Digital Medievalist 9 (2013). ISSN: 1715-0736.
© Tarrin Wills, 2014. Creative Commons Attribution-NonCommercial licence

Semantic linking of the Pre-Christian Religions of the North

[ Skip to Abstract | Return to Top ]

Peer-Reviewed Article

Accepting Editor: Malte Rehbein, Universität Passau, Passau.
Recommending Reader: Øyvind Eide, Universität Passau, Passau.
Recommending Reader: Arianna Ciula, University of Roehampton, London.
Received: July 4, 2013
Revised: April 3, 2014
Published: June 17, 2014

[ Skip to Navigation | Return to Colophon ]

Abstract

The Sources strand of the Pre-Christian Religions of the North project aims to bring together and document the primary evidence for the early religions of Northern Europe. In order to be useful for researchers these sources must include metadata or semantic links that can connect them to concepts relevant to the study of religions. This paper outlines ways in which some very different source types, including textual sources, place-names, pictures and figurines, can be incorporated into a single resource. The aim is to both facilitate analysis and to separate documentary processes from interpretative ones. This structure has been tested using a web interface for both searching and data entry. Some possibilities for representing the visual and spatial data associated with the proposed resource are also demonstrated.

Keywords: Old Norse; Semantic Web; Myth; Runes; Place-names.


[ Return to Navigation]

1. Introduction

§ 1    The Pre-Christian Religions of the North (PCRN) project aims to document and analyse the mythology and religious practices of Northern Europe prior to the introduction to Christianity. The project was established in 2006 and has over 30 contributors internationally (a draft description can be seen at http://abdn.ac.uk/skaldic/bm.php?371). The 'Sources' strand of PCRN involves producing a digital resource to document the disparate sources for the pre-Christian religion, giving access to the evidence of the early religions of the North to a wide audience (version in progress at http://abdn.ac.uk/pcrn). A challenge for this project is to make meaningful links between these sources and central concepts in the study of the religions, as well as to the other two strands of the project ('History and Structures' and 'Research and Reception'). This challenge is also an opportunity to take this project and related ones to an advanced technological level by the semantic linking of data from a broad range of sources. This requires the linking of interpretative elements to the corpora that provide the evidence for the understanding of the religions.

§ 2    This paper will take a number of different phenomena potentially representing pre-Christian religious belief and practice in Scandinavia and propose ways in which machine-readable semantic links can be made between the sources for the religions and their analysis. The sources include mythological narratives, place-names, pictorial representations and some other phenomena including folkloric material, poetry and burials. I will focus on the god Þórr/Thor and some sources associated with him, but draw examples from other areas.

§ 3    In addition I will also present ways in which this approach can be implemented through data design and eventual web display of the material. This requires structures that allow different phenomena to be connected, such as an object and a narrative. This model can incorporate links to Web and Semantic Web resources, and also be exported itself as a Semantic Web resource. The structures proposed here have been tested and implemented using a relational database with a web interface.

§ 4    The approach presented here shares some similarities with comparative, formalist or structuralist approaches to understanding Old Norse religion in that it requires some level of interpretive abstraction in order to compare equivalent phenomena. It shares the drawbacks of such approaches in that, by reducing extremely complex phenomena to a series of discrete and comparable elements, there is always the risk that the complexity is lost. The present approach, however, minimizes this risk by providing a flexible structure where the level of analysis allows for multiple interpretations. It has the added advantage that the evidence base, including geographical and temporal data, is always fundamentally linked to the analysis.

2. Background

§ 5    With some level of abstraction, the relationships between evidence and interpretation can be encoded electronically. Perhaps the most obvious and commonly used method is a relational database which can be edited and viewed as an interlinked series of web pages. This platform is the basis for a range of complex resources that are available on the Internet, especially user-contributed resources such as Facebook, Wikipedia and Twitter. The Skaldic Project (abdn.ac.uk/skaldic) uses this platform for representing and editing the complex data and relationships of the corpus of Old Norse skaldic poetry. It uses a generalizable web-based user interface to the database (the Early North Data Service — ENDS) which has been adapted for other projects, allowing multiple users to generate and publish structured and linked data for different corpora (Wills 2013). The PCRN database is built on the same platform and user interface and draws on existing relevant structures and content. At the time of writing there are six research assistants in Aberdeen and Reykjavík testing the proposed structure and populating the resource.

§ 6    The difficulty with the networked relational database approach is that there are no or few standards for representing these relationships across different corpora, and a web interface normally hides (largely for good reasons) the complexity of the linking between different data sets. Semantic Web technologies provide a means by which the relationships themselves can be encoded and exported in a machine-readable way and queried using a common interface.

§ 7    There are a number of other techniques for modelling the structures which can encode these relationships and their external reference points which the present approach draws upon. There are three general areas in this model which require linking: physical objects, textual phenomena and the mythological/religious system. The last of these is specific to PCRN as a project and should be the end point of this project, as is discussed below. The semantic representation of objects and texts is already well-established, albeit in two quite distinct ways. Physical objects, normally managed by museums and other cultural heritage bodies, can be linked to interpretative phenomena according to standards such as CIDOC CRM (The CIDOC Conceptual Reference Model http://cidoc-crm.org). Textual phenomena, including those that relate closely to physical objects (e.g. facsimiles and transcriptions), can be documented according to standards such as The Text Encoding Initiative http://tei-c.org (TEI).

§ 8     The Skaldic Database, originating and in continuous development since 2001, predates CIDOC CRM and mature Semantic Web and Linked Open Data technologies. It is nevertheless envisaged that the database structure can largely be made compatible with such initiatives. Since 2007 the XML representation of rich marked-up text has been replaced in the database with a relational data representation. The relational data model is better suited to simultaneous updates from a large contributor base and can be adapted to represent largely the same semantic and linguistic structures as conventional TEI/XML (Wills 2013).

§ 9     A number of cultural heritage bodies (e.g. British Museum http://collection.britishmuseum.org, Swedish cultural heritage bodies http://ksamsok.se) have produced Semantic Web resources which can be used to search collections for meaningful concepts within their collections. These have been increasingly incorporated into larger-scale projects such as Europeana which allow for the simultaneous searching of multiple collections.

§ 10     For organisations which deal mainly with non-textual phenomena, this approach is important for a variety of users to find materials by textual searches without having to know inventory numbers or other non-semantic identifiers. The British Museum, for example, allows the searching of the collections according to period, find spot, material and names and events associated with the object, organised using the CIDOC Conceptual Reference Model. A SPARQL interface is also supplied.

§ 11    One such object is the Franks Casket (British Museum, inventory number 1867,0120.1), which will appear in searches on the British Museum site with links to and from whalebone objects, middle Anglo-Saxon objects and representations of the legendary smith Weland and other figures on the casket. The ability to find such objects by semantic searches is extremely useful, but is still problematic for a systematic study of the northern myths and legends depicted on the casket. Some of the pictures on the casket are not indexed, particularly where the images are not reliably identifiable. The lid, for example, has a scene with the runic caption ægili (possibly the name Egill) The picture no doubt relates to a legendary narrative but is not linked to any index, presumably because of the uncertainty of the attribution.

§ 12    Additionally, the picture on the Franks Casket which is identified as Weland by the web resource is not self-evident: in this case there is no caption identifying the persons represented or the scene. The process by which an object is linked to a figure who appears in textual sources is a complex one involving several stages of interpretation based on literary sources. The identification of Weland on this object is generally accepted (e.g. Motz 1993, 713) but may preclude other interpretations, particularly ones that do not link an image to a particular figure or narrative.

§ 13    There have been various attempts to link such approaches, including more abstract discussions of the problem (Eide 2008; Ore and Eide 2009), and attempts at semantic analysis of fictional works (Boot 2006; Zollner-Weber 2011). These approaches, although complex, involve different challenges from the present project, with its diverse source types and analysis.

