THE WATER DRAGON AND THE SNAKE WITCH Two Vendel Period Picture Stones from Gotland, Sweden

The style and iconography of two well-known picture stones are re-analysed. The Hablingbo Havor II picture stone shows a motif that occurs frequently in Gotlandic art from the Vendel Period onwards: the “Water Dragon”. It is suggested that this relates to an ideological connection between the dragon and the sea, where the sea is the dragon that ferries ships to distant shores. This is reflected not only in picture stones, but in Viking Age art in general. The iconography of När Smiss III (the “Snake Witch”) has been interpreted in a variety of ways, but special consideration is given to Peel’s (1999) suggestion that it relates closely to the Vitastjärna myth from the 13th-century Guta Saga. The artistic style of the zoomorphs on both stones (Style II) is typically dated to the Vendel Period. It is suggested that Sune Lindqvist’s insistence that the stones date from before AD 600 comes from a long-standing debate with Nils Åberg over the date and context of the east mound at Uppsala, and by association, the date of the artistic style found on Hablingbo Havor II and När Smiss III. This debate has been resolved in favour of Åberg’s interpretation. These two picture stones represent an artistic tradition that should be dated conservatively from the beginning of the 5th century AD to the middle of the 7th century AD.

The style and iconography of two well-known picture stones are re-analysed. The Hablingbo Havor II picture stone shows a motif that occurs frequently in Gotlandic art from the Vendel Period onwards: the "Water Dragon". It is suggested that this relates to an ideological connection between the dragon and the sea, where the sea is the dragon that ferries ships to distant shores. This is reflected not only in picture stones, but in Viking Age art in general. The iconography of När Smiss III (the "Snake Witch") has been interpreted in a variety of ways, but special consideration is given to Peel's (1999) suggestion that it relates closely to the Vitastjärna myth from the 13th-century Guta Saga. The artistic style of the zoomorphs on both stones (Style II) is typically dated to the Vendel Period. It is suggested that Sune Lindqvist's insistence that the stones date from before AD 600 comes from a long-standing debate with Nils Åberg over the date and context of the east mound at Uppsala, and by association, the date of the artistic style found on Hablingbo Havor II and När Smiss III. This debate has been resolved in favour of Åberg's interpretation. These two picture stones represent an artistic tradition that should be dated conservatively from the beginning of the 5th century AD to the middle of the 7th century AD.

THE WATER DRAGON AND THE SNAKE WITCH
The picture stones of Gotland are the best-preserved examples of post-Roman era Nordic artwork. Indeed, the picture stones and the images they convey today are closely bound with modern concepts of Scandinavian antiquity and social identity. Viking Age picture stones with their rich content, much of which is interpreted through the lens of Norse mythology, are best represented in the scholarly literature. Two of the most striking examples of pre-Viking Age style are found on Hablingbo Havor II and När Smiss III. Though the morphology and symbolic structure of these two picture stones are similar to the oldest examples, their artistic style is significantly different.
It is well known that Sune Lindqvist painted many picture stones from the 1920s onward as a means to study and communicate his interpretations (Lindqvist 1941(Lindqvist , 1942. Today there are non-invasive techniques that preserve the integrity of the stone and its carved images (Åhfeldt 2012, 2013). Fortunately, these new methods can also be employed on stones that are already painted, potentially allowing us to re-interpret the delicately preserved carvings. But Hablingbo Havor II and När Smiss III have very well-preserved markings. A visual inspection quickly reveals that his painting matches the visual appearance of the carvings quite accurately. The images we can see today on Hablingbo Havor II and När Smiss III are what style historians call "Style II" following a typology first published in 1904 by Bernhard Salin. Salin's style system is still in use today virtually unadulterated in form. Most of the other Type A stones show "Style I" artwork, thus it is very easy to detect how Hablingbo Havor II and När Smiss III differ from the earlier picture stones.
A quick review of the literature on Style II, which was originally designed to categorize metal artefacts recovered throughout the "Germanic" culture area, reveals that in Sweden this style is associated with the Vendel period (c. AD 540-790). Late Style II, such as we see on the Hablingbo Havor II and När Smiss stones, appears well into this period (c. 630-690). Why, then, did Lindqvist insist that Type A stones dated from AD 400 to 600, and not to AD 700? A further investigation of previous interpretations of the picture stones, and an historical analysis of how Lindqvist came to his conclusions, allows for a better understanding of his rationale.

