TRADITIONS AND INNOVATION IN FINE METALWORK IN THE MIDDLE DANUBE REGION IN THE SECOND HALF OF THE 5 TH AND THE EARLY 6 TH CENTURIES A.D. ONCE MORE ON THE ‘SZARVAS’ BROOCH GROUP: NEW FINDS, NEW HYPOTHESES

: The aim of this paper is to examine a group of brooches whose numbers have been increasing in recent years to determine their origins, their relationship to each other and their role in the fine metalwork, goldsmith practice of the period. These brooches and pairs of brooches were found in ten sites scattered across a large geographic area (Szarvas, La-Rue-Saint-Pierre, Bernhardsthal, Uppåkra, Narona, Hemmingen, ‘Italy’, Collegno, Domoszló, Nagyvárad). The artefacts share common features that can aid in determining the areas of production for objects within the group. We can confidently date them to the second half of the 5 th and the early 6 th centuries A.D. and examine their role in the development of the so-called Thuringian-type brooches. Furthermore, they allow us to investigate changes in female attire and shed light on the relationships between the Middle Danube region and Southern Sweden (Skåne). niello footplate footplate side runs two filled niello. In two five double semi-circles can be headplate is semi-circular, and upper edge, where it meets the two side bands of the is curved. (‘fanlike’) ribs fill the headplate, each is with transverse engravings (pseudo-filigree). At the pinnacle of headplate laterally, lower are three settings with garnets one is a two sides

the centre protruding band on the part of the footplate adjacent to the bow. The other part of the footplate terminates in an animal head with accentuated eyes. -The bow is arched and proportioned lengthwise. A protruding central band runs the entire length of the bow. It is flanked by two longitudinal fields. -The headplate is relatively small and is decorated with stone inlays. Three flush settings with stone inlay (stone inlaid settings in cast recess, 'pseudo-cloisonné) can be observed on all of the brooches. Naturally, many other characteristics of form and technique can be observed in this group of brooches, even if most of them can only be examined through photographs and descriptions. Interestingly, the ten brooches and brooch pairs that belong in this analysis can further be divided into smaller 'sub-groups' of two or four based on more or less similar features not shared with the other brooches. These are: 1. Szarvas and La Rue-Saint-Pierre ( Fig. 1.1-2) 2. Berhardsthal and Uppåkra (Fig 1.3-4) 3. Narona, Hemmingen, 'Italy', Collegno (Fig. 3.5,6,7,8); and within this group: a) Narona and Hemmingen appear closely related, as do: b) Collegno and 'Italy' 4. Nagyvárad-Guttman brick factory (henceforth Nagyvárad) and Domoszló-Víztároló (henceforth Domoszló) (Fig. 3.9, 10) The similarities and differences found in the sub-groups

The brooches of Szarvas and La Rue-Saint-Pierre
The features shared by the brooches of Szarvas ( Fig. 1.1) and La Rue-Saint-Pierre ( Fig. 1.2) have already been discussed in detail in earlier analyses. Here, I will refer only to those elements that link these two while distinguishing them from the other members of the group: -A 'Y'-shaped strip runs along the entire end of the foot plate, across the head of the animal. The strip contains a single row of punched triangles inlaid with niello. The slanted, oval-shaped eyes are accentuated with stone inlays. -The central, protruding band on the bow has two rows of punched decoration: alternating triangles, their vertexes directed inwards, inlaid with niello. The two outer bands of the bow each have a deeply engraved zigzag pattern running lengthwise. -The remains of the pin mechanism are visible on the reverse: three small lugs were cast together with the plate as a continuous unit; these lugs were once furnished with a spring that later rusted and was lost.
On the La Rue-Saint-Pierre brooch, a spring axis can be seen that also appears to be made of silver and connected to the two outer lugs. The spring may have been coiled onto this axis. No trace of anything similar can be seen on the Szarvas brooch. -Size, length: the Szarvas brooch is 4.95 cm and the La Rue-Saint-Pierre is approximately 5 cm (this measurement is based on the photograph).