§ 14    One existing approach that appears to offer a method for dealing with sources on pre-Christian religions is the factoid (Bradley and Short 2005). Factoid-based approaches can be very useful for developing ways to link sources and other phenomena such as people and places (Pasin and Bradley 2013). Independently of Bradley's and others' work, the original model for this project, developed by John McKinnell (unpublished), used a similar concept to the 'factoid': a documented and abstracted source extract representing a particular mythological or religious narrative or event. McKinnell's model included what were effectively links to the source extract's different textual versions, with keywords for aiding analysis and a bibliography. I implemented this model in the electronic resource as a 'sources' table, with links to texts, mythological agents, physical objects (manuscripts), places and keywords. The model was originally developed by McKinnell to deal with short extracts from textual sources, but its application beyond that has not proved to be very useful. The interpretations involved, such as identification of people and places, are well-defined for textual sources, but are much more tenuous and complex when dealing with non-textual sources. The model does not give sufficient detail to connect narratives, for example, which appear in both written and visual form.

§ 15     The Skaldic Project is, like PCRN, based on a body of evidence that requires a large amount of analysis and interpretation to be made accessible to a wide audience of scholars and the general public. The skaldic database links manuscripts, inscriptions, transcriptions, editions and a textual and critical apparatus in order to make accessible every stage of the editing and analytical process in understanding the poetry. It brings together as transparently as is possible the connection between the material evidence for the poetry and the putative oral form of the poetry and its context of production.

§ 16    While complex, skaldic poetry is nevertheless a textual and linguistic phenomenon; the study of pre-Christian religions requires the analysis of visual and spatial phenomena and has as its goal the reconstruction of belief systems and human practices. The corpus of materials for mythology and religion is therefore even broader and requires different approaches for different sources. This paper proposes ways in which these different approaches can be synthesized in a coherent data structure.

3. Central concepts and principles

§ 17    There are some central concepts that we can use to link together the analysis of the religion to its sources. These will provide common reference points in order to link different sources together. The following is not an exhaustive list of the types of concepts that can usefully be linked to the source corpus, but represents some of the main phenomena that end-users are likely to find useful to explore through an interface to the resource.

Beings

§ 18    The religions deal with a range of named beings, objects, and places which share many similarities with things observed in the world of their practitioners, but which clearly do not belong to the human world. The most obvious are anthropomorphic figures (gods, giants, and other beings), as well as mythic animals and objects. This category also includes mythological places: unlike 'real' places, mythological places cannot be easily mapped and their importance is in relation to other entities in the mythological system rather than within a common geography with the human world. In order to understand items under the category of beings, we have to have ways of linking them together when names and representations differ.

§ 19    Many of the gods and other beings have multiple names, and in some cases the identification of a named figure with a particular god may be controversial. The category of beings must therefore be independent of the different names that individual beings are given, but connect to those names and the evidence for them.

§ 20    Pictorial representations of these beings rarely have text which identifies them as a particular being. The identification of a mythological being in a picture must be done by different means. This can be encoded as a different set of links between the sources and this category.

Attributes

§ 21    One of the main means of understanding the gods is by means of the various attributes assigned to them which distinguish them from each other and in some cases from the human world. These attributes can be visualisable, concrete phenomena such as an object (Thor, for example, carries a hammer), or physical attribute (Thor's beard is often referred to, as are his piercing eyes). Thor's name itself (< PGmc *thunraz) is the word for a physical phenomenon: 'thunder'.

§ 22    These attributes are often disparate but can be linked to the being in question by a range of usually written sources. The textual evidence for linking an attribute (such as a hammer) to a particular being (such as Thor) can then form the basis of identifying a visual representation of a male being with a hammer as Thor, or associating a hammer-shaped amulet with the god.

§ 23    Attributes may include more abstract phenomena that are associated with particular beings, such as legal assemblies, virginity or immense strength. These can be linked to the other phenomena (beings, sources) by the same structures as the more concrete attributes.

Narratives

§ 24    The religions involve stories about the various beings and their interactions with the world and each other. These form common narratives which help to build a picture of how the gods and other mythological beings interacted, and may form the basis of various pictorial representations from early Scandinavia.

§ 25    Like the previous concept of attribute, narratives normally have as their evidence base textual phenomena that describe various identifiable beings and their actions and interactions. Also like attributes of gods, this evidence base can be used to identify narratives in non-textual phenomena, and can be linked to beings in a similar way to attributes.

Practices

§ 26    A fourth concept that will be discussed here in a more limited way is the human element of worship and religious practice. This involves interactions between humans and their environment, including objects, the landscape, and particular sites. This may involve such practices as naming a place after a god, performance, passing around a horse phallus, divination rituals, worshipping a Germanic god reinterpreted as a Roman one, or using a religious or magical word or image in particular contexts.

§ 27    The issue of religious practice is difficult to reconstruct and represent because this kind of information is likely to be lost in cases where the practice has been discontinued for some time before the source appears. There are nevertheless some sources that can provide this information. The following discussion will touch on some of the religious practices that can be incorporated into the proposed resource.

Principles

§ 28    This paper defines a structure that encompasses a large proportion of these concepts and the way they can be linked to their evidence. The way in which it does this is according to certain principles of how the concepts, sources, and links are treated, namely:

  • The above concepts should be an interpretative end point: beings, attributes, narratives and practices should be the result of the analytical process of designing and populating the resource, rather than behaving as concepts that the analytical process depends upon.
  • The concepts should be linked ultimately to the evidence base in the form of items localisable in time or space to a certain extent. These include manuscripts and objects, which in most cases can be dated and located geographically; known authors, for whom similar information is often also available; and anonymous texts to which often a putative date and/or location can be attached.
  • The interpretative and analytical steps in linking the interpretative concepts to the localisable items should be explicit and as detailed as practicable. This means including the encoding of intermediate stages of analysis, some of which will be dependent on the source type and/or religious concept.
  • The implementation should use wherever possible existing authoritative public resources. Where possible Semantic Web services should be used; in other cases, semantic links should be made to conventional Web resources.

§ 29    The following discussion will be structured largely around different types of sources and how these can be linked to the above concepts. I will start with textual sources and progress to a range of other source types.

4. Narratives and textual sources

§ 30    Textual sources can give detailed and complex information about religion, but because of the dependence on literacy acquired through the church and Christianisation, they are problematic and need to be treated in a way that presents the evidence with sufficient apparatus to evaluate it.

An example text

§ 31    Snorri Sturluson (d. 1241) gives an account of an episode involving Thor and the giant Hymir where Thor attempts to catch and kill Miðgarðsormr, the serpent that encircles the world. This episode and the various forms of evidence for it are treated at length by Preben Meulengracht Sørensen (1986) and other scholars (cf. Abram 2011). The poem Hymiskviða includes this story within a broader narrative, but the clearest presentation of this part of the story is in Snorra Edda . Part of the episode reads thus:

And when Thor had shipped his oars, he got out a line that was pretty strong, and the hook was no smaller or less mighty-looking. On to this hook Thor fastened the ox-head and threw it overboard, and the hook went to the bottom … The Midgard serpent stretched its mouth round the ox-head and the hook stuck into the roof of the serpent’s mouth. And when the serpent felt this it jerked away so hard that both Thor’s fists banged down on the gunwale. Then Thor got angry and summoned up his As-strength, pushed down so hard that he forced both feet through the boat and braced them against the sea-bed, and then hauled the serpent up to the gunwale. And one can claim that a person does not know what a horrible sight is who did not get to see how Thor fixed his eyes on the serpent, and the serpent stared back up at him spitting poison. And just at the moment when Thor was grasping his hammer and lifting it in the air, the giant fumbled at his bait-knife and cut Thor's line from the gunwale, and the serpent sank into the sea. (Faulkes 1987, 47)

The translation gives an aid to interpretation for an audience not familiar with Old Norse. Ideally the text would be linked to the primary materials, which would ultimately include the manuscript(s), viewable as digital images of them. The basic structure that can represent this is a series of links representing a one-to-many relationship between images of the physical object and ultimately the repository or institution which is responsible for it. This can be represented by a relational data model (Figure 1). This model is adapted from the Skaldic Database (cf. Wills 2002) but has more recently been addressed by TEI (TEI 5 2014, §§11.1-2, 338-51). The joins represent one-to-many relationships (the small square is the many side). In other words, one repository may contain one or more collections; one manuscript page can have many images. The tables can include multiple items that are linked individually to an item in the 'parent' table. Users in this way can browse and search the various categories in order to locate manuscripts and view images of individual pages.