THE TYPE A FORM
Hablingbo Havor II and När Smiss III are referred to as "Type A" picture stones using the typology developed by Lindqvist in his two-volume monograph Gotlands Bildsteine, published in 1941 and 1942. The Type A Form, or Abschnitt A in the original, is the earliest of Lindqvist's 5 major types of picture stones (Lindqvist 1941:22-35). There are three main compositional elements that nearly always appear on complete Type A stones: (1) A large round centrepiece, often dominating the composition; (2) pairs of facing figures, human or animal; and (3) a continuous border of interlace, lines, or shapes at left, right and top (Figure 1). Additionally a number of them have oared vessels, with or without oarsmen, but never a sail. The stones range in height from slightly less than a metre to as large as 3.3 m tall. They are always sub-rectangular, 2 to 4 times higher than they are wide. Their sides are slightly concave and their top is slightly convex; overall this shape has been described as like the upturned blade of an axe (Nylén & Lamm 1988:11). Lindqvist identified numerous morphological subtypes, including the so-called "grave orbs", to which the previous description does not apply. All of them, he argued, could be placed in the Migration Period or early Vendel Period, which he dated from AD 400 to 600 (Lindqvist 1941:108-115).

HABLINGBO HAVOR II
Hablingbo Havor II is one of the most widely recognized picture stones, its central motif, the quatrefoil knot, having been co-opted as the official symbol marking culturally significant sites on maps and roadways throughout Sweden (Figure 2). It has all the basic elements of a Type A stone: (1) an "axe-blade" shape, (2) a border framing the composition, (3) a circular emblem in the upper frame, and (4) facing figures in the lower frame. The major deviations are (1) the circular emblem is a looping knot rather than a geometric swirl, (2) the linear borders are natural- istic zoomorphs exhibiting a wave-like structure, rather than the usual interlace pattern, and (3) the twins are opening their mouths towards a third "eye" like their own. Although there is no boat in the lower frame, the stone exhibits maritime symbolism in the form of dragon-headed waves along the lateral edges.
Lindqvist did not attempt to interpret the meaning of the stone, but he did note that its style was quite different from other Type A stones (Lindqvist 1941:112-115). He considered it a masterpiece without a local equivalent. Indeed, he found the stone's uniqueness important enough to devote roughly half the typespace for the dating of Type A to account for its unique style. He insisted, however, that although the artistic elements of the stone had strong affinities with Salin's Style II (which Salin argued was a 7th-century form) that it dated to the 6th century (Salin 1904, Lindqvist 1926. Two key elements anchor the stylistic interpretation: the quatrefoil knot in the middle of the composition, and the border of continuous waves with the crests of the waves transformed into "Water Dragons" that form the lateral edges. The quatrefoil loop is a common symbol in Germanic art, but it only appears in this well-developed form in the Vendel Period. For Gotland, Nerman (1975) shows a number of examples of earlier quatrefoil knots that he estimates date to between AD 600 and 650 ( Figure 3).  The top border shows the late Style II dragon motif. Here we can clearly see the four appendages of the beast, though its head form clearly links it with the wave-crest creatures. This upper symbolic representation portends the complex interwoven gripping dragons seen on the later Type E picture stones, in Style III, and in the Urnes style which belongs to the Early Medieval period.
The lateral borders of dragon-headed wave crests are also representative of late Style II. Åberg (1946) first recognized that the continuous waves of animal heads first evolved from earlier forms of interlace patterns ( Figure 4). Like Nerman, Åberg placed late Style II in the 7th century, but believed the motif disappeared after this time. Note that the left and right borders are slightly different in form. The left border shows dragon heads with forelocks flowing backwards, conveying a sense of motion. These same forelocks are seen on the opposing figures at the centre, and the creature forming the top border. In contrast, the wave-crest heads forming the right border are more gracile and without forelocks. In this way the lateral borders are reminiscent of the opposing geometric figures in earlier Type A stones that show slight variations (e.g. Sanda Kyrka IV). Given that the Hablingbo Havor II stone already has two twin opposing figures at the centre, I suspect the differences between the lateral borders convey another meaning significant to the artist.

