27 April, 2015
This week On the Reading Rest I have a small report in Danish from an excavation in Jutland. This kind of report is called a developer’s report. It is a short description of an excavation partly paid for by the developer because prehistoric remains would have been damaged by the planned construction work. In this case the archaeologists responsible for the excavation and the report have chosen among other things to present a house from the period 400 to 550 CE, that is the Early Pre Carolingian Iron Age. The site is called ‘Domdalsgård’ or ‘Dooms dale Farm’ and although this direct translation is probably not correct I shall comment upon the house calling it Doomsdale.
Terkildsen, Kamilla Fiedler and Boddum, Sanne 2011.Domdalsgård: gravplads fra ældre romersk jernalder, landsby fra ældre og yngre germansk jernalder (Domdalsgård: Cemetery from Early Roman Iron Age, Village from Early and Late Germanic Iron Age). Bygherrerapport nr. 61. Viborg Museum. ISBN 978-87-92778-11-6 http://www.viborgstiftsmuseum.dk/data/imagemanager/pdf/61_domdalsgaard.pdf
On today’s farm the dwelling houses are small in comparison with dairy barns and manure basins because milk production at an industrious scale shapes the farm and its farmland, as well as and the agricultural firm called I/S Søndergaard that runs the estate. The relation employee:milk-cow is c. 1:75. It’s modern market-sensitive Denmark.
In the 5th century when Doomsdale was built, times were different and for the first time in centuries the byre or ‘the dairy barn’, in those days part of a main house, stated to become shorter than the dwelling quarters. This change was pointed out OtRR: 16 April, 2012, and exemplified by the St Darum village. In this regulated village the change was more marked and probably the result of a change in the agricultural economy favouring crops over cattle. The byres shrank and the dwelling quarters grew – the cattle became fewer and there were more people in the household. The number of heads would have balanced each other.
At Doomsdale which might be a somewhat earlier farm the proportions dwelling:byre are a little more balanced, but Doomsdale is still a typical EPCIA building because of its formalized planning, but also a little odd.
Usually in the PCIA the proportions between the lengths of the rooms were expressed in even numbers of feet. The central main room with the hearth was the largest room. The private quarters above the heath room were larger but divided into the gable room and the chamber. In most PCIA farms the outer room was not smaller than the chamber and frequently larger probably as a reflection of the number of farm hands in the household. In Doomsdale this is not the case although a minimal change from the proportions 8-12-17-11, 5 to 8-12-17-12 would have been a natural solution if proportions mattered as much as they usually did.
Other proportional solution such as:
A four room residence with lovely chambers and a capacious central room featuring 8-12-18-10 ft. proportions …
The classical balanced 48 ft. room solution: 8-12-16-12 ft., often favoured on better farms in Southern Jutland …
simply didn’t fit Doomsdale.
The size of the gable room is a consequence of the roof construction, but the three other rooms are there to express a social and practical balance. When we visit one of the evenly proportioned farm we understand that there are quite a number of farmhand and that the private quarters in the upper end of the house are capacious. The central room, therefore, may be a bit crowded, but that was probably just as it ought to be on a busy modern farm. Visiting Doomsdale, on the other hand, would give the impression that the farm hands were few or their living quarters crammed, the central room was large and the chambers capacious. This would have reminded one of a Late Roman Iron Age situation being an echo of the time when cattle were more important, and large farm owners larger.
In the byre part of the house there are no simple proportions other than the ones that were dependent upon the construction of the house. This is quite a modern trait compared to Southern Jutland where the post setting in this part of the house was functional, mirroring the construction of the cow’s compartments, dividing this part of the main farm house into byre and barn. At Doomsdale the post pairs were probably spread out evenly east of the entrance room in order to support the roof in a rational way. This solution probably involved measuring, but also small factions of the foot which cannot be measured on an excavation plan. The general impression of the main house would thus have been a mixture of rooms mirroring the social order of the household as well as rational and practical solutions to house construction. Compared to the RIA the byre part of the house would have lost a little of its ideologically inspired Early Iron Age design.