Figure 1: Links between manuscripts, collections and images Links between manuscripts, collections and images

§ 32    The same structure can be used for any physical object which has a visual or textual representation of a religious phenomenon. In most cases the relevant holding institution will have web-based materials that document the various phenomena here and can therefore be linked as URIs (web links). Depending on the collection, some of these structures may have to be incorporated into the electronic resource itself containing the necessary metadata to fit the materials into the structure. Eventually such data can be replaced by an authoritative external resource.

§ 33    This structure provides a means of electronically identifying the physical objects which provide evidence of the religions, and a means of incorporating information about their provenance and date which can then be incorporated into the analysis by electronic links.

§ 34    At this point we can introduce a specifically textual environment which is linked to the data associated with the physical object. Abstracting the text from the manuscript allows for incorporating and linking different versions of the text. It requires linguistic and textual divisions (words, sections, chapters) to be useful, rather than the physical divisions of the material item (codex, page). The most flexible way of creating the link between the object and text is to use the most specific physical phenomenon which contains text, namely, pages. There may be different images of a single page, but the text on it is the same for each image of the whole page. The page can form the basis of linking to the texts that are represented on it and which may occur in more than one manuscript. It is technically feasible to link the text to individual lines in the manuscript, or to sections of text by pixel coordinates in an image of the manuscript page (cf. TEI P5 2014, §§11.1-2, 338-51), but this would require a very large amount of additional data entry and processing. The page is therefore the most specific practical link that can be made between a longer text and the physical object.

§ 35    In order to effectively link pages to different parts of the text we need a relatively small but meaningful division of the text in its edited form. Most editors use chapter divisions which normally are found in one or more manuscripts, or which have been established by editors. Chapters therefore provide a useful subdivision of the text for the purposes of linking information. The basic structure can be represented as a series of relationships that link the 'page' resource above to the textual environment (Figure 2).

Figure 2: Links between texts and manuscript pages Links between texts and manuscript pages

Because there is a many-to-many relationship between chapters and pages (one chapter may occur on more than one page, and one page may contain more than one chapter), a separate table ('Text on pages') is needed to record these links in a relational data model.

§ 36     The model, originating in the Skaldic Project, addresses the same problem as the TEI sections on digital facsimiles (TEI 5 2014, 343 ff). The proposed model does not distinguish between 'surface' and 'page'.

§ 37    Different manuscripts, editions and translations will sometimes use slightly different divisions and may result in different numbering. The resource therefore needs to select a single set of subdivisions in order to accurately link information to the text. A table containing at least the metadata necessary for identifying the particular relevant piece of text will be necessary for linking the chapter to other phenomena. This will contain numbering that is unambiguous; ideally this will be based on a scholarly edition, the text of which will provide a way of identifying the correct subdivision of the text for linking authoritative versions and other phenomena. For internal reference, a representative text should also be incorporated into the resource. Digitised out-of-copyright texts of relevant sources are often available on the internet and can serve this purpose.

Figure 3: Fragment of tables linking texts and manuscripts Fragment of tables linking texts and manuscripts

§ 38    The links between rows in each table can be seen in the Figure 3, which represents a fragment of the overall schema. Within each row, columns store information associated with each item, such as dates and locations (for manuscripts and objects), biographies for authors, introductions to texts, editions and translations for chapters, and so on. The lines represent links between rows in each table, with one-to-many relationships shown by links between single rows in one table and multiple rows in another. A user interface can provide various means of searching and browsing the relevant texts and content, with links to the manuscripts and images of manuscript pages which contain the relevant text.

§ 39    There is considerable potential for populating such a structure because of the large number of resources that have been digitised and are available via relatively stable URIs. For editions and translations of Gylfaginning ch. 48, for example we have publicly available Faulkes’ edition (p. 44: http://www.vsnrweb-publications.org.uk/Edda-1.pdf#page=75 – accessed 19 June 2013), Finnur Jónsson’s edition (ch. 32, p. 62: http://www.septentrionalia.net/etexts/finnur_SnE.pdf#page=131 – accessed 19 June 2013), Brodeur’s translation (p. 69: http://www.northvegr.org/the eddas/the prose edda - brodeur translation/069072.html – accessed 19 June 2013) and Guðni Jónsson’s text (ch. 48 http://www.heimskringla.no/ wiki/Gylfaginning – accessed 19 June 2013). In some cases, such as particularly important and relatively short extracts, members of the PCRN project may make editions and translations of their own.

§ 40     Some of these resources are bibliographic and further reference can be made to an internal bibliography or external resources such as openlibrary.org. The references can be incorporated into the relational model (Figure 4).

Figure 4: Links between texts and editions Links between texts and editions

In some cases the text will form part of another text, such as þættir and sagas in compilations (e.g. Gylfaginning forms the first part of Snorra Edda ). These relationships can be encoded as a recursive reference to the texts table within the texts table.

§ 41    The same basic structure can be applied to poetry. In the case of stanzaic poetry (the overwhelming majority of Norse poetry), the most useful division is the stanza itself. The different types of text division are indicated, where necessary, by another column in the table. For non-stanzaic poetry such as is found in other Germanic languages (e.g. the Old English poem Beowulf) another set of divisions may be needed in order to break longer poems into more manageable units for referencing. For Beowulf the 40-odd numbered sections in the manuscript are used.

§ 42    Runic inscriptions normally have a one-to-one relationship between the text and object which makes the structures above redundant for representing the relationship between the text and object. However, this is not always the case: some objects contain more than one runic inscription, and some inscriptions are repeated on more than one object, or very occasionally extend over more than one object. In addition, the text may be recorded on more than one side of the same object. This paper proposes retaining this structure for runic inscriptions, which additionally allows for linking images to specific sides of the inscription.

§ 43     Non-textual phenomena can also fit in this structure, albeit with some conceptual adjustments. Physical objects with pictorial representations of potentially religious phenomena can normally be divided into one or more separate pictures (such as the panels and sub-panels of the Franks Casket). These pictures can be treated as the equivalent of chapters, stanzas and inscriptions as linkable units for linking interpretative data.

5. Narratives in texts

§ 44    Returning to the myth of Thor and the World Serpent, in order to create links between the story represented in Gylfaginning 48 and other versions, the narrative must be divided into elements that may or may not be present in possible other versions of the narrative. These may include:

  1. Thor goes out in a boat
  2. Thor and Hymir row out to sea
  3. Thor fastens ox-head to line
  4. Thor casts line
  5. Miðgarðsormr bites bait and pulls
  6. Thor puts foot through hull
  7. Thor pulls up Miðgarðsormr
  8. Thor and Miðgarðsormr glare at each other
  9. Miðgarðsormr spits poison at Thor
  10. Thor raises hammer
  11. Thor strikes Miðgarðsormr with hammer
  12. Hymir cuts fishing line
  13. Miðgarðsormr sinks into sea

The element (k) does not appear in Snorri’s version but does appear in at least one other version; Gylfaginning therefore contains elements a, b, c, d, e, f, g, h, i, j, l, m. Elements (a) and (b) are here separated for reasons that will be apparent in the next section.