NÄR SMISS III
När Smiss III, commonly called "the Snake Witch," is one of the more enigmatic of the picture stones (Nylén & Lamm 1988). While it has the same basic structure of a Type A stone, the images are strikingly unique ( Figure 5). The elements it shares with other Type A stones are (1) the "axe-blade" shape, (2) a border framing the composition, (3) a circular The Water Dragon and the Snake Witch emblem in the upper frame, and (4) facing "twins" in the lower frame. The major differences are: (1) the circular emblem is a zoomorphic triskele rather than a geometric disc, (2) the facing twins are naturalistic zoomorphs rather than geometric-abstract ones, (3) the stone lacks a lower division containing a boat, and (4) a human figure in a birthing position faces the viewer, grasping the snake-like zoomorphs.
The main diagnostic style element of När Smiss III is the triskele occupying the wheel locus on other Type A stones. That particular style The När Smiss III triskele differs from the ornamental perforated discs in that these examples are completely symmetrical and have identical animal heads. However, stylistically the form of the När Smiss III triskele most closely resembles Figure 6c-e. Unlike the ornamental perforated discs that are symmetrical, the När Smiss III triskele shows three distinctive heads interpreted as a boar, a wolf or serpent, and a bird respectively (Lindqvist 1955;Peel 1999:19).
To some extent, all picture stones are unique, and the similarities mentioned above are enough for most scholars, including Lindqvist, to assign it to the Type A picture stone group. In the initial publication of the När Smiss III stone, Lindqvist (1955) searched in vain for a connection with other regional art forms. He found only vague parallels with Celtic and Germanic art, finding whimsically that the image of the woman in particular had closer affinities to a Minoan snake goddess or a Parisian dancer! Ultimately, he placed the stone with the other Type A stones purely on its structural and morphological similarities.

DATING THE STONES
Assigning a date to individual picture stones is somewhat problematic, as few of them have been discovered in their original context. In fact, while a few in situ stone fragments have been suggested as possible Type A "roots," no definitive Type A picture stones have been found in their original use context. The stones themselves are not directly datable, so the practice has been to cross-date them. If a picture stone had preserved artistic elements, Lindqvist compared them with artistic styles commonly preserved on gold and bronze brooches, bracteates, belt fittings, etc., found at Iron Age archaeological sites throughout the northern region (Lindqvist 1941(Lindqvist , 1942. The artistic styles used by Lindqvist and other mid-century archaeologists were originally defined and described by Salin (1904) for the northern region, and correlated to Gotland-specific archaeological materials by Nerman (1919).
Lindqvist noted the obvious stylistic differences between most of the Type A stones and the Hablingbo Havor II stone, but considered them to be close in age (Lindqvist 1941:114-115). Using Salin's (1904) typology of Iron Age styles, Lindquist saw most of the Type A stones as having more similarities to Style I, whereas the Hablingbo Havor II motifs were clearly late Style II. Based in part on correlating the great barrows of Old Uppsala to lineages outlined in Ynglingatal, Lindqvist preferred a relatively early date for the introduction of Style II, which was represented by fragmentary evidence in the eastern mound (Lindqvist 1917(Lindqvist , 1949. Although Salin (1904) put Style II in the 7th century, Lindqvist argued that Styles I and II were "sisters", attributing both to the 5th and 6th centuries 1 (Lindqvist 1922, Lindqvist 1926. Thus, it was quite natural for Lindqvist to place Hablingbo Havor II with the other Type A stones in the 5th and 6th centuries (i.e. AD 400-600; Lindqvist 1941).
Lindqvist's contemporary, Nils Åberg strongly disagreed. He argued on strong archaeological grounds that Salin's initial assessment that 1 "SL hat jedoch seinerseits geltend zu machen versucht, dass Stil II nicht eine Tochter, sondern eher eine, wenn auch später gereifte, Schwester von Stil I ist." Translated by the author: "Sune Lindqvist contends that Style II is not a daughter, but rather, though more mature, a sister of Style I" (Lindquist 1941:115).