The brooches of Bernhardsthal and Uppåkra
The artefacts from Bernhardsthal ( Fig. 1.4) and Uppåkra (Fig. 1.3) are the best analogies to the brooches discussed above (for a brief description see the earlier publications 8 ). These two brooches have characteristics that are exclusive to them and others they share with the two discussed above.
-The proportions and form of the animal head at the end of the footplate are similar in all four brooches.
However: the stone inlays are missing from the slanted, almond-shaped eyes in the Bernhardsthal and Uppåkra brooches. A description of the latter reveals that 'a dark mass, but not niello' was applied to the narrow Y-shaped line on the head. There is no mention of anything like this on the Bernhardsthal brooch. Another common feature is the long, rectangular flush setting that occupies the centre of the other part of the footplate, which is flanked by a series of cross-length, inclined, engraved ribs. The bow of the brooches is proportioned lengthways, but only the Uppåkra brooch is described as having an 'inlay of a dark mass that differs from the usual niello' in a longitudinal line in the middle strip.
The Bernhardsthal brooch has a zigzag line in the middle strip and the two exterior bands also have an engraved zigzag decoration. -If we examine the headplates, we find the two brooches correspond almost perfectly with respect to the settings and inlays. In all four brooches, the basic form is the same: in the centre is a cast setting that tapers downwards into a point, while the two outer settings have curved edges. However, in the Szarvas and La Rue-Saint-Pierre brooches, the central setting has a pentagonal shape and the lower walls of the two outer settings are also curved; in the other two brooches, the central recess is rhombus-shaped and the outer settings have straight edges along the bottom. Moreover, in the photographs, the walls of the settings visible in the Bernhardsthal and Uppåkra brooches are wider and flatter than in the other two. In all four, there is a narrow field on each side, beneath the panel of inlaid stones: in the Szarvas and La Rue-Saint-Pierre artefacts these fields are also curved with peaked ends and each has a lengthwise groove; in the Uppåkra, this feature is similar but more accentuated, and in the Bernhardsthal brooch it has a rectangular shape articulated by two grooves. The description of the Bernhardsthal brooch refers to 'glass inlays' throughout, but for this question, it would be better to rely on an expert examination of the piece. -On the reverse of the Uppåkra brooch, the spring and pin mechanism are mounted on two lugs. Since the spring and pin are completely intact, it is clear they were made of silver. The side drawing of the Bernhardsthal brooch allows us to conclude that it perhaps had a different spring and pin mechanism: the spring may have been mounted on a single lug, but the description also mentions that, based on the remains, the spring was made of iron. -The Bernhardsthal and Uppåkra brooches was shorter (3.7 and 4.1 cm) than the Szarvas specimen and its companion (4.95-5 cm). However, the photograph and drawing show they have wider and stubbier bows and footplates than the Szarvas and La Rue-Saint-Pierre brooches. Observations of the details of the four brooches reveal that the most convincing similarities are found between the Szarvas and La Rue-Saint-Pierre brooches. While the correspondences between the head and footplates of the other two, smaller brooches and their resemblance to the Szarvas and La Rue-Saint-Pierre items are striking, these smaller specimens also have features that distinguish them not only from the other pair but, in some respects, from each other; in other words, they possess distinct features in terms of both form and technical execution.