§ 45    The division of the story into elements allows us to link other versions which may not have the complete story or the same details. The earliest textual evidence for the story is in a series of stanzas attributed to Bragi Boddason, a ninth century Norwegian poet. These are traditionally taken to be part of Ragnarsdrápa (Finnur Jónsson 1912-15, BI, 3-4; numbered 14-19) but in the forthcoming new edition of the skaldic corpus have been categorized separately. The stanzas contain many of the elements above: Stanza 1: Thor goes to fight Miðgarðsormr (a, b); Stanza 2: fishing line tightens (implies d, e); Stanza 3: Thor recognises Miðgarðsormr (h), raises hammer (j); Stanza 5: Thor drags up Miðgarðsormr (g); Stanza 6: The giant cuts Thor's fishing line (l)

§ 46     A more abbreviated version occurs in the Icelandic poem Húsdrápa by Ulfr Uggason, dated to 938 by Finnur Jónsson (1912-15, BI, 128). Four stanzas deal with this episode, including: Stanza 4: Fierce glares between Thor and serpent (h); Stanza 5: Miðgarðsomr glares (h) and spews venom (i) from the gunwale (implies g); Stanza 6: Thor strikes Miðgarðsomr (k); Miðgarðsormr’s head comes off and sinks into the sea (m). The fact that Laxdœla saga (Ólafur Sveinsson 1934, ch. 29, p. 80) attributes the origin of this poem to a description of wood panels in a house in Iceland gives this a possible further corroboration, but because the original objects are not extant this cannot be incorporated into the structure.

§ 47    Another version is to be found in the anonymous poem Hymiskviða (before c. 1170, cf. von See et al. 1997, 277): Stanza 20: Hymir doesn't want to row further out (b); Stanza 21: Thor drops line (d); Stanza 22: Thor uses ox head for bait (c); serpent bites (e); Stanza 23: Thor strikes with hammer (k); Stanza 24: serpent sinks back into the sea (m).

§ 48     The structure can be represented fairly simply, with links between the narrative elements and the divisions of the text (chapters and stanzas) using a separate table to indicate instances of a narrative element in a particular text division (Figure 5).

Figure 5: Links between texts and narratives Links between texts and narratives

The linking of multiple forms of the narrative by individual elements gives a means of comparing different versions. Because the narratives are linked by these means to the manuscript evidence, they can be assessed by the number and complexity of medieval versions of a narrative (both as texts and manuscripts), as well as the temporal and geographical distribution of these narratives.

§ 49    When this information is displayed for an end-user, there are various ways of bringing together these connections, such as simple lists of narrative elements for each text, or lists of texts with links for each narrative. The geographical and visual data linked to the physical objects can also be used for maps and images. The display of such information will be demonstrated below.

6. Pictorial representations of narratives

§ 50    In order to explicitly link narratives to non-textual sources, we need a way of using elements of the narrative which are not dependent on names, as names rarely appear in pictorial representations. Part of this process should be to determine objective elements which could be represented visually; in many ways this process will anticipate the kinds of pictures which have previously been linked to the narrative.

Figure 6: Altuna runestone, retouched (Wikimedia Commons) Altuna runestone, retouched (Wikimedia Commons)

For example, the Altuna stone (Riksantikvarieämbetet (RAÄ) Altuna 42:1, Upplands runinskrifter (U) 1161: c. 1150 — Figure 6) has an image which appears to relate to the narrative in question. There is a (probably) male figure in a boat; the man holds a hammer; a foot is represented through the bottom of the boat's hull; the man appears to be holding a fishing-line; the line has (possibly) an ox-head on the end; a serpent-like figure appears below the boat and may be biting the ox-head-like object on the end of the line.

§ 51    Despite there being only one figure in the boat, this is clearly a representation of Thor's fight with the World Serpent. However, there is an element of interpretation — there is nothing linguistic on the stone or inscription to identify the male figure and serpent with Thor and the World Serpent. Thor is a male figure who is often represented with a hammer, which may in itself provide sufficient information to identify the figure in the boat. However, there is also a collection of elements that provides the link with Snorri's account of the story. The identification here is based, firstly, on the identification of a figure with a hammer as Thor; and secondly and potentially independently, on the basis of elements that match the narrative given in Snorri and supported by other sources.

§ 52    The encoding of this relationship then involves a collection of narrative elements or features rather than the use of a name. In order to encode the special status of this inscription as containing both a fairly clear Thor figure (man with hammer) as well as elements of a narrative about Thor (man in boat with foot through bottom, fishing for serpent, etc.) we potentially need two sets of entities here. I will firstly discuss the representation of narratives as individual pictorial elements, and then the representation of individual figures from the mythology.

Narrative representations

§ 53    There is potentially a one-to-one correspondence between the narrative elements discussed in the previous section and those represented in the picture, and the division of narratives into elements should take into account the potential divisions of visual representations of the story.

§ 54    The divisions allow us to connect the story with others. It should be noted that elements a, b, d, and l may be represented in a generic fishing episode. The presence of the foot through the hull, hammer, ox-head, and possibly serpent, however, probably represent this particular narrative as they would be somewhat unusual in a story of a fishing trip in a different context.

§ 55    The Altuna runestone (U 1161; RAÄ Altuna 42:1 — Figure 6) contains a, j, d, c, e, f; the last three should be linked indicating a degree of uncertainty. The second figure (Hymir) is missing, perhaps because the surface on which the picture is inscribed is very narrow.

§ 56    The Gosforth stone (The Corpus of Anglo-Saxon Stone Sculpture (CASSS) Gosforth 6 http://www.ascorpus.ac.uk) shows two men in a boat (a, b); one holds a fishing line (d) with an ox head (indicated by horns) on the end (c); the other holds probably a hammer (a bit unclear; j); there are fish in the water (not part of the story and probably used for decorative purposes). This image lacks the foot-through-hull motif and the serpent, but has the second figure.

§ 57    The Hørdum stone is a much simpler picture with two men in a boat (a, b); one holds a fishing line (d); below him is a foot through the hull (f); a second figure holds a hook or small sickle (l); there are two curved lines below the boat that may have originally represented a serpent, but now identifiable only with a great degree of uncertainty. The only certain element that is unique to the narrative is the foot through the boat hull, but the overall connections should provide enough links to the narrative (cf. Meulengracht Sørensen 1986, 265‑6).

§ 58    Not discussed here is the Ardre VIII (Statens Historiska Museet (SHM) 11118:VIII) stone from Gotland which may represent more than one scene in the narrative (Meulengracht Sørensen 1986, 269).

§ 59    The structure that can be used to represent this is the same as for the textual narratives above: the pictures are treated structurally in the same way as texts, with links to the narrative elements via a third table. This structure must explicitly include the assumption, when linking and analysing these relationships, that the attribution to a non-textual source is on the basis of the narrative elements alone, without reference to individual named beings.

§ 60    The various materials — texts, manuscripts, objects, narratives — can be brought together as a series of linked resources for a web interface, including maps of the geographically locatable sources and images of the material sources.

§ 61    The different representations of what appears to be the same story point to the diversity of the religion: various aspects of the story can be left out (such as the foot, second figure, etc.) but it remains a recognizable story. This diversity is thus encoded into the structure of the resource described here.

Representations of individual figures

§ 62    For the identification of mythological figures such as gods it may be sufficient to use individual attributes to identify them. This process is frequently represented in interpretations of individual figures from the period and include: a man with a hammer = Thor; a man with one eye = Óðinn; a horse with eight legs = Óðinn’s horse, Sleipnir; and a man with an enlarged phallus = Freyr.

§ 63    These relationships can be represented by a table of attributes for gods which represent sufficient conditions for interpreting images and texts that do not identify mythological beings by name. This is problematic for two main reasons: it may encompass representations of features that could apply to other interpretations (one-eyed men or men with hammers may represent something other than a god); and the basis for attributing features to the gods may itself be problematic. For example, a large number of sources indicate that Thor has a hammer, but we have one potentially unreliable source (Adam of Bremen’s Gesta IV.26) for Freyr being represented with an enlarged phallus. This representation is the basis for identifying the Rällinge figurine from Södermanland (SHM 14232) with Freyr.