Frederic B. Pearl
Style I preceded Style II was correct. Furthermore, he argued that the change was gradual and occurred at the beginning of the 7th century, 100 years later than posited by Lindqvist (Åberg 1922, 1924, 1947). Åberg's chronology was not given consideration by Lindqvist in his picture stone monographs of 1941 and 1942. If it had, the chronology of Type A stones would likely have been extended an additional 100 years (see Gaimster 1998, Ljunkvist 2008or Nielsen 2012 for fascinating accounts of Lindqvist vs. Åberg). Both Åberg and Lindqvist recognized a distinct Continental influence in the development of Style II, and predicted its origins could be found there. Today most scholars agree with Åberg's argument that Style II occurs in the Vendel Period (c. AD 540 to 790), which developed from Style I, occurring in the Migration Period (c. AD 450 to 540) (Høilund Jørgensen & Jørgensen 1997;Rundqvist 2010;Ljunkvist 2008).
The main counterpoint to this argument is the quatrefoil loop on Hablingbo Havor II. A nearly identical symbol occurs on the double bracteate from Lyngby, east Jutland (Figure 3). The period of bracteate manufacture in southern Scandinavia generally precedes the Vendel period, although in Gotland this is not necessarily the case (Axboe 1999). However, this particular bracteate (IK 297) is given a rather early position in the seriation, indicating that it would have been produced in the late 5th or early 6th century (Axboe 1999;Pesch 2007:90-93, 438). However, this does not change the fact that Style II animal symbolism on Hablingbo Havor II occurs a century later.
Excavations at Valsgärde in the 1920s produced copious quantities of Style II artefacts. In what was the first major publication of the artefact catalogues from Valsgärde, Arwidsson (1942) showed successive Vendel Styles. At the end of his long career, Nerman produced the most detailed volume of Style II metal artefacts from Gotland (Nerman 1975). He numbered the phases VII:1 through VII:5 (the "VII" standing in for Montelius phase VII). He suggested an age of AD 650 to 700 for VII:3. However, these ages have been refined by subsequent scholarship by correlating the VII:3 type to Vendel grave VII, which dates to AD 600 to 670 (Nørgard Jørgensen 1992; Jørgensen & Jørgensen 1997).
The VII:3 stylistic elements on Hablingbo Havor II and När Smiss III place them in the heart of the Vendel sequence. The Hablingbo Havor II and När Smiss III stones should therefore also be placed in the mid 7th century on stylistic grounds, making them contemporary with other VII:3 artefact types (cf. Arrhenius and Holmqvist 1960).