As mentioned, the technical construction of the building shows in the distance between the short end of the house and the nearest post pair – eight feet. In reality, however, this distance is linked to a relation between the wall plate in the corner of the house and the first roof-supporting trestle. Doomsdale was built according to a proportional system which starts with a rule of the thumb stating that if the distance from the short end to the first trestle is equal to 2 and the distance from the wall plate to the side ridge equals 1, then all the roof angles in the house will fit a thatch roof, if the height of the post in the trestle is 4, given that 2 equals the wall height. In this case the central ridge will be 5 units above the ground. Moreover, owing to the rounded corners, and supported by the first trestle and the side ridges, the rafters from the short end will meet with the rafters from the side roofs in one and the same point at the central top ridge. This will not happen if the wall height isn’t equal to the distance from the short end to the first trestle. Consequently, wall height at Doomsdale equals 8 feet, i.e. 2×4 feet. The total height of the house was thus 4×5 = 20 feet, i.e. 0.312×20 = 6.24 meter.
In the ever more open Early Iron Age landscape finding the necessary timber in order to build a house gets more and more difficult and there is thus a great point in being able beforehand to figure out the dimensions needed to build the house. In tandem it becomes all the more important to build large main houses as a symbol of the status of the owner. Social competition is in other words important and little by little architecture turns away from the functionalism of the Pre Roman Iron Age to the formality and rational technique of the Early Pre Carolingian Iron Age. There is a short revival of the functional ideal in the beginning of the Late Carolingian Iron Age, but Harold Bluetooth, who else, sees to it that formalism gets back in business (cf. OtRR 11 August, 2014).
9 February, 2015
Once again this week On the Reading Rest I have a report on the excavation of a small Iron Age village. Once again in the easternmost part of Uppsala, Old Uppsala a suburb incorporated in of today’s city. Once again the primary reason is the possibility to define new local time frames, i.e. new historical situations. The secondary, overarching reason, has to do with research history and the way the present, which holds the power over the historical records, shapes the past making it more or less interesting – that is more or less historical in a contextual sense.
When the Greater Stockholm area went through an unprecedented expansion 1965 to 1975 archaeology was generally confined to excavating cemeteries and visible monuments, i.e. graves. Emphasizing traditional heritage categories, and the same procedure as usual, during a vogue of pressing development, rather than focusing on developing the concept of ancient monuments, is typical of metropolitan areas, especially that of the Capital. In these areas, in addition to the pressing needs, the conservative influence of the national boards of antiquities, most often situated in the Capital, is manifest simply because these categories were once pointed out by the boards as ‘our’ significant monuments of the past. During a large-scale modernistic urban development project, history simply isn’t not called upon to rock the boat.
Later on, in a less modernistic period, when provincial towns expanded, local autonomy played a greater role in tandem with a central wish to monitor local administration and see to it that they didn’t overlook any ancient monuments on their road to modernity. One of the effects of this situation was the introduction of new categories expanding the concept of ancient monuments. Adding invisible settlement remains in arable land to the categories was a typical example of development and the reason why expanding towns such as Västerås, Helsingborg and Uppsala can boast so many IA farms in their outskirts. Sometimes small is less ugly than large.
Göthberg, Hans; Frölund, Per and Fagerlund, Dan. 2014. Gamla Uppsala – åter till Berget. Om undersökningen av en förtätad bosättning från äldre järnålder med begravningar från äldre bronsålder till romersk järnålder. Fornlämning 614:1, Uppsala Gamla Uppsala 21:52 Uppland. [With contributions by] Thomas Bartholin, Ylva Bäckström, Stefan Gustafsson & Emma Sjöling [Upplandsmuseets rapporter 2014:16]. Uppsala, Upplandsmuseet. Acronym: GUB.
There is a summary in English and the report can be found at:
Berget – the mountain in English – is a small rocky hillock. Once in the beginning of the Early Bronze Age (EBA) it was surrounded by water, but gradually, owing to the isostatic uplift, by arable land sloping north and northeast facing a narrow sound that was turned into bay. Today this water has become the small river Samnan. In the IA the farms at Berget would have benefitted from the low-lying meadows north of the village.
The history of the site starts in the BA when rituals in connection with funerals and burials took place by the blocks on the hillock. Twenty nine context were excavated. Some may reasonably be called graves others are less evident. Now and again from c. 1400 BCE until the end of the 300s CE a few people were buried on the hillock and a number of contexts have probably disappeared during Iron-Age and modern occupations (cf. Göthberg & Frölund GUB:51ff.). Being already a site in the human landscape the first house was built here in the very beginning of the EIA (cf. Göthberg GUB:109:Hus 33, fig. 78). As expected this was a One-Generation House, never rebuilt and thus standing c. 30 years without major repairments before the house was demolished and people moved somewhere else.