§ 64    These problems cannot be overcome in themselves, but the evidence base for the attribution can at least be represented by the data model. The most neutral representation for the visual image of a potential mythological being is simply to link the attributes to the object itself. For the Eyrarland figure interpreted as Thor (Þjóðminjasafnið, Reykjavík (Þjms) 10880), this will mean linking a male figure and a hammer to the figure itself (Figure 7). For an image such as this, the page/side links are redundant but are included in order to maintain a consistent structure: some figures may include more than one image on one or more sides of the object.

Figure 7: Links between attributes and objects Links between attributes and objects

§ 65    The textual evidence for Thor as a male figure with a hammer consists in linking the attribute to a specific named being. This requires a slightly different structure which, firstly makes the link between the god and the attribute, and secondly links the attribution to the corpus. For the interpretation of a male figure with a hammer, the interpretation of Thor as a male figure is self-evident — despite his cross-dressing activities in Þrymskviða , he is always represented as male. Thor has a hammer according to a number of sources, including Snorra Edda (Gylfaginning ch. 48; Faulkes 2005, 47), Þjóðólfr ór Hvini, Haustlǫng 18 (Finnur Jónsson 1912-15, BI, 18), Bragi Boddason, Þórr’s Fishing 3 (formerly Ragnarsdrápa 15; see Finnur Jónsson 1912-15, BI, 3), SHM 25654 (Södra kvinneby amulet) and Þrymskviða 1 (Neckel and Kuhn 1983, 111). The evidence base for identifying a man with a hammer as Thor is shown in Figure 8. In cases where the attribution is obvious (e.g. Thor is a male being), this status can be indicated in the table (here represented as an asterisk, but occupying a separate field in the table).

Figure 8: Links between texts, beings and attributes Links between texts, beings and attributes

§ 66    This system separates but links the evidence base (references) to the interpretative connections; the amount of evidence and relative dating can be incorporated into the analysis by the links; thus the Eyrarland image is connected to Thor by the image of a man with hammer; this shares the same connection as a number of textual sources, including a runic inscription and a ninth-century poem. The Rällinge statuette identified as Freyr, which has an enlarged phallus, shares a connection with the god based on a more limited range of sources.

Other objects

§ 67    The examples above include anthropomorphic figures which can be identified with mythological beings or narratives, however tenuously. There are other types of objects which do not fall into these categories which nevertheless may be useful for understanding the pre-Christian religions. One such example is the Thor’s hammer motif that is found on a large number of objects from Scandinavia. These objects consist normally of an amulet-like object made of metal with a two-dimensional representation of a double-sided hammer with a very short handle.

§ 68     Such objects do not need to be explicitly linked to a particular mythological being to be of interpretative value. The Attributes category can be used to label objects with this type of hammer image and can be linked to the object via Attributes in texts table. Where multiple objects are extremely similar or identical, perhaps because they are modelled on the same image or even cast with the same die, these can be represented by a single entry in the category of pictures with multiple links to the objects themselves.

§ 69     Folkloric sources which mention named beings can be linked to the resource according to the structures outlined in the previous section (5). For sources which have clear connections (such as narrative motifs) with mythological narratives but do not use identifiable mythological names, the structures outlined in this section can be used to link them to the interpretative concepts.

7. Linking the religion: names and beings

§ 70    In the extract above we have three mythological beings participating: the god Thor, the giant Hymir, and the World Serpent (Miðgarðsormr). Indexing their names gives an automatic link between the text and other instances of the being or name 'Thor', 'Hymir', 'Miðgarðsormr'. His hammer also has the name Mjǫllnir. This indexing can be performed by a direct link between the word in the text and the being in question (Figure 9).

Figure 9: Links between texts and beings Links between texts and beings

§ 71    There is a potential problem in some other sources: these beings may be represented by other names. For example, the World Serpent is sometimes called Jǫrmungandr (possibly 'the great monster'), and Óðinn regularly adopts other names. There is therefore a need to link a more abstract entity for these beings than simply the name itself, but we should also keep the name as a concrete link to the texts. In some cases the evidence for the name applying to a particular god is equivocal or contradictory, in which case a further link from the application of the name to the textual material will help the user to determine the reliability of the attribution.

§ 72     For example, the identification of the eponymous subject of Rígsþula (Rígr) with the god Heimdallr occurs in the introductory prose to the poem in the Arnamagnæan (AM) manuscript AM 242 fol. This structure is represented in Figure 10. The identification of Rígr with Heimdallr has been disputed by scholars (e.g. Simek 1993, 264). The database structure includes the source material, including date (the ms. is fourteenth century) which helps to evaluate the evidence of the identification. This could contribute to further analysis, for example, of the god Heimdallr.

Figure 10: Links between texts, names and beings Links between texts, names and beings

§ 73    In other instances the alternative naming of a god will have a better foundation, such as the lists of names found in the þulur . The primary name of a god should be indicated as such in the 'Names' table in order that the structure does not presume it requires further justification. The names, however, can provide a linguistic link from the mythological being to other phenomena such as place names, which themselves may provide only the linguistic connection with a god or other being.

8. Place names

§ 74    Place names present a different set of problems for semantic linking. They have considerable potential in the study of pre-Christian religion by connecting lexical information about the religion to the geography of Northern Europe (Brink 2007, 106‑7).

§ 75    Place names have a similar base in textual evidence as the other phenomena represented here. The textual sources for names can therefore be incorporated into the same structure. The link between the text, a religious being or phenomenon, and a place is complex and may be structured in the following way: a section of text contains a series of words, one or more of which may be a relevant place name. The words are linked separately, and the place name itself is then linked to a particular place (the modern place, or other names for the place, should be distinguished from the linguistic information). The words can also be linked to a table of dictionary headwords and broken into the elements required for analysis relevant to religions, namely, two or more components of a compound, where relevant. The dictionary headword can then be linked to the names table presented above, which makes the linguistic connection between a word and a being or other phenomenon relevant to the project.

§ 76    For example, the place Torsburgen on the island of Gotland is an identifiable location. It is first mentioned in Guta saga in chapter one, which is found on fol. 43v of Stockholm ms. B 64. The name used is 'þors borg' which can be linked to the dictionary headword 'þórsborg' (using the West Norse form for convenience, but different versions in cognate languages can also be linked in the same entry). This can be linked by a further table back to the dictionary lemmas as two parts of a compound ('Þórr' and 'borg'). The word Þórr' can be linked to the god via the names table (Figure 11).

Figure 11: Links between beings, place names and their sources Links between beings, place names and their sources

§ 77    The level of reference to original texts may be beyond the scope of the proposed project, as place name archives draw upon large numbers of sources which may otherwise be irrelevant to the project. However, even without such detailed reference this structure will be able to provide sufficient potential for analysis.

§ 78    The links to the text provide the earliest known and subsequent dates for the naming of a place, as well as potentially the geographical location of the evidence for the naming. This can be used to build maps of theophoric names. Similarly, other place-name elements (such as lundr 'grove') which may have significance for religious beliefs and practices can be used to build maps by the same mechanism. For example, theophoric place names which relate to groves as potential sites of worship can be represented on a map (see section 12 for examples).

9. Equivalence with Classical figures

§ 79    Useful early information about the religions of the North and their gods can derive from early medieval processes of interpreting equivalences between the Germanic gods and Classical ones. The terms given to these processes are the Interpretatio romana and Interpretatio germanica , the former being where people connected with the Roman world interpret an apparent Germanic deity as a Roman one; the latter where the reverse process takes place by Germanic people (Simek 1993, 174‑5).