INTERPRETATIONS
Assigning meaning to symbols and icons is considered one of the most difficult and subjective pursuits attempted by the archaeologist (Robb 1998). In terms of picture stone iconography, as noted by Burström (1996), they carry different meanings to different people in different times and contexts. When we ask what the symbolism on a stone means, are we asking what the artist wanted to convey? I am always suspicious of symbolic interpretations when done without any cultural reference; nonetheless, they obviously contain some kind of structured message.
For a theoretical foundation I am inclined to invoke Leslie Webster's (2003) discussion on Anglo-Saxon styles from the same period: My argument is that we are dealing here with a particular kind of visual literacy, with its own enduring grammar and vocabularies, which is wholly attuned to the reading of complex artifacts from at least the fifth century to the end of the ninth. Anglo-Saxon style, in its widest sense, constitutes a form of visual language developed in the migration period, which possesses the scope and recombinant flexibility to shape and control the expression of complex ideas in very particular ways. (Webster 2003:12-13) This theoretical viewpoint is sensible. The artwork on the Type A stones appears in a highly structured way, consistently showing the same artistic elements but with variations on the artistic theme. This strongly suggests that a coordinated "meaning" lay behind not only the icons, but the placement of the iconography itself. The Vendel period Hablingbo Havor II and När Smiss III stones hold vivid symbols of "Water Dragons" and a "Snake Witch," but we have very limited historical tradition with which to interpret them. The Icelandic sagas and heroic myths have proved fertile grounds for interpretations of the Viking Age stones (e.g. Helmbrecht 2012; Ney 2012), but have been less helpful for understanding the earlier ones.

The Snake Witch
In Christine Peel's 1999 translation of and commentary on the 13thcentury Gotlandic Guta Saga she introduced an interpretation of När Smiss III that is worthy of consideration. In her notes on the translation Peel (1999:18-19) draws a comparison of Vitastjärna's dream to the iconography of När Smiss III, without definitely stating that the images are a representation of the story. Guta Saga was first written down in the mid 13th century, but presumably existed in oral forms before that time. While much of the story deals with the emergence of Christianity in the 11th century, the first hundred stanzas or so deal with an older mythic past. The stories of Þieluar, his son Hafþi, his wife Vitastjärna and their three children were probably distant memories in the time in which the later saga is set: This same Þieluar had a son named Hafþi, and Hafþi's wife was called Huitastierna [Vitastjärna]. These two were the first to settle in Gotland. The first night that they slept together, she dreamed a dream. It was just as if three snakes were coiled together within her womb, and it seemed to her as though they crawled out of her lap. Translation by Peel (1999:3) The implication of Peel's direct comparison suggests the possibility that even though the artist of När Smiss III and the author of Guta Saga are separated by five centuries they were influenced by the same, or similar, local legends. This rationale is similarly applied by other scholars to use Icelandic sagas and heroic myths to help interpret Viking Age stones (e.g. Helmbrecht 2012; Ney 2012).
Peel does not elaborate on the details of the comparison, which presumably are that (a) the figure in the lower third is that of Vitastjärna giving birth to the snakes as per her prophetic dream from Guta Saga, and (b) though she only holds two serpents, her legs splayed wide suggest she is still in the process of giving birth. Arrhenius and Holmqvist (1960) assert that the composition and style of the snake-witch bears a strong likeness to the Daniel in the Lion's Den representation on the purse from Sutton Hoo. They also tentatively support the idea of placing När Smiss III in the 7th century. An alternate parallel is a small human figure between two snakes on the Horn of Gallehus (5th-century Denmark). Hauck (1970:281-282) draws this comparison (among others), and suggests that in these cases the human figure is an aspect of Odin. As noted previously, Lindqvist (1955) found only vague similarities to other Germanic artwork from the 5th and 6th centuries.
If Peel's interpretation is considered, however, it opens the possibility that the artwork on the remainder of the stone may also relate to that myth. For example, in this context the serpentine triskele in the upper frame might represent the three boys: Guti, Graipr, and Gunfiaun. What was the artist trying to convey by giving the birthing mother the distinctive headgear or hairstyle? Undoubtedly this indicates her status, but which status? Is it her social role, her social position, marital status, ethnicity, or another social indicator? For the most part Lindqvist and his successors satisfied themselves with understanding the direct iconic meaning. I believe the answers to these more complex questions are encoded in its symbolism, but can only be interpreted in the context of gender roles and status in Vendel Period Gotland. These would be productive areas for further research.
Another interesting sidebar to Peel's note is that she interprets the three zoomorphs of the triskele as a bird of prey, a boar and a serpent or dragon (Peel 1999:19). Other scholars including Lindqvist (1955) and Nielsen ( , 2012 clearly see a wolf rather than a serpent in the design. Although the eagle-boar-wolf motif arguably appears repeatedly in Continental archaeology (Werner 1963), both Lindqvist and Nielsen consider the motif rare in Scandinavia. Nielsen (Nielsen 1999:333) acknowledges that over time in Scandinavian Style II the wolf symbol is gradually displaced, first by a horse, and then by a dragon-like creature. Considering the Vendel Period date for this picture stone, it seems likely that the topmost zoomorph on När Smiss III is a serpent.

