strahlenwall_low.jpg).

On the excavation plans some dark areas, which might denote the position of beams, can be seen. The length of one of these beams is c. 1.6-1.8 m long, then the feature disappears. Assuming on this basis a width of 1.7 m for the original wall, the resulting height would be 1.9 m, which is much more likely (see Strahlenwall_high.jpg). As before a ramp was attached to the wall8.

The inner structures of the rampart were built according to Rittershofer. Only part of the lowest layer is preserved and I had to chose how the construction would look further up. I decided that the wooden framework (again for stability reasons) would be repeated in close layers, so as to avoid the rubble and sand on the inside being washed out. Other constructions are possible, for example a variant similar to the Ehrang-type but with a wooden front, because no stones were found at the site.

During the excavation four round soil marks were found, their diameter was between 10 to 20 cm and they marked a straight line. The distance between one post and the next was between 65 and 70 cm. The marks were just a few centimeters deep and then disappeared. It is not clear whether these marks are postholes and whether they stand in any relation with the Strahlenwall. I included the findings in the form of a palisade into my smaller, more accurate model of the Strahlenwall, but their sense is highly questionable9.

In the model of the entire oppidum I placed a palisade on top of the Strahlenwälle, even though there is no archaeological evidence to support this.

The only other rampart for which some reliable data exists is the topmost wall (see topwall_back.jpg and topwall_front.jpg on the CD). According to Janke's description a construction of the Preist-Altkönig type seems possible10, but once more a wooden facing has to be postulated due to the lack of stones on site.

The height and width of the wall were estimated, by using the sections drawn by Reeh (p. 57-58), and calculating the volume from the remaining bank only (because no data from the ditch was available). A maximum height of 8.9 m and a width of 6.6 m were estimated, and this time no ramp was added11. The omission of the ramp was according to the reconstruction drawing of Collis (fig. 4), even though it is not clear whether there was really no ramp at the back of the topmost wall.

Once more a palisade was added to the rampart, as well as a ditch, the latter being mentioned by Janke as being 2 m wide.

There is almost no data available for the middling wall and the banks t, u and v.

For the middling rampart I once more used Reeh's (p. 67-68) section drawings. A height of 5.1 m and a width of 3.1 m plus a ramp were estimated. I assumed a similar layout of the structure to the bottommost rampart12. Due to the lack of reliable data no detailed reconstruction was made.

The same accounts for the banks t, u and v, for which no data exists at all. Reeh (p. 62) mentions them to be now about 65 cm high, from which I approximated a height of about 1 m for the original rampart13 to which a palisade was added.


As laid out in section Platforms there are almost no data available for the platforms and the houses. In this case I had to rely on parallel finds from other oppida. Lorenz (p. 40-44) offers reconstructions of houses from Budenbach, from which I chose a half-timbered style14 with wattle and daub. Budenbach even provides evidence that the walls were whitewashed, sometimes even coloured.

For the roof I chose shingles, because wood was a readily available material. Another just as probable kind of roofing would have been thatching. The streams around the Dünsberg would have offered abundant amounts of reeds for this.

Two houses were reconstructed in detail (see House.jpg and Houses.jpg on the CD). For this I chose the measurements of two houses described by Schubert (p. 150). They are both small buildings with six posts. One is 2.79×3.72 m, the other 3.72×4.96 m15 on the ground plan.

For the model of the entire Dünsberg I chose an even smaller house, out of the necessity that this house had to fit even the smallest platform. Based on Schubert's (p. 138) construction schema I, I picked a relation of 6:816, which results in a 1.86×2.48 m basis for the house. The height of the house is 1.80 m without the roof, and 2.78 m with a roof.