3. Brooches from Narona, 'Italy', Collegno and Hemmingen More distant analogies to the brooches above come from Narona 9 (Fig. 2.4), 'Italy' 10 ( Fig. 3.7) and Collegno 11 (Fig. 3.8). Here I need to reiterate that the Narona brooches have a striking analogy from Hemmingen 12 (Fig. 2.2). However, this latter example, despite corresponding almost completely in form, only partially meets the criteria of this brooch group: one of the most important characteristics, the stone inlaid in a rectangular setting on the foot, is missing. Nevertheless, the Hemmingen brooch pair needs to be included in the analyses of this group. The similarities and differences between the brooches and brooch pairs from these four sites and those above are the following: -The almond-shaped eyes on the animal head of the Narona brooch pair are not slanted, but rather are positioned vertically, without stone inlays; pseudo-filigree decoration accentuates the eyes. The animal head on the 'Italian' brooch tapers strongly, but the Y-shaped strip running alongside and between the eyes is adorned with rows of punched triangles; the animal's mouth is triangular and is accentuated by a stone inlay. The Y-shaped strip on the animal head of the Hemmingen brooch pair is similar to that of the Szarvas and La Rue-Saint-Pierre artefacts, but is simpler in execution. The animal heads on the footplates of the brooch pair from Collegno are similar in design to those of Narona and Hemmingen, but they have a setting at the end, also seen on the 'Italian' artefact; however, the settings on the Collegno pair are semi-circular, while the setting on the 'Italian' brooch is triangular. -All of the brooches in this group have bows adorned with two rows of alternating punched triangles, but while the two outer fields on the 'Italian', Hemmingen and Collegno artefacts are empty, the Narona brooches have a series of punched, double semi-circles. 9 buLJević 1999. 10 sALin 1904, 194. 467;Åberg 1922, 109. -The headplates on all the brooches are half circles, but on the Narona and Hemmingen specimens, a curved field above the main decorated surface, similar to that on the Szarvas, La Rue Saint-Pierre and Uppåkra brooches, can be seen. The headplates on all of the artefacts in this group have chip-carved designs; on the Narona and Hemmingen specimens, parallel grooves are arranged in a radial pattern, while the 'Italian' and Collegno specimens are adorned with two symmetrical scrolls. The headplates of these latter two brooches also have borders adorned with rows of triangular niello inlays. What unites these brooches and clearly distinguishes them from the previous group are the three oval (semi-circular in the Collegno pair) flush settings with stone inlay cast along the edges of each headplate. A conspicuous characteristic of the Narona brooches is the pseudo-filigree, engraved line decoration. Cross ribbing can be seen in the upper part of the footplate in the Hemmingen pair, but these brooches have no setting with stone inlay. -The Narona brooches are 5.8 cm long, the 'Italian' brooch is 5.7 cm (based on the drawing), and the Hemmingen brooches are 5.8 cm: thus, their lengths correspond almost perfectly. The Collegno brooches are slightly longer (6.6 cm). As for their fasteners, on the Narona brooches, all we know is there was an iron pin mechanism on the reverse. On the Hemmingen brooches, we know that an iron spiral was mounted on a spring axis supported by double lugs located 'near to each other'. In summary: the common features (chiefly the shape of the headplate and its three separate oval settings with stone inlays) of these brooches and brooch pairs from four different sites suggest they have a closer relationship to each other than to the other brooches in the larger circle. However dissimilarities can also be observed within the subgroup (varying patterns on the headplates, somewhat different uses of punched decoration, and, in the Hemmingen pair, the absence of the rectangular setting on the footplate). A conspicuous similarity can be seen in engravings (scroll designs) on the headplates of the 'Italian' and Collegno brooches; yet the shaping of the animal head in each specimen is different. Here too we should note their almost identical sizes, although the Collegno specimens are slightly longer than the others.
4. The brooches of Domoszló and Nagyvárad The similarities and differences between the remaining brooches, a pair from Domoszló 13 (Fig. 2.1) and one from Nagyvárad 14 (Fig. 2.3): -The 'turtle-like', round animal heads at the end of the footplates are cruder than in the previous subgroup of brooches; the eyes, too, are round and inlaid with red stones. The decorations above the eyes on the Domoszló specimens are perhaps meant to be ears. -Decoration in the two bands flanking the central strip on the bow can only be observed in the Nagyvárad brooch: a row of four S-shaped spirals. -The headplates have similar structures: in both pairs, the half-moon-shaped panel of the semi-circular head has three settings, each inlaid with red stone. The headplate in the Nagyvárad brooch has three animal-head knobs that resemble the head at the end of its footplate. The headplates of the Domoszló pair have five knobs: round settings at the end of shafts radiating from the headplate. -The drawing suggests the Domoszló brooches are much thinner, sheet-like. The Nagyvárad specimen is 5.8 cm long (in his book, R. Harhoiu mistakenly states the length as 8 cm), while the Domoszló specimens are 5.4-5.5 cm. The width of their headplates, including the knobs, is similar: approximately 3.5 cm. It is obvious from the above information that the brooches from these two sites are more similar to each other than to the others in the larger brooch group (semi-circular headplates with three flush settings inlaid with stone and three or five knobs, either animal shaped or on shafts projecting from the plate). 13 bónA-nAgy 2002, 27, Taf. 4.7-8. 14 hAMPeL 1905csALLány 1961, 109, Taf. CCVIII.4, hArhoiu 1998, Taf. C/3a-c.