The Water Dragon
The När Smiss III snake twins bear a strong resemblance to the central twins motif on Hablingbo Havor II, suggesting the possibility that the orb between the snakes on the latter has the same symbolic meaning as the human figure on the former. The twins motif is one of the most important diagnostic features of the Type A style (Andrén 2006(Andrén :54-55, 2014. Andrén further contends that the opposed twin motif appearing on Type A picture stones is a representation of the sunrise and sunset, possibly offspring of the sky.
The two snakes on Hablingbo Havor II resemble the Water Dragons composing the left border, yet their size and shape are juvenile by comparison. Following the fertility theme of När Smiss III suggests the possibility that the left and right borders represent male and female aspects of the Water Dragon, and that the central twins represent their offspring.
Åberg notes that after making a late appearance on the Type A stones the Water Dragon motif disappeared completely (Åberg 1946). He must have been principally referring to the wave design appearing on Vendel Period metal artefacts, especially strap-fittings, because on Viking Age picture stones the dragon motif reappears on ships' figureheads and as the diamond-headed wave crests beneath the dragon boats on some picture stones (Figure 7). Indeed, the dragon motif is the only motif that occurs in one form or another in each era of picture stones. Though it is difficult to say what the dragon symbolized in ancient times, it is clear that by the Vendel Period the dragon was linked very closely to the sea, either as a benefactor protecting and aiding ships along their way, or as an ever-present danger on which the ships must travel, or perhaps both. Frederic B. Pearl Figure 7. Type C/D Viking Age stone from Hejnum, Riddare. The clearest example that the dragon symbolizes the sea, carrying the Viking ship to its destination (Lindqvist 1941: fig. 79).

The Water Dragon and the Snake Witch
This coincides with the mythic legend of Jormungandr, the Midgard Serpent, that lives in the sea and is so long that it encircles the world (Gylfaginning §33). Myrberg suggests that the spiral wheel so common in Type A style depicts a concept of the Earth (or Midgard) circumscribed by the dragon, and that the water, snake and dragon are conceptual siblings (Myrberg 2006). Hablingbo Havor II would seem to provide the perfect example of that concept.
While I am reluctant to accept a direct parallel between the Jormungandr of Gylfaginning and the Water Dragon motif on the picture stones as they are separated so distantly in space and time, I agree that they both reflect a pan-Nordic association between the dragon and the sea. Since maritime symbolism conspicuously reappears through time across Scandinavia's landscape, we can surmise that a relationship with the sea was of paramount importance to the ancient peoples of the north and deserves to have special consideration when interpreting archaeological landscapes (Westerdahl 1992;Cook 2001;Skoglund 2008;Wehlin 2010;Westerdahl 2011).
The concept of the dragon was so embedded in Northern culture that it is no surprise that it strongly influenced medieval art and ideology. I would argue that the westernized Christian concept of the dragon owes more to its Germanic/Scandinavian roots than its Mesopotamian/ Mediterranean ones.