Interpretatio romana

§ 80    In Tacitus’s Germania (c. 90 CE) there are a number of gods named as worshipped by the peoples he describes. In one instance (ch. 9; pp. 142-5) he names the principal gods as Mercury, Hercules and Mars. This does not provide particularly solid evidence for the identification with particular Germanic deities in itself but the features of these gods can be linked to the Germanic gods Óðinn, Thor and Týr by various means, including as a triad of chief male deities.

§ 81    Hercules, although a demi-god, has similarities with Thor that can be expressed by certain attributes, including using a blunt weapon (Hercules carries a club; Thor a hammer), fighting monsters, and otherworldly creatures (Hercules fights the Hydra and captures Cerberus; Thor fights giants, the World Serpent, etc.).

Figure 12: Interpretatio romana Interpretatio romana

§ 82    The constellation of attributes can be mapped to the tables outlined above (Figure 12): beings, beings' attributes, attributes and attributions in texts; the last of these is built on a broad range of evidence which may not need detailed incorporation into the database in order to be useful. The classical material may be incorporated more sparingly, as the Northern material will generally have more detail relevant to the project.

§ 83    To represent the Hercules-Thor interpretatio , we can record: the two gods as beings, the attributions for each of them that coincide, and the texts that evidence these attributions together with their primary witnesses. The equivalence of the gods can then be built on the linking of these attributes and the source base for it.

§ 84    Particular manifestations of worship of these equivalent deities is found in votive inscriptions, such as that to Hercules Magusanus which occurs in a number of locations (cf. Simek 1993, 141-2). Here the second element (magusanus 'mighty') provides a further corroboration, although disputed; the relevant interpretations (e.g Wagner 1977) should be incorporated into the structure.

Interpretatio germanica

§ 85    The process by which Roman gods were interpreted by Germanic-speaking people as equivalent to or identical with their own gods has found its way in particular into the weekday names found in most Germanic languages. These provide a way of linking the worship of certain manifestations of Roman deities with those of the Germanic world.

§ 86    A similar structure to the one in Figure 12 can be used for the interpretatio germanica . The primary attributes are the day of the week that each god is associated with (e.g. Jupiter and Thor for the fifth day of the week). These can be linked to both the Roman and Northern gods in the 'Beings' attributes' table, and references to the weekday names in sources can be linked to the textual and inscription corpus using the 'Attributions in text' table.

10. Runic inscriptions

§ 87    Runic inscriptions can be incorporated into the structure using the same tables as discussed for textual materials, as discussed in section 4 above. As a unique object containing a primary witness of a text, a runic inscription is taken as equivalent to a manuscript. The sides of the inscription which may contain text are taken as equivalent as pages in a manuscript, using the numbering (usually alphabetic) from the standard edition of the inscription. The whole inscription is taken as the textual unit which is to be analysed or linked to religious phenomena in the same way as chapters or stanzas for manuscript texts. The structure may be represented as in Figure 13 using the runestone from Glavendrup, Fyn as an example.

Figure 13: Links between runic texts and objects Links between runic texts and objects

§ 88    The text can be treated in the same way as other units of text such as chapters and stanzas. For example, runic inscriptions that mention gods and other named beings can be incorporated into the structure in the same way as place names and attributes: Danmarks runeindskrifter (DR) Fyn 26 (lavendrup, formerly DR 209) appears to call upon Thor to consecrate the runes. The reference to Thor can be built into the structure as represented in Figure 14.

Figure 14: Links between runic inscriptions, beings and attributes Links between runic inscriptions, beings and attributes

§ 89    Some words, particularly in older inscriptions, appear to have some kind of religious or magical significance, including the nouns alu 'ale', laukar 'leek' and verb vígja 'consecrate'. These can also appear without any apparent significance, but where some significance is presumed because the word is used in isolation or with other words suggesting religious or magical significance, these can be linked by a table to the words in the texts. Such phenomena are shown in Figure 14 as the 'Practices' table, which links lexical units to named beings and attributes via the 'Beings' attributes' table.

§ 90     These structures can be used to generate maps and links to various phenomena associated with a god or a practice, such as all inscriptions mentioning Thor, or all inscriptions mentioning consecration, or a combination of the two.

11. Burials

§ 91    Burials are highly complex phenomena which can provide evidence for various religious and cultural belief systems. They can comprise human remains and clothing, a burial context, furnishings and other material, and spatial phenomena which may be of interest individually or in connection with each other.

§ 92     The electronic resource as already described can incorporate spatial data, objects and descriptive information, so that the physical phenomena of burials can in themselves be incorporated into the resource. The interpretation of the burials needs further elaboration and its full implementation is beyond the scope of the present paper; I will, however, present some ideas for representing some of the interpretative possibilities associated with burials.

§ 93     An example is the Birka burial Bj. 834 which is discussed at length by Price (2002, 132-9) in relation to the pre-Christian practice of seiðr . This contains a woman's skeleton and may have also contained a man's, along with a number of objects including jewellery and an iron distaff. The connection between the skeletons, objects and their positions within the burial are what give rise to the interpretation.

Figure 15: Interpretation of a burial Interpretation of a burial

§ 94     Figure 15 maps some of the relevant phenomena to the interpretation of this burial. The physical objects and materials of the burial are entered in the normal table for these objects and connected to the site and burial using the collections table (with an appropriate type description). The objects are linked to a description of the burial (entered in the same category as descriptions of pictures) which should provide a non-interpretative description of the various objects and their placement. This in turn provides a means for linking the relevant attributes and the concept that is to be discussed in relation to this burial, namely seiðr . The structure can also be used to provide analysis and evidence for other phenomena.

§ 95     An issue with this particular method is that the description removes the direct semantic connection between some of the objects and the attributes category (e.g. the iron staff). In such cases a direct link between the object and attribute can be made.

§ 96     High-status burials such as Oseberg and Sutton Hoo represent much more complex phenomena. The overall features nevertheless fit roughly to the categories in the structure as currently conceived: real humans (here those buried rather than known authors or poets); objects which may individually or collectively represent cultural practices; attributes of the burial such as position of body and location in relation to other sites, objects and burials.

12. Populating the resource

§ 97    The proposed structure is highly complex and covers a range of specialist disciplines such as philology, runology, onomastics, archaeology, palaeography and codicology, and classical studies in addition to history of religions. Populating such a diverse resource is potentially a very large undertaking. Fortunately there are existing resources that can be incorporated either directly or indirectly or through Semantic Web and other technologies.

§ 98    The Skaldic Database includes a range of materials which can be linked to the PCRN database and provide a basic resource for building the Old Norse material related to pre-Christian religion, including material from the Ordbog over det norrøne prosasprog (dictionary headwords, prose works, manuscripts, etc.); the project's own corpus (skalds, poems, stanzas, text, translation, notes, manuscripts, textual and critical apparatus); some 30,000 ms. images digitised by or for the project itself; a bibliography of some 3,000 items; and an index of mythological references (esp. heiti and kennings) in the skaldic corpus.

§ 99    There are other existing digitised resources which can be linked to the proposed resource on religions. The Samnordisk runtextdatabas is one such data set: this resource, which includes text, translation and information about all Scandinavian runic inscriptions, it is planned, will have Semantic Web resources and can be linked to the PCRN resource. Handrit.is provides detailed manuscript information and images of a very large proportion of the Old Norse manuscripts held in Iceland and elsewhere. Additional images are supplied by the Stofnun Árna Magnússonar and can also be linked. The Monumenta Germaniae Historica has been recently digitised and includes a very large amount of source material in different languages. This does not provide semantic linking of these sources, but semantic links to this resource can be supplied by the PCRN project and linked to the digitised version. In addition, Riksantikvarieämbetet supplies data resources for cultural heritage materials from Sweden. There are a great many more such resources, particularly for non-manuscript sources.

§ 100    Using the main example from this paper for textual sources (the Thor narrative), a large proportion of the data structure's components can be linked to authoritative materials on the Web, including:

§ 101     Figure 16 shows the full database schema. Data sets which have significant materials available on the internet are marked with a ‡. In addition, data sets that have the potential for dating or geographical information are also indicated.