Divisions within the brooch group
By surveying the brooches and brooch pairs from the ten sites, we have been able to conclude the following about their formal and technical characteristics: The brooches from Szarvas and La Rue-Saint-Pierre stand out from the group not only because of the high quality of their design and execution, but also because of the degree to which the minute details of these two artefacts correspond. Minor differences can only be seen on their reverses: the La Rue-Saint-Pierre brooch has a silver (?) spring axis integrally cast with the double outer spring-holding lugs. This feature is missing on the Szarvas brooch, and we can find no explanation for this technical difference. 15 Nevertheless, the spring and pin mechanisms are identical. This method was rarely used on other brooches: three (rather than one or two) lugs supported the spring: double lugs were positioned laterally on the headplate, while the third was placed between them and below. 16 The uniformity between the two brooches extends to the size of the punched triangles and their placement in the central band of the bow and along the protruding Y-shaped strip on the animal's head. The number of cross-grooves on both sides of the stone inlay on the footplate is also identical, and the same is true for the engraved zigzag lines on each side of the bow's surface. The dimensions of the two brooches likewise show the greatest uniformity. Furthermore, we have every reason to posit that these two brooches were not only cast at the same time, but presumably the same hand added (afterwards) the minute details and decorations. Of course, the two have minor differences, but these are a result of their 'later life'. Both brooches bear traces of wear and their stones are damaged, cracked. The cracking of the bow in the Szarvas specimen, however, may have occurred after the brooch was acquired by the museum. 17 The Szarvas and La Rue Saint-Pierre brooches, which serve as the starting point for this analysis, are most closely related to another two specimens: the brooches of Bernhardsthal and Uppåkra. These latter two, however, do not correspond exactly to the other pair or to each other. Although the two pairs bear a strong resemblance with respect to important typological details, as a whole they do not display uniformity. The discrepancies are already apparent in the dimensions: compared to the Szarvas and La Rue-Saint-Pierre pair, the Bernhardsthal and Uppåkra brooches are shorter and stubbier. The settings with stone inlays on the headplates of the Bernhardsthal and Uppåkra brooches, however, match in the smallest details (the shape and size of the settings are uniform), while the footplate and the arch and proportions of the bow also correspond. Their differences, though, are also striking: for example, the decorations on the bows are not the same and the upper edges of their headplates also have different shapes. The greatest variation can be found in their spring and pin mechanisms. While their dimensions are different (the Bernhardsthal brooch is 3.7 cm long, and the Uppåkra brooch is 4.1 cm), both are shorter than all the other brooches in the group. From this we can perhaps conclude that the Bernhardsthal and Uppåkra brooches display the kind of typological similarities to the Szarvas and La Rue-Saint-Pierre specimens that indicate they may have been products of the same master or workshop. The Bernhardsthal and Uppåkra brooches, however, may have been made from different casting mould(s) on different occasions, perhaps for different customers and were crafted with different details.