IMPLICATIONS AND CONCLUSIONS
Leaving the vagaries of symbolic interpretation behind, the more grounded conclusions of this analysis are that Type A picture stones were most definitely being made from (at least) the 5th century well into the 7th. Lindqvist knew that the Hablingbo Havor II and När Smiss III stones were both late Style II, but he held fast to the notion that Style I and II were overlapping and fully within the Migration Period. Today, Åbergs's hypothesis that puts Style II in the Vendel period, chronologically after Style I, is well accepted (Stenberger 1964;Nerman 1975;Arrhenius 1980;Ljungkvist 2005Ljungkvist , 2008. A lengthened chronology of Type A stones has a few other interpretive implications. Notably, if Type A stones are produced almost up until the Type C/D stones are produced (c. 750-790), where do Type B stones fit in the sequence? One possibility is that they are indicative of an independent or regional artistic school contemporary with the one that produced the later Type A stones (ironically recalling Lindqvist's "sisters" analogy between Style I and II). Alternatively, late Type A stones may reflect a renaissance of Type A style -a re-emergence after some period of dormancy. I cannot prefer one of those hypotheses over the other at this stage of my research, but it deserves future consideration.
What about an earlier date for Type A stones? This hypothesis, suggested by Manneke (1984) and again by Andrén (2012Andrén ( , 2014, is based on two important facts: (1) the similarity between the "solar disc" on all Type A picture stones and Roman tombstones from south-west Europe, and (2) the preponderance of Roman imports such as coins, metals, glass, and ceramics that begin occurring on Gotland in the first few centuries AD. Both authors also note the occurrence of undecorated picture stone "roots" first encountered at Vaskinde, and then at Grötlingbo, which Manneke argues date from the first to third centuries AD (Manneke 1984:86-88). Manneke did not provide evidence that this discovery was from a Type A picture stone, but rather only offered it as a possibility. Alternatively these roots may be examples of one of the many untyped and undecorated burial markers found occasionally in larger grave-fields on Gotland (Per Widerström, personal communication). Andrén also points out similarities between the roundel on Stenkyrka Kyrka I (Lindqvist 1942:115) and the so-called "swastika-fibulas" from the third and fourth centuries (Andrén 2014:121-122).
Lindqvist placed the Type A stones in the Migration Period based on solid stylistic grounds. He recognized the similarity of the sun emblem on both Type A picture stones and the Roman headstones from Spain and France. However, he rejected the notion that solar representations on Type A stones were borrowed directly from the Roman examples. Lindqvist considered the entire composition, especially the use of highly structured linear motifs of zoomorphic and geometric ornaments that characterize Migration Period styles. He recognized the Roman connection, but suggested, I believe correctly, that those elements were brought in via Germanic connections, especially from Continental sources (Lindqvist 1941:114-115). Swanström (1993) has linked the solar motifs on type A picture stones to post-Roman, Visigothic sculpture in Spain. Roman artefacts in Scandinavia reflect the massive and complex trade networks that had been established in the Bronze Age (or earlier) for moving goods back and forth across Europe, not direct merchandising by Roman ships (Lindqvist 1941;Holmqvist 1970;Bitner-Wróblewska 1991;Swanström 1993;Kristiansen 1998). The stylistic evidence that Type A stones are a Migration Period phenomenon is so strong, however, that if there is an earlier tradition of picture stones or stone burial markers they probably constitute a heretofore undeclared type. While Type A stones may turn out to be older than AD 400, more archaeological evidence is needed to strengthen the argument.

The Water Dragon and the Snake Witch
One explanation why the iconography of Hablingbo Havor II and När Smiss III is strikingly distinctive from earlier Type A picture stones is that cultural rules regarding picture stone content changed during the 7th century, and broader mythical content began to be incorporated into what had been a relatively stable artistic tradition. This tradition eventually evolved into the Viking Age picture stones (Types C/D) that have been routinely interpreted in the context of Norse myths and legendary stories as preserved in medieval Icelandic literature. The continuity of the Water Dragon through time supports the prevailing notion that underlying the cultural changes occurring in the 7th and 8th centuries there was substantial cultural continuity as well.