If we map the earliest 14C-dates we see a dispersed settlement structure with One-Generation Houses in the eastern part of the settlement area. Circa 1950-1900 bp (i.e. in the beginning of the Common Era) the farms move westwards and started to cluster. If we look at the latest 14C-dates, circa 1650-1500 bp (the century around 500 CE), we see a more clustered settlement in the western part of the settlement close to what will eventually become the situation of the historical farm. In the 5th c. some farms have already been pulled down, but the surviving ones, on the crowded plots in the west, respect the larger deserted plots in the east such as that of Farm D, inasmuch as they do not spread out to occupy them.
Between the long sporadic beginning of the settlement and the end phase there lies a dynamic period with a massive and varied cluster of 14C-dates. With the Bcal program (1), this beginning and end can be modelled quite well and it would seem that the central dynamic period of occupation commences c. 180 and ends c. 400 CE. Perhaps the time limits are less wide. This period is not just an intensive settlement period, it is also the period characterized by tar production. Cause and effect are impossible to judge, but the correlation adds to the dynamics of the village and indeed to seeing these two centuries as an era in its settlement history.
Göthberg & Frölund have analysed the farms and summarized the whole history of the settlement and its farms (GUB:251ff.). Owing to the numerous and strategically chosen 14C-dates a more precise chronology of the individual farms may nevertheless be obtained by means of Bayesian statistics. In the following Farm D will be taken as an example (cf. GUB:259-61).
All these numbers, plans and diagrams boil down to a small history of farm and property. Farm D occupied a site in close spatial relation to an ancient place, a historical site, and developed into one of the oldest farms when Berget became a village with permanent farms in the second century CE. In the 200 dynamic years in the third and fourth c. CE, the main house at Farm D was the largest in the village and in the human landscape its situation was quite prominent. We would perhaps have expected that this farm should be the last to be demolished, but that was not the case. On the contrary it was pulled-down before the smaller neighbouring farms and the farm was succeeded by a barn, erected to claim the property. This indicates that the owners of one of the economically more important units chose to move somewhere else without giving up their land. Since we have never excavated a large solitary farm established in the 5th c. CE we don’t know where the land owners lived, but in all probability it wasn’t at Berget. This means that land could be owned by people who didn’t live on the land. The original idea, emanating in the beginning of the EIA, stating that a family could settle for a generation or two sustaining itself on the land surrounding it, was replaced by an ownership that wasn’t based on the presence of a household. Ownership to a property became more abstract. Since the small farms couldn’t occupy the large abandoned plot of Farm D, we may suggest that they were in some way or other dependant on the non-present landowner although they may of course have been autonomous land owners themselves. Already in the 5th c. it would seem that there was an organisation of plots that forced farmer to build their farms on a plot that reflected the size of their property. This didn’t create a formal pattern at Berget, but plots were nevertheless respected. The barn on the plot of Farm D, i.e. House 20, was the first example of a formal definition of a plot – not its boundaries – but its centre representing presence and non-presence at the same time. Berget, Farm D is a prime example of the early centuries in the development of landownership, which eventually allowed land owners (probably the large ones) to live on one property and control an estate composed of several unoccupied properties – claiming their right with reference to a concept of ownership that wasn’t based upon land use.
11 August, 2014
This week On the Reading Rest I have an article about the large 10th c. royal monument in Jelling, Denmark and its enclosure. Speaking of its dimensions and the fact that the palisade surrounds a number of religious monuments there is something almost Pharaonic about it. But instead of dressed stones Harold used wattle and daub, wooden structures, earthen mounds and natural boulders to construct his monument.
Dengsø Jessen et al. 2014. Mads Dengsø Jessen, Mads Kähler Holst, Charlotta Lindblom,Niels Bonde and Anne Pedersen. A Palisade Fit for a King: Ideal Architecture in King Harold Bluetooth’s Jelling. Norwegian Archaeological Review, 2014 Vol. 47, No. 1, 42–64.