Figure 16: Full database schema with mixed examples Full database schema with mixed examples

§ 102    For resources that are not available by Semantic Web or similar means, the data entry process will involve some level of semantic analysis on the part of the individuals populating the resource. In particular, the phenomena associated specifically with pre-Christian religion (generally the left-hand materials in Figure 16), will require some training and expertise in the field of pre-Christian religions. This includes narratives and their elements, attributes, beings and other religious phenomena, and the links between them and the texts. In addition, as most of the textual material has not been linked to dictionary headwords, this process will also have to be performed by the PCRN project where necessary.

§ 103    The data can be entered through a web interface. A similar interface is used for the Skaldic Database and has been adapted and implemented for all the data sets described in this paper (login required; see Figures 17 and 18). Although the structure appears to be highly complex, the process by which the different elements are linked is greatly simplified by the interface. Modern web interfaces such as this include mechanisms for quickly searching and linking items and can prompt users based on previous and likely or possible choices. There are various mechanisms for one-to-many linking, such as drop-down menus and Ajax lookups (Figure 17 gives an example of the editing form for linking narrative elements to text). Many-to-many links can be made using tag prompts such as those used for entering addresses in modern email software (Figure 18 shows the editing form for general information about narratives, including mechanisms for linking participants in the beings table to the narrative).

Figure 17: Data entry for narratives: linking narrative elements to texts and images Data entry for narratives: linking narrative elements to texts and images

Figure 18: Data entry for narratives: linking narratives to participants Data entry for narratives: linking narratives to participants

13. Accessing the resource

§ 104    The semantic links described in this paper can be represented in a variety of ways, the simplest being as a series of web pages generated by an interface to the database. The pages can contain links to other materials in the resource, or other media where relevant.

§ 105     As discussed in the previous section, Figure 16 shows a number of phenomena that can be geographically locatable, such as places and objects, and these are in turn linked to the religious phenomena represented by the central concepts of the resource. These links can be used to generate, for example, a map of the theophoric place names in Scandinavia, containing elements referring to particular gods. An example of such a map is shown in Figure 19 (place names are taken from Brink 2007 and supplemented from other sources; the author has added coordinate data). In this map yellow dots represent Óðinn place-names, red dots represent Thor, green: Freyr, and blue: Njǫrðr/ *Niærþer. On an online version of the map, the dots can be linked to the original name forms and sources that contain them in the database. References to other place-name elements (such as lundr 'grove') which may have significance for religious beliefs and practices can also be used with the same mechanism to produce maps of, for example, theophoric place names which relate to groves. This is a fairly simple example but more detail can be provided, for example, by providing representations of insecure etymologies or second elements, such as the interactive map of Freyr - place-names at abdn.ac.uk/skaldic/db-map.php?id=172.

Figure 19: Theophoric place names in Scandinavia Theophoric place names in Scandinavia

§ 106    Geographical data and other media can be combined into synopses of particular phenomena such as beings or narratives. Figure 20 shows a fragment of a web page generated from the resources described in this paper about the narrative of Thor’s encounter with the World Serpent. This includes some basic information about the narrative, along with links to the main material records of the narrative (manuscripts and picture stones), a map of locatable objects and texts of the narrative, and links to individual narrative elements.

Figure 20: Resources associated with the example narrative Resources associated with the example narrative

§ 107    The links to primary materials and map data in Figure 20 are implemented using the relational structure described in sections 4-6 of this paper. Here the map shows only the four visual representations (Altuna, Gosforth, Hørdum and Ardre). The right column of Figure 20 shows links to the textual sections with the narrative using thumbnail images of the stones or manuscript pages upon which the story is recorded. The objects are ranked according to the number of links between the narrative and the object.

§ 108    For users who need an overview of sources on broader particular topic, sources can be ranked according to the number of links made between the topic and the physical sources; these could also be weighted according to other factors such as date. These kinds of weightings lie behind the rankings in Google searches, for example. The same process can be used, for example, when selecting a particular source: topics associated with the source can be ranked according to the number of intermediary links.

14. Conclusions

§ 109    A wide range of phenomena can be linked and exported according to various models, despite the extremely disparate body of sources for the pre-Christian religions of the North. Although other source types and problems arising from them are not documented here, it is likely that with minor modifications they can be incorporated into the structure.

§ 110    This paper has proposed that the central concepts relating to religious phenomena can be abstracted from the texts and images that bear witness to them; and that the texts can be abstracted from the material evidence for them. This results in an analytical structure for the pre-Christian religions in which the religious phenomena, the textual basis for them and the material evidence for them are all linked, and the interpretative processes are separated from the non-interpretative processes. This provides a firmer foundation for the study of pre-Christian religions than projects which do not integrate the material evidence and analysis.

§ 111    The structure allows for the investigation of new questions that hitherto have been difficult or required duplicating data gathering, such as: How diverse was the worship of certain figures and other religious practices? What is the evidence base for accepted religious attributions and categories? What changes occur in worship over time (linking to datable resources such as archaeology, dated texts, etc.)? What are the geographical distributions of worship (linking to placeable resources such as place names, runic inscriptions, etc.)?

§ 112     There are difficulties in populating such a large and diverse resource, but the large amount of existing material currently available on the Web makes this process considerably easier than would have been possible even very recently. As the resource is developed there will no doubt be a need for further revisions of the structure and relationships, but this paper shows that some of the main analytical processes for the PCRN project can be represented in an electronic form.

Appendix

The following documents the proposed structure (Figure 16) as a relational database as currently implemented. Most categories are represented as tables in a relational database. Columns in these tables that link to other tables are given in italics. This is not meant to be an exhaustive list of tables and columns for the proposed resource but lists some key information types associated with each phenomenon.

Some categories are implemented using the generic 'rdf' table, which is structured like the XML RDF application as a subject-predicate-object triple. The subject is a reference to a row in a particular table, as is the object; the predicate is a link to the 'predicate' table in the database which defines the relationship between the subject and object.

Where there is a correspondence with CIDOC CRM class definitions, these are shown using their identifiers (e.g. E23). Other classes are implicit in the values given in table columns: many such tables include a column for dating an entry, corresponding to class E4 Period (CIDOC CRM, 3). Property values correspond to certain column types in the relational model. For texts, the language may be identified in a column corresponding to P72 'has language' (CIDOC CRM, 55). In other cases, the property may be expressed by a relational link to another table.

Repository (E40; The body responsible for holding and/or documenting the object — not yet implemented)

  • Fields: Name (P1), Abbreviation, Location, URI.
  • Existing resources: Various institutional websites.

Collection (E24; A manuscript collection, or a standard or convenient grouping of other object types — 'collections' table)

  • Fields: Link to Repository , Name (P1), Abbreviation, URI
  • Existing resources: handrit.is, www.bl.uk, etc.

Manuscript (E84; Any unique object such as a manuscript, runic inscription, artefact, picture stone or burial — 'mss' table)

  • Fields: Link to Collection , Siglum/shelfmark/inventory number, Name (P1), Date, Provenance, Description
  • Existing resources: handrit.is, www.bl.uk, britishmuseum.org, raa.se, CIL, etc.