These four brooches pairs are more distantly related in terms of form and technique to the Narona, 'Italian', Hemmingen and Collegno artefacts; conversely the members of this latter subgroup bear typological similarities to each other, although none of them can be considered identical. The headplates in this subgroup are not 'tulipshaped', but rather semi-circular. Where it was customary at the time to have three knobs, here instead there are three more or less regular, oval (or semi-circular) settings with stone inlays studding the edges of the headplates. Nevertheless, the 'common features' described in the previous subgroup also apply here, linking the two: in each brooch the footplate is the same width as the bow and has a long, rectangular setting inlaid with a stone. Only the 15 The spring was perhaps wrapped around an iron spring axis, and perhaps this and the spring are what deteriorated. See also horváTh 2012, 199-200. 16 The spring chord connecting one end of the coil spring to the other was probably threaded through the smaller, third lug. In the western Merovingian region (for example, of the Franks), only mechanisms with one or two lugs are differentiated: this too is an important typological difference. The specimens with a single lug were earlier and were already made during the proto-Merovingian period. The specimen with double lugs however, only appeared in the early Merovingian period. The latter structural solution ensured greater stability of the pin. Meanwhile, the springs of the smaller brooches required only a single centrally positioned lug: generally they did not need wider springs. Koch 1998, 513 17 B. TóTh 1999 It can be seen with other brooch pairs too that the two members of the pair are abraded in different places and to different degrees depending on how they were worn. For similar examples, see Koch 1998, 510: the mode of dress and placement of the brooch may explain the differences in some cases. Nevertheless, the author also considers it possible, in cases where the difference is great, that one brooch was a copy of the earlier brooch, which was no longer in use.
Hemmingen brooches lack this feature, but like the 'Italian' and Collegno specimens, this pair, too, has cross grooves on the footplate. However, the formation of the animal head shows marked differences not only between the two subgroups but also among the members of the present subgroup in question. At first glance, the Narona and Hemmingen brooches bear the greatest similarity, but in their case, too, there are technological differences: their animal heads are shaped differently and the Hemmingen brooches do not have a row of punched decoration in the two outer bands of the bow. In these artefacts, unearthed from four different sites, workshop traditions identical to those of the previous subgroup are blended with others that originated elsewhere: numerous formal-technical features correspond to those of brooches and brooch types from outside the group (see below). Nevertheless, it is still astonishing that not only are the Narona and Hemmingen specimens the same size (length: 5.8 cm), but so too is the 'Italian' specimen (5.7 cm, based on the drawing).
The brooches of Domoszló and Nagyvárad are the most distantly related with respect to typology, but they possess the minimal number of elements to belong to this circle of brooches and furthermore share some similar characteristics. Their most distinctive common feature is their curved, half-moon panel with stones inlaid flush settings in cast recesses on a semi-circular headplate. The animal heads on these brooches are rough and 'turtle-like'. A generally widespread characteristic of the period can also be seen: the knobs along the edge of the headplate. The three knobs on the Nagyvárad brooch are shaped like animal heads, while the five on the Domoszló pair resemble 'the tuft of a peacock'. 18 Another fashionable innovation is the row of S-shaped spiral engravings on each side of the bow: this suggests that the Nagyvárad specimen was the product of a workshop or goldsmith who incorporated numerous other (local?) elements in addition to the few seen in this subgroup. The Domoszló brooches, on the other hand, certainly appear to be a lower-quality copies of the Nagyvárad-type: although their plates are cast from silver, the former ones appear too thin for chip-carved decorations; in other words, decorations that required a certain thickness in the base material could not be applied. Furthermore, the animal heads of the Domoszló brooches are shoddily executed. Nevertheless, the brooch pair is characterized by the uniqueness of the headplate design ('peacock tufts') and the richness of the stone inlays: they display the most abundant use of stone inlays, not only on the headplate surface, but along the edges, too. Furthermore, the spring and pin mechanism can be clearly observed in the drawing from the original publication. 19 It appears that the spring was mounted on double lugs and the catch plate was conspicuously long. In the case of the Nagyvárad brooch we can only conjecture that the reverse had just a single lug upon which the spring was mounted. Nevertheless, these brooches are of similar length: the Nagyvárad specimen is 5.8 cm while the Domoszló brooches are 5.4 and 5.5 cm.
All of this indicates that the Nagyvárad brooch and its Domoszló copies may not have been produced by the same master or workshop as the others.