DJal sum up the excavations at Jelling and put the site into a larger chronological, archaeological and political perspective. The short-lived but gigantic enclosure, covering c. 12.5 ha, was built in the 960s and early 70s. It may be that (1) building the enclosure and the houses within it, as well as (2) the opening of King Gorm’s mound and grave, (3) his reburial in the new church tagged by (4) the erection of the large rune stone advertising Harold’s good work and finally, perhaps (5) the relocation(?) to Jelling of the c. 20 year old stone commissioned in memory of Gorm’s queen Thyra, Harold’s mother, who died before Gorm – all this took place in the decade around 965 CE. The enclosed Jelling monuments were short-lived, but grand enough to match Harold – ‘the continental-style ruler’ (DJal:59). Jelling’s Pfalz-character or function as a royal assembly place is not a farfetched interpretation and its stress on formalism and geometrical planning catches the eye. Moreover, the fact that the Jelling monument is older than Harold’s rigidly planned circular ring forts, suggests that Jelling was the start of something great – significant and sufficiently provocative to be short-lived and destroyed in an arson attack.
Bandholm (2013) (1) has shown that the lozenge-shaped palisade in Jelling is a rhombus with sides c. 360 m long comprising four large identical Pythagorean triangles with sides equaling 3, 4 and 5 units. Five units equal 360m and the four right angels are situated in the center of the rhombus. For what it is worth, there are geometrical coincidences with the great pyramid at Giza or vice versa – at least in principle. With a ruler, a setsquare and a pair of compasses this quadrilateral can easily be constructed on the drawing board. In principle, therefore, it can also be illustrated on a computer screen.
DJal shows that in the Jelling topography, materializing the theoretical ideal was not straightforward and they point out that the observed measures do not always fit the theoretical model – but then again why should they when close enough in Jelling is most impressive.
Further excavations may give us more clarity in these matters and DJal, representing the Jelling Project, are cautious in their discussions, as indeed they should be. On the Reading Rest is less committed and free to wonder how the architect(s) and/or surveyors went about plotting out the enclosure – not least its corners. Since ideologically inspired formalism in the construction of past-and-present as lavish monuments, is still with us, one may argue that Jelling was once a significant manifestation. Consequently, ruining the monument by setting fire to it reflected its importance and appalling, in effect, significant character. Since it is not self-evident how the monument was constructed, its construction itself might be outlandish.
In Jelling, the main problem when it comes to setting out the rhombus is contextual, caused by the monuments that occupied the site before the construction of the palisade. There are four musts to the initial construction work:
1) The center of the large mound predating the enclosure, must also be the centre of the rhombus. Since the centre is situated on top of a mound above the surroundings it is difficult to measure directly from it.
2) This centre must be aligned to an axis that is also the axis of the ship setting, which predates the mound. It is not given that the endpoints of the ship setting are aligned to the actual centre of the mound. Centre and endpoints must in other words be aligned in a best fit.
3) The distance between the centre and the endpoints of the axis must be defined as a fixed number of feet. In some way or other the precise distance from the center to a given point on the axis must be measured, preferably with some sort of triangulation.
4) North and south of the mound this axis must be level. This is because the distance to the points where the axis intercepts the palisade of the enclosure, that is, the northern and southern sides of the rhombus, must be measured directly on the ground, that is, by some sort of trilateration.
The mound is situated more or less at the centre of a ship setting as if the monument was a ship grave with a central chamber more or less where the mast would have stood. In this way, the monument is similar to ship graves such as Oseberg or Gokstad. The chamber represents the mast and somewhere above the chamber, the centre of the Central axis should be situated. The ship may have been ‘sailing’ towards the NNE or the SSW diverging c. 22.5° from North towards the East. Together the ship setting and the mound make up King Gorm’s grave monument. Gorm and his Queen Thyra, Harold’s parents, were to be commemorated in the enclosure, while at the same time Harold made a point of displaying his power, his Christian inclination and indeed his master builder capacity. The latter was a crucial part of his royal image and to less lavish-minded citizens, as well as the odd modern researcher, it borders on obsession (cf. OtRR 15 October, 2012). Since Harold wanted to reshape Jelling, there was a point in letting Gorm’s mound in the ship setting become the centre – the original root of the monument – after Harold had excavated it and moved his father to Harold’s new the Church where once a hall stood (2). Acculturating Pagan and Christian traits it was probably respectful to let the enclosure of old and new monuments proceed from the axis and centre of the first monuments. The cultural trait that would eventually combine the old and the new ideology was a hetero-normative elementary family – the married couple and their successful son. Talking to his architects, we can imagine that Harold would have listed the following operative paragraphs having said
A) Protects the old monument and incorporate it in the new as central historical objects, that is, heritage and origin.