Page (A roughly two-dimensional part of the object containing text, image or other relative information, or a whole object — this is currently not represented by an actual table but by references to page numbers in the Reference and Image tables)

  • Fields: Link to Manuscript (P59), Number, Alternative number

Image (E38; An image of a side or page of the object — 'images' table)

  • Fields: Link to Page , URI or data, Type, DPI, Source
  • Existing resources: handrit.is and similar sites

Reference (A link connecting the Page to a picture, chapter, stanza or text of an inscription — 'refs' table)

  • Fields: Link to Page, Link to Chapter/Stanza/Picture

Chapter/Stanza/Rune/Picture (A single piece of semantically separable text or image — 'verses' table)

  • Fields: Link to Text/Poem (where relevant), Number, Alternative number, Title, Title translated, Rough Text, Rough Translation
  • Existing resources: heimskringla.no and similar web sites

Editions of text (Versions of the chapter text or translation — 'edrefs' table, partially implemented)

  • Fields: Link to Chapter/etc., Link to Bibliographic work, Number, Title, Text/translation

Bibliographic work (A standard bibliographic reference with appropriate fields — 'bibl_works' table, with links to 'bibl_authors', 'bibl_journals', 'bibl_images', etc.)

Texts/Poems (E33; Prose works or poems — 'text' table)

  • Fields: Link to Author (where relevant), Title, Introduction, Dating, Provenance

Author (E21; The author of a primary work — 'skalds' table)

  • Fields: Name, Dates (b., d., fl.), Biography

Word in text (Words that appear in the text segments — 'word' table, also requires 'line' table)

  • Fields: Link to Chapter/etc ., Link to Lemma , Link to Place, etc ., Relative ordering, Word form, Translation in context

Place (E53; A real named place, not necessarily identified — currently implemented as part of the 'thing' table)

  • Fields: Link to Word in text , Modern name, Other names, Type of name, Latitude/Longitude coordinates

Lemma (Dictionary headword — 'lemma' table)

  • Fields: Lemma, Part of speech, Gloss
  • Existing resources: onp.hum.ku.dk, Fritzner’s Ordbok, etc.

Lemma part (Parts of compound words in dictionary lemma — 'lemmapart' table)

  • Fields: Link to Lemma (compound), Link to Lemma (part), Relative order of part

Name (E41; A link between a word and a being or thing — 'name' table)

  • Fields: Link to Lemma, Link to Being, Name

Name in text (A link between the application of a name to a being and the source of that application — currently implemented using the 'rdf' table)

  • Fields: Link to Name, Link to Chapter/etc.

Being (A mythological being or other unique phenomenon — part of 'thing' table)

  • Fields: Main name, Type, Description, Links to Being (as mother, father)

Being's Attribute (The linking of a being to an attribute — 'attribution' table)

  • Fields: Link to Being, Link to Attribute

Attribute (A personal attribute of a being, or some sort of attribute of any other phenomenon in the corpus — 'attribute' table)

  • Fields: Name, Description

Attribution in Text (A link between the attribution of a feature to a being, or similar phenomenon, and a source of that attribution — currently implemented using the 'rdf' table)

  • Fields: Link to Being's Attribute, Link to Chapter/etc., Description

Attribute in Text (A non-specific link between a series of attributes not attached to a particular named being, such as in a picture or analogue or text sharing mythological motifs — currently implemented using the 'rdf' table)

  • Fields: Link to Attribute, Link to Chapter/etc., Description

Action in Text (A link between a narrative element and a particular instance of that action in a text or image — 'action' table)

  • Fields: Link to Chapter/etc., Link to Narrative Element , Classification as linked by identified name, visual features or generic motif, Description

Narrative Element (Ordered part of a repeated narrative which can be used to link to individual versions of the narrative — 'narrativeelement' table)

  • Fields: Link to Narrative , Label, Description, Relative ordering

Narrative (A narrative that may occur in more than one text or image — 'narrative' table)

  • Fields: Name, Summary, Context, Commentary

Participant (An identified being in a narrative — 'participant' table)

  • Fields: Link to Narrative, Link to Being, Other information

Works cited

Abram, Christopher. 2011. Myths of the Pagan North. London: Continuum.

Boot, Peter. 2006. Decoding Emblem Semantics. Literary and Linguistic Computing 21 (Suppl. Issue), 15-27.

Bradley, John and Harold Short. 2005. Texts into Databases: The Evolving Field of New-style Prosopography. Literary and Linguistic Computing 20 (Suppl. issue), 3-24.

Brink, Stefan. 2007. How Uniform Was the Old Norse Religion? In Learning and Understanding in the Old Norse World: Essays in Honour of Margaret Clunies Ross. Eds. Judy Quinn, Kate Heslop and Tarrin Wills, 105‑36. Medieval Texts and Cultures of Northern Europe 18. Turnhout: Brepols.

CIDOC-CRM: Le Boeuf, Patrick, Martin Doerr, Christian Emil Ore, Stephen Stead. 2013. Definition of the CIDOC Conceptual Reference Model. CIDOC CRM Special Interest Group.

Eide, Øyvind. 2008. The Exhibition Problem. A Real-life Example with a Suggested Solution. Literary and Linguistic Computing 23, 27-37.

Jónsson, Finnur, ed. 1912-15. Den norsk-islandske skjaldedigtning. B: Rettet tekst. Copenhagen: Villadsen & Christensen. Rpt. 1973. Copenhagen: Rosenkilde & Bagger.

Faulkes, Anthony, ed. 2005. Edda: Prologue and Gylfaginning. 2nd edn. London: Viking Society for Northern Research.

Meulengracht Sørensen, Preben. 1986. Thor's Fishing Expedition. In Words and Objects: Towards a Dialogue between Archaeology and History of Religion, ed. Gro Steinsland, 257-78. Oslo: Norwegian University Press and Instituttet for sammenlignende kulturforskning.

Moltke, Erik. 1985. Runes and their Origin: Denmark and Elsewhere. Trans. Peter Foote. Copenhagen: The National Museum of Denmark.

Motz, Lotte. 1993. Vǫlundr. In Medieval Scandinavia: An Encyclopedia, ed. Phillip Pulsiano et al. New York: Garland, 713.

Neckel, Gustav and Hans Kuhn (1899), eds. 1983. Edda: Die Lieder des Codex Regius nebst verwandten Denkmälern. 2 vols. I: Text. 5th edn. Heidelberg: Winter.

Ólafur Sveinsson, Einar, ed. 1934. Laxdœla saga. Íslenzk fornrit 5. Reykjavík: Hið íslenzka fornritafélag.

Ore, Christian-Emil and Øyvind Eide. 2009. TEI and cultural heritage ontologies: Exchange of information? Literary and Linguistic Computing 24, 161-172.

Pasin, Michele and John Bradley. 2013. Factoid-based prosopography and computer ontologies: Towards an integrated approach. Literary and Linguistic Computing (Advance Access, 29 June 2013).

Peel, Christine (ed.). 1999. Guta saga. London: Viking Society.

Price, Neil. 2002. The Viking Way: Religion and War in Late Iron Age Scandinavia. Aun 31. Uppsala: Dept. of Archaeology and Ancient History.

Simek, Rudolf. 1993. Dictionary of Northern Mythology. Trans. Angela Hall. Cambridge: D. S. Brewer.

Tacitus, Germania: trans. M. Hutton. 1970. Tacitus I Cambridge, MA: Harvard University Press.

TEI P5: The TEI Consortium. 2014. TEI P5: Guidelines for Electronic Text Encoding and Interchange. Edited by Lou Burnard and Syd Bauman. Charlotteville: Text Encoding Initiative Consortium.

von See, Klaus, Beatrice La Farge, Eve Picard, Ilona Priebe and Katja Schulz. 1997. Kommentar zu den Liedern der Edda. Vol. 2: Götterlieder. Heidelberg: Winter.

Wagner, Norbert. 1977. (Hercules) Magusanus. Bonner Jahrbücher 177, 417-22.

Wills, Tarrin. 2002. Electronic Editing of the Skaldic Corpus. Skandinavistik 32, 25-30.

Wills, Tarrin. 2013. Relational Data Modelling of Textual Corpora: The Skaldic Project and its Extensions. Literary and Linguistic Computing (Advance Access, 11 August 2013).

Zöllner-Weber, Amélie. 2011. Text encoding and ontology—enlarging an ontology by semi-automatic generated instances. Literary and Linguistic Computing 26, 365-70.