Here we should address the question raised in the introduction: how should we view the ten brooches and brooch pairs in terms of typology? I consider the terms 'brooch group' and 'brooch circle' the most suitable, as these objects, from a typological standpoint, cannot be categorized as only one 'type', since they have significant differences in addition to their many identical features. The various subgroups felt the effects of other, contemporary impulses. A. Koch described a similar phenomenon as a 'group of shapes' (Formengruppe); that is, a small group of brooches that presented a rather wide range of variability. This is characteristic primarily of the proto-or early Merovingian period: objects made in small numbers without uniform appearances. In contrast, true 'brooch types' display a combination of recurring formal elements and were made in large numbers. 20

Typological connections to the brooch group
When more than a decade and a half earlier I examined the origin, development and connections of the Szarvas brooch type, I determined that the starting point for its evolution was the Krefeld-type brooch, a conclusion that was consistent with the generally accepted views in the professional literature. This brooch type was widely distributed (in the Alemannic, Frankish and Thuringian regions), and in the second half of the 5 th century displayed 18 nAgy 2007, 23. 19 bónA-nAgy 2002, Taf. 4.7-8. 20 Koch 1998, 13. In this sense, Eisleben-Stössen (to which, for example, the Ószőny specimen also belongs) is a true brooch type. a wide variety of forms. 21 We are acquainted with this rather broad group chiefly through H. Kühn's brooch typology, 22 but well before his work, N. Åberg had defined a brooch type that was only partially identical to Kühn's (narrow footplate terminating in an animal head), in which he placed the La Rue-Saint-Pierre and 'Italian' brooches. 23 He believed the former brooch was made in a 'Central European style'. Today we believe that the Åberg and Kühn brooch types, because of their 'expansive' nature, can only be viewed as distant antecedents to the brooch circle summarised in this study.
Also during my search for formal-typological precursors to the Szarvas brooch, I came to the opinion that the cicada form widespread in the Carpathian Basin in the 5 th century also played a role in the development of the head plate. My conclusion was further corroborated in recent times. In J. Tejral's overview of the pre-Lombardian period, the author discusses the Szarvas, La Rue-Saint-Pierre and Bernhardsthal brooches and refers to the possible role of the cicada form, but noted that the bird head 'pointing downwards' ('Bügelfibel mit nach unten beissenden Tierköpfen') seen on some Thuringian brooches may be behind this strongly stylized form. Moreover, for comparison, the author includes in the tables a similarly formed, mostly silver cast cicada type (Ringelsdorf type), noting that the cicada was originally a Danube-region, proto-Merovingian motif. 24 In a later work, the author returns to these conclusions and declares the cicada motif to be the origin of the three oval settings on the edge of the headplates of the Narona and Hemmingen brooches as well. 25 Most recently he reiterated this opinion, again citing the Szarvas, La Rue Saint-Pierre and Bernhardsthal brooches, which he considers a super-regional ('überregional') brooch form. He refers to his conjecture that the 'much-loved cicada motif' played a role in the development of the later Central-German brooches with pincer-like headplates ('zangenartige Spiralplatte'). 26 In connection with a pair of cicada brooches from grave no. 26 in Altenerding, H. Losert and A. Pleterski determined that the cicada motif had a role in the development of a series of brooch headplates. 27 It is worth surveying again the various opinions arrived at by researchers -decades earlier or recentlyabout the possible classifications of the brooches analysed here, their typological connections, and perhaps grouping. After making these comparisons, we then need to ask where we stand. Maybe the simplest and easiest way to proceed is to sum up the views of the various scholars in order of the analyses. D. Csallány in his volume on the Gepids posited that the 'narrow-bodied brooches with stone inlays' form a coherent group: he included the Nagyvárad and Ószőny brooches along with the Szarvas specimen; 28 thus he too had observed their typological similarities. In I. Bóna's brief summary, he presents the Szarvas brooch as an 'unarticulated Gepidic brooch' and did not give any analogies. 29 After analysing the trade and foreign relations of the Gepids, K. Mesterházy determined that 'the Szarvas and Nagyvárad brooches with footplates of the same