B) Encloses them in the new monument thus exemplifying how the past corresponds to the present.
C) Builds a cenotaph, a mound commemorating his parents, his mother south of his father’s mound.
D) Builds a church between the mounds where the hall stood and moves the remains of his parents into a grave in the church.
E) Moves the old rune stone commemorating his mother to Jelling.
F) Erects a rune stone commemorating himself next to the church and his mother’s rune stone.
And the architect team came back with the idea of the lozenge-shaped enclosure. The head architect pointed to Bandholm’s theoretical construction, or a similar formulation of it, and its beauty, as a satisfying model. The King gave his consent, but pointed out once again that they were not to tamper with Gorm’s mound. The architect, reassuring and convincing as always, answered that there was no need to do so, because he had another triangle up his sleeve if needs be. The mound would not be damaged.
OFF THE TEAM went to establish the axis of the ship setting as the historical roots of the monument constructing the new monument on the basis of the axis of the old one – a theoretically simple, but practically difficult task. They started by establishing the axis as a line-of-sight between two points on the flat top of the mound. They define this line with two gormas and then they aligned arbitrary points to the North and the South and marked out the Central point, Cp.
Having created the centre of the mound as well as the axis, they defined the fixed the points N’ and N as well as their distance to the Central point (Cp) by means of measurement (trilateration) and triangulation. Then they decide upon the length of the central axis making it longer than the ship setting.
Since this length was crucial to the construction of the outline of the palisade they sought a specific length long enough to enclose the ship setting and definable as a specific number of modules befitting their geometrical model. In some way or other they must have studied geometry and in all probability they had a book where they could look up different geometrical solutions to architectural problems. Rough measurement in the field and their model would tell them that if the sides of their rhombus were c. 1152 feet, then the palisade would be large enough to enclose the ship setting. Having done their calculations in advance they let the endpoints of the central axis be 1237 feet apart. This meant that the distance N’ – N must became 309 ¼ ft. They constructed S’ and S in the same way as they constructed N’ and N. In points N and S the Central axis intercepted the palisade line.
In our measuring system 309 ¼ ft is slightly less than 100 m and that is a distance possible to measure directly. Nevertheless, creating the Central axis and the Cp is probably difficult and it is a great advantage if the ground is level around the mound – not least since points N’ and S’ must be level. To solve these problems they would probably have used some sort of chorobates or large scales or a device with water or whatever (2).
With their 5 points on the central axis they bring in a new triangle (sides 1 – 2 – sqrt 5) and define the Northwest corner (NW) of the enclosure along a line that is perpendicular to the central line in N’.
Having defined the NW corner they can make two lines of sight – the northern side of the rhombus, NW → NE, and its NW → SE diagonal.
The Northwest point, NW, is situated 618½ feet from N’, that is twice the distance between N and N’. Theoretically the hypotenuse of this triangle is sqrt(12 + 22), that is, the square root of 5. The present case is a practical application of this quasi-Pythagorean triangle:
sqrt (309.252 + 618.502) = sqrt 691,5040 … or ≈ 691,5 ft
The difference between 619,5040 and 691,5 ft is c. 0.1 cm. The architect must have had some practical knowledge about the proportions of the 1 – 2 – sqrt 5 triangle and understood that in some situations it could be useful.
Creating the sixth point, NW, implies in the following: The 619½ ft between NW and N equals 3 in a Pythagorean triangle with sides: 3 – 4 – 5 and its right angle in the Centre on the mound (Cp). Both NW and N are situated on the northern line of the enclosure. A line-of-sight from NW over N will reach the point NE 461 ft behind N (619,5/3 * 2 = 461) because NW-NE equals 5 in the Pythagorean tringle. This is a long distance, difficult to measure directly. The architect, therefore, anchored the 1 – 2 – sqrt 5 triangle in the points N’ and N (see above) and that enabled them directly to measure a point, NE’, situated close to N’, but also between Cp and NE. This means that a sight line from NW over N and from Cp over NE’ will meet in NE – the northeastern corner of the enclosure. In this way the architect has created the Pythagorean 3 – 4 – 5 triangle with the points NW-Cp -NE. This is the triangle which Bandholm (2012) refers to.
In this construction, the distances that must be measured directly are relatively short, c. 97 and 48 m respectively. They combine four level points: NW’, N’, N and NE’.