Reconstruction phases


Several software packages had to be used for this reconstruction: AutoCAD R14 or AutoCAD Map2 were used for digitizing and extruding. ArcView 3.2 was used to create a TIN from the isolines. Chisel was needed to simplify the surface model, 3D Studio Max 2.5 and 3D Studio Max 4.2 helped to create the model and to make the animations, and the finished model was then exported to the Viewpoint media player. Adobe Photoshop 5.5 and Corel Photopaint 9 were used to create textures, bumpmaps and to crop the mapsheets. Conversion of images to other formats (especially to .eps) were done with the GraphicConverter 4.0.


K.-F. Rittershofer from the RGK17 had provided me with tiff images of the 1:1000 maps of the Dünsberg by the FH-Frankfurt. Even though one of the mapsheets had already been digitized and imported into MapInfo, by one of the employees at the RGK, I decided that it was better to digitize the whole 13 mapsheets in a new AutoCAD file and to include not only the isolines, but also the platforms, streets, springs and ramparts.

First it was necessary to crop the mapsheets using Photoshop, so that no overlap occurred between them. Then I created a grid in AutoCAD, which was set up according to the Gauss-Krueger co-ordinates18, into which I inserted my prepared mapsheets. First I digitized most of the available features on the plan: isolines and platforms (which were lifted to their appropriate height) springs and ramparts. The parapets of the topmost and bottommost rampart were recorded as spotheights, and the same was true for the Strahlenwälle. Only the middling rampart had no height values assigned to it, so I had to digitize it as a line. Also not all of the Strahlenwälle were recorded in the same way: some had exact measurements of the parapet and the foot, others were only recorded through hachures19. In the latter case it was necessary, to record them as lines, too. I also digitized the roads, which were taken by Reeh as being original. They are also included on his plan (see figure Reeh). The second step was to incorporate features from the plans published by Reeh. The scanned images from the publication were made available to me by C. Nickel. After scaling figure 113 and figure 160 by Reeh (p. 137, 200), in which the positions of over 600 platforms are recorded, and trying to match them up with the plan made by the FH-Frankfurt, I found out that they were not identical. The sketch made by Reeh was so distorted that it was impossible to digitize the platforms from it immediately. Trying to rubbersheet the platforms from Reeh's plan to the FH-Frankfurt's plan proved to be futile. Some of the round platforms were distorted to lines. The attempt to rubbersheet just a few of the platforms with more orientation points made the result even worse. Finally the platforms were moved by hand to their approximate position on the already digitized plan. It might be argued that some accuracy was lost through this process, which is certainly the case, but regarding the fact that Reeh's plan had just been a sketch (Reeh, 2001: 137-138), the accuracy of the original drawing is already questionable. Where possible it was attempted to match Reeh's platforms with the ones recorded by the FH-Frankfurt, to reintroduce some of the accuracy.

The height values assigned to these platforms were given according to their proximity to the neighbouring isolines (see also ).

After having digitized the most important features, I tried to import my data into MapInfo, but unfortunately none of the possible methods for importing the files worked, because my AutoCAD file appeared to be corrupt.

Surface model in ArcView

Even though I was unable to import my data into MapInfo it was still possible to import the platforms and isolines into ArcView 3.2. There a TIN20 was created from the features.

It seems now that the platforms recorded by the FH-Frankfurt were more often not modelled as flat surfaces than the platforms recorded by Reeh. The reason for this could be that some of the platforms recorded by the FH-Frankfurt used only 3 to 4 lines per platform, while for recording Reeh's platforms always 5 to 6 lines were necessary.

The TIN was then exported as a VRML 2.0 file, using the 3D Analyst extension to ArcView.

VRML and minimizing

The VRML scene could not be readily imported into 3D Studio Max, as the surface had more than 32,767 vertices, which is the maximum supported. So I had to import the file into Chisel, which can clean up and reduce VRML code21. Unfortunately the available documentation for this software is not very elaborate on behalf of the settings, so that I usually accepted the default. First I let the software clean up the VRML code produced by 3D Analyst, and then removed the smallest edges and the smallest triangles. I also tried unsuccessfully to split up the elevation grid into smaller grid tiles, which were meant to correspond with the 13 initial map tiles. Even though the program would not let me do this, my surface was now—through the triangle and edge reduction—small enough to be imported into 3D Studio Max.

Reconstruction in 3D Studio Max and AutoCAD

The main aim of the further reconstruction work in 3D Studio Max was to incorporate houses and ramparts into the model. I had also intended to include the springs and water reservoirs into it, but experienced some difficulties22.


I realized that the spotheights, with which I had recorded the topmost and bottommost rampart as well as some of the Strahlenwälle, were of little use for the reconstruction, because they recorded the status quo. I had already calculated how high the different ramparts must have been, and the excavation drawings indicate that the front of the original wall lies about 3 m down the hill from the modern parapet.

I decided to digitize the ramparts again in AutoCAD, this time using polylines, which had a sufficient distance from the modern parapet. First I used three layers to record the ramparts:

included banks t-y

contained banks o-s

consisted of banks a-n and A-W

These lines were imported into 3D Studio Max 2.5 and merged with the surface. To keep the model flexible and the filesize small the ramparts had to be constructed as simply as possible. I decided to build a palisade which could be used as the front view showing the posts and the wooden facing, and to have a simple rampart on the back. To construct the palisade I extruded the lines in 3D Studio Max. To give them the appropriate height over the ground it was necessary to conform them to the underlying surface. In 3D Studio Max two possibilities exist to conform objects, one is by creating a compound object, but this would result in the palisade and the surface being one object and I could not have mapped different materials to them easily. The second method, which was finally chosen, is to bind the palisade via a spacewarp to the surface. In this case the surface and the palisade stayed two separate objects23. For the conform modifier to work, it was necessary to flip the normals of the surface, then the palisades would penetrate the surface by the height assigned to the “standoff distance” field. This meant that now banks a-n and Strahlenwälle A-W had the same height, the same accounted to banks w-y and banks t-v. To change this it was necessary to return to AutoCAD. In the file “maps_and_gridPalisades” I transferred banks t-v and Strahlenwälle A-W into separate layers24. Now it was possible to extrude the lines and conform them to the surface with the appropriate height.

To create the ramparts behind the palisades I needed to “attach a thickness” to the lines. First I tried to create loft objects along the lines in 3D Studio Max, but the software would not let me pick the lines as the path for the loft object. I returned to AutoCAD and produced the ramparts in the file “maps_and_gridExtruded”25. In the beginning I assumed that it would be possible to create the ramparts with a ramp already attached to them, and then to conform them in 3D Studio Max to the surface, but this is not possible. There are two ways to use a conform modifier. One is to let the extruded/lofted object penetrate the surface, and the height assigned to the “standoff distance” field would act as an barrier against which the object is pressed. Thus the ramp would not be visible, because for most of the time the vertices denoting it would be pressed against the “standoff distance” barrier. The other way to conform an object to a surface is to select only the bottom vertices of the object, which will then be bound underneath the surface. The top of the object would be where the “standoff distance” tells it to be. This would preserve the form of the ramp, but would distort the middle part of the object by stretching it, while the top of the object would always be on the same height (which is not the same height away from the surface). So neither of the two possibilities suited me, so that I had to build the ramparts like walls, without the ramp26.

Nevertheless the reconstruction tried to account for some differences in thickness, for example bank n was designed not as massive as the other banks of the bottommost rampart, because it is only enclosing the spring of the Schulborn, but not shielding it from the population inside the oppidum. Strahlenwälle E and I were constructed more massive than the other banks around the Grinchesweiher. This is consistent with a drawing by Reeh (fig. 75), who depicts the inner wall E to be more massive than the outer one (D).

Unfortunately during the process of extruding in AutoCAD the outer lines of the ramparts were deformed, so that the ramparts were not matching up with the palisades any more. This caused some unpleasant effects in 3D Studio Max, where the palisade is sometimes sunk into the rampart instead of always fronting it.

Height of: palisade rampart
Topmost rampart (w-y) 9.9 m 8.9 m
Low rampart (t-v) 2.5 m 1 m
Middling rampart 6.6 m 5.1 m
Bottommost rampart 5.9 m 4.4 m
Strahlenwälle 3.4 m 1.7 m
Height of the ramparts and palisades in the model


To keep the model small, the houses were built in a very simple way. They consist of a box, with a wedge as a roof. All the other features were mapped on the surface via materials.

As mentioned in section Houses the houses were meant to fit even the smallest of the platforms, and I decided to built them in a standard size, to minimize work27. The buildings were merely meant to represent occupation of the site, and the schematic representation, which looks unrealistic, helps to express the uncertainty. On the other hand the buildings help to make out the platforms on the surface, so that it is possible for the observer to get an impression of settlement density or settlement dispersion.

After the prototype of the house was finished I copied it and distributed the houses over the hill. In total 724 houses were put on the corresponding platforms. Two platforms could not be populated by houses, because they did not show up on the surface, or were for some reason shown as holes. Several other platforms were not represented as flat surfaces (see section Surfacemodel), so I had to sink the houses into them. Unfortunately some houses appear still to be hovering above the surface, where the surface model did not fit the actual platforms I had digitized28.

Springs and Roads

Springs and cistern had been digitized in AutoCAD, using either polylines, for the features recorded by the FH-Frankfurt, or circles to record the cisterns mentioned by Reeh. Unfortunately 3D Studio Max was unable to extrude or otherwise modify those lines, because they were constructed as two different kinds of primitives(?) and imported as one object. Nevertheless it was possible to conform these features to the surface so that they show up as blue lines in the model.

The roads, which were digitized in AutoCAD (see section Digitizing), were also imported into 3D Studio Max, but I finally decided not to include them into the model, because they might bias the perception and imply roads which might not necessarily have been in use at the time29.

Animations and stills

One long and several short animations were rendered. The long animation (walkthrough.avi) is a walk over the hill, which gives an overview over the model30, and shows how the settlement might have been experienced by an observer. The path for the animation is shown in figure Animation.

The short animations look along the Strahlenwälle and try to determine how a person approaching the oppidum might have experienced them. Through these animations I wanted to find out whether the Strahlenwälle could have directed traffic towards the gates, or whether they were defensive features. Another small animation shows the view from gate 7 at the area outside the oppidum.

The stills show views from banks o-y towards the Strahlenwälle. Through them I wanted to test how far a person could see, and also how the Strahlenwälle were perceived from the inside of the oppidum. Another group of stills explores the smaller more accurate models of the different ramparts (see section Ramparts) and the houses. Some of these stills have only a black background, while others are displayed more naturalistically with sky and grass, depending on whether the view of the model allowed the more naturalistic looking background to be arranged easily. The observer will be inclined to notice that there is a difference in perception of these stills, depending on a naturalistic background or a plain black one.

All the animations and stills can be found on the CD.

Viewpoint Media

The model was designed to be accessible to other researchers as well, especially as not all of its features can be covered by stills and animations. VRML was offering an easily accessible form of displaying the model and making it available to other researchers, who would on their part be able to explore and modify the scene.

Unfortunately it was not possible to export the whole model from 3D Studio Max to VRML. The obvious problem was that the conformed objects were distorted, and had lost their conformed shape.

Therefore it was necessary to export the model as a 3ds-file, which just records the position of the vertices and not the conform modifiers themselves31. Again it was not possible to export the whole of the model at once. Instead parts of it had to be exported as different files, which then had to be assembled correctly in a new max-file. This process introduced further inaccuracies into the model, especially the Strahlenwälle had to be positioned by hand, so that their precision in height and position is possibly not faithful to the original model32. Still it proved to be impossible to export the model without distortions into VRML. Finally the decision was taken to export it to another file format provided by Viewpoint Media33, which would export the model perfectly. The exporter creates three files, among them an html-file, which can be opened to view the image.

Shortcomings and flaws of the model

It was already mentioned that there are flaws in this model. I want to use this section to remark in detail what could have been done in a better way and what is still missing.

Ramparts and ditches

It would have been desirable to have a more accurate picture of the ramparts. In particular ramps and ditches should have been included into the model. As mentioned before it was not possible to model these features by simple conforming, but I could have included them into the surface model in the same way, as I included the platforms. This means that I should have digitized them in AutoCAD and together with the platforms and the isolines imported them into ArcView, where the TIN could have modelled these features.

Having the ramps and ditches included into the model, it would have been possible to assess how obstructing they really were. For example: in the current model it seems as if it was possible to walk through the gap between Strahlenwällen L and M. but indeed there must have been the ramp of Strahlenwall L and the ditch of Strahlenwall M obstructing this passage34.

It is also not clear for all the Strahlenwälle which side they were facing, and whether there was a ditch in font of them at all. It is also not proven that all of the Strahlenwälle were constructed in the same way. They might have varied in height and appearance considerably.

On the other hand for the ramparts a-y it is attested that they were of different heights and width at different points of the fortification (Reeh, 2001). It is possible that their construction varied in different places as well35. At least the data for different heights had been available through the section drawings by Reeh (see section Ramparts).


None of the gates were modelled even though I had collected material from correlates. First it is not really necessary to see the gate as a structure for answering any of the questions I had posed at the beginning of my research work, and second none of the gates has been excavated, so my reconstruction had to be based on mere assumptions regarding the construction of the gates. Still it would be interesting to see the difference a simple gate or a whole gate tower would make to an entrance. Especially gate towers would make the gates much more prominent and outstanding features in the fortification. They must have been widely visible as well, and thus making an impression on visitors of the oppidum.


I had planned to model the springs and water reservoirs as outstanding features, so that their impact on the appearance of the inner features of the oppidum could be assessed. Unfortunately I was not able to achieve this (see section Springs and Roads). The lack of these features is especially apparent at the Grinchesweiher, where the absence of the basin leaves the viewer orientationless36. Modelling the actual basins and showing all the springs in the model would certainly have made a difference in the perception of the site, and would maybe show the relation between Strahlenwällen and water reservoirs more clearly.


The uniformity of the houses on the hill has a touch of the surreal. It would have been advantageous to have different sizes of houses, and maybe different types of buildings as well, for example barns, stables and Grubenhäuser, which show different activities. But for this is would be necessary to have more excavation data first, otherwise the model would seem to be a faithful reconstruction, even though we actually know nothing about the buildings on the hill.

Using the model

To test the model animations and stills were prepared. Still it would be desirable to be able to interact even more with the model. I had the impression that it was not possible to capture the whole site through the means of recorded walkthroughs and panoramas. The animations often seem jaggy37 and it is not possible to direct the view to whichever direction is interesting at the moment.

This situation is changed if the user has access to the original model and to 3D Studio Max, new paths and cameras can be set up to investigate questions and to generate new impressions of the site.

Also the perception changes greatly with the focal length of the camera used to view the scene. Experimenting by changing the focal lense for the same scene can focus or broaden the view on objects and their surroundings.

For experiencing the model even more freely the viewpoint media representation is a step in the right direction. It is possible for researchers to explore the model on a low-cost basis, maybe download it from the internet, and manipulate the views on their browser 38. Refinements and changes to the model could open new ways of viewing the site in the landscape, and by adding and hiding some of the ramparts previous stages of construction could be visualized and tested. Also changes in colour could help to experience the model differently. I had the impressions that I could view the features in the model more clearly when freezing most of the objects in 3D Studio Max (they become grey then) and only the ramparts were displayed in green.

I think that the model is very complex and it would take a long time to explore it in its entirety in order to understand the specific layout of its features. Still this would be a worthwhile exercise, because the model gives us insights which cannot be experienced otherwise.

This leads us straight to the conclusion.

1 The site of the Glauberg is of an earlier date, but I still think that this is a valid parallel.

2 The 725 platforms on the Dünsberg, stand in no relation to the neighbouring Heidetränkoppidum, which is with 130 ha (Maier, 1980) larger than the Dünsberg, but contains only about 161 Platforms (Maier, 1985).

3 Digital Elevation Model

4 Several sites are visible from the watchtower installed on the top of the Dünsberg, but the view from the tower is not the same a person on the ground would have had.

5 See images Lowestwall_... and Strahlenwall_... in folder Small_models on the CD.

6 This is according to my calculations and the measurements I took from the excavation drawings.

7 The soil on the hill is not very suitable for detecting changes in colour and other features. The marks left by roots of trees and animal dens make the identification of archaeological features even harder.

8 The volume of the ramp was included in the calculations for the height and width of the rampart.

9 See all images Strahlenwall_... in folder Small_models on the CD.

10 The reconstruction drawing by Rybova (fig.11) for Hradiste by Stradonice, also conforms to Janke's description of the inner structures, only the stone facing any maybe the ramp would have to be changed.

11 The existence of a ramp would reduce the height of the rampart.

12 The lack of easily observable inner features and the layered appearance of the sediments inside the rampart made me think that they could have been of similar design.

13 This is a rather arbitrary estimate, as there were no section drawings to guide me.

14 The other possibility had been simple walls made of wattle and daub.

15 This corresponds to his fig. 4b and fig. 4c.

16 Schubert proposes some common construction principles for the buildings at Manching. The basis is a measure of length based on feet. One foot has a length of 30.9 to 31 cm. Further design elements are based on the Pythagorean theorem (especially the numbers 3, 4 and 5) or on compass patterns. These were used to construct right angles and standardized proportions.

17 Römisch Germanische Kommission des Deutschen Archäologischen Instituts

18 The Gauss-Krueger co-ordinate system is the German standard grid. It is based on a transverse mercator projection, using the spheroid bessel. The false easting lies at 3500000 m, the false northing at 0 m.

19 It should be mentioned at this point that the plans used for digitizing and the plan FH are not identical. The plans which I used for digitizing are much more detailed in respect of the ramparts and the platforms.

20 Triangular Irregular Network

21 A free trial version of this program is available from .

22 This is discussed in more detail in section Springs and Roads.

23 The conform modifier tries to conform the vertices of one object to the surface of another object. The more vertices both of the the objects have the more accurate the fit will be, as the modifier has more points to work with: only the vertices are adjusted the area in between will be flat.

24 When importing the lines into 3D Studio Max, the software believed the lines to be a single object, when they were on the same layer.

25 To loft an object in 3D Studio Max means the same process as extruding an object in AutoCAD.

26 It was also impossible to generate a ditch in this manner.

27 We also know almost nothing about the size of the houses on the Dünsberg and whether several of them could be expected on one platform.

28 I had used the platforms as an aid for positioning the houses. Checking each of the 724 houses in detail in the perspective view would have been to time intensive.

29 Mapping the roads onto the surface would have been an easy task, by using a conform modifier.

30 Path “animpath” and camera “animcamera” are included in the 3D Studio Max file on the CD.

31 The same result could be achieved by exporting the model to AutoCAD as a dwg-file.

32 Most of the ramparts (a-y) were inserted at the same place as the platforms. The platforms on the other hand can be positioned exactly on the surface model. The palisades could be placed along the ramparts, but might not have the exact height of the original model.

33 The Viewpoint Media Exporter for 3D Studio Max 3.1 and 4, as well as the Viewpoint Media Player can be obtained free of charge for non commercial uses from .

34 Another possibility would be that there was no ditch in front of Strahlenwall M in this part.

35 Leicht (p. 137-138) mentions several sites where this is the case.

36 At least this is my impression.

37 Something which could have been solved with higher frame rates.

38 For a theoretical discussion see for example Gillings.