To complete the rhombus the corresponding level points and triangles SE, S’, S and SW’ must be constructed and that turns out to be virtually impossible if the South mound had already been built. The mound is in the way for those who want to measure the distance between S’ and S
As long as the corners SW and SE have not been excavated it is difficult to know what’s right or wrong when it comes to these points – and even after an excavation it may be impossible to know. Archaeologically speaking, nevertheless, it seems that the South mound had not yet been built when the palisade was laid out (DJal:62:Fig. 11). The archaeological context makes the present model possible, but there are many ways of construction the rhombus. In the 10th c. the best models ought to have been inspired by Ockham’s razor inasmuch as they should involve little direct measurement between few level points and sight lines over points that are relatively far apart.
There is another consequence of the model that might be interesting. The hall building, which ante-dated Gorm and Thyra’s(?) grave church, could be seen as parallel to the northern or southern side of the palisade. The alignment is not perfect, but possible. If so, then the house was probably built after the lozenge was laid out, and after the North mound was built.
This would mean that some time, say 965 CE, the enclosure was planned and the hall south of the North mound erected and the palisade about to be built. The whole monument was thus a non-Christian environment – a classical scene: the hall next to the mound, in the ship setting enclosed by a palisade. Then – and that must have been quite suddenly comes Harold’s conversion to Christianity and the need to make the monument a Christian one with parents seemingly moved from their mounds to the new grave church built upon the hall – another classical scene!
Or the hall, a more or less East-West orientated building stood there before the enclosure was contemplated and Harold wanted its axis to be reflected in the new monument. In that case this turned out to be only not quite possible because the surveyors could not measure arbitrary angels.
Or the present geometrical model is wrong!
WHEN THE SOUTH mound was built is must have stood out as an echo of the past – a monuments from the 950s when Queen Thyra died and thus older than Harold’s monumental enclosure. If the mound was accepted as old it would have made the layout and construction of the enclosure enigmatic and perhaps wonderful, because it would have seemed that Harold’s surveyors in the late 960s could see through mounds.
It stands to reason that any charismatic ruler opposed to his father and his father’s quest for systematic and enigmatic formalism as well as show-off dressed up as Christianity, such as Svein Forkbeard, would burn down the palisade.
How many of the above boxes were ticked by the architect(s?)? four – perhaps all six:
√ A) They protected the old monument and incorporate it in the new as central historical objects, that is, heritage and origin (Perhaps they started to do this already when the hall under the church was built).
√ B) They enclosed them in the new monument as an example of how the past corresponds to the present.
? √ C) They built a cenotaph, a mound commemorating Harold’s parents, his mother south of his father’s mound.
√ D) They built a church between the mounds where the hall stood and move the remains of my parents into a grave in the (grave-)church.
?√ E) They moved the old rune stone commemorating Harold’s mother to Jelling.
√ F) They erected the rune stone commemorating Harold next to the church and his mother’s rune stone.
DJal’s article and the lozenge tag a new Jelling research and the role of Harold Bluetooth, the successful son, in the transformation of South Scandinavia. His antagonists may have burn down Jelling and made it a monument securely anchored in the past, but in so doing the future they embarked on nevertheless had to fit the place he staked out as his Denmark rather than the indefinite ‘Denmark’ mentioned e.g. by Alfred the Great. Harold’s ring forts lack a historical dimension in their construction, being thoroughly modern. Jelling, on the other hand, is a shrewd example of the present as rooted in and incorporating the past as well as of royalty rooted in farmsteads and manors (cf. DJal:59) (3). A contemporary of Ottonian Renaissance Harold’s monarchy boasted rural origins and systemic modernity partly resting on eternal geometrical principles.
(1)Bandholm, N., 2012. The Arcane Eshøj Ell: applications from royal Egypt to royal Jelling. Acta Archaeologica, vol. 83: 275–86.
(2) The scale in the illustration is inspired by and article by Tom M Apostol:
Apostol, Tom M. 2004. The Tunnel of Samos. Engenering and Science 2004:1 pp. 30-40
(3) For a detailed argumentation see Holst et al. 2013. Mads Kähler Holst, Mads Dengsø Jessen, Sten Wulff Andersen, and Anne Pedersen, A., 2013. The Late Viking-Age Royal Constructions at Jelling, Central Jutland, Denmark: recent investigations and a suggestion for an interpretative revision. Praehistorische Zeitschrift, vol. 87:2: 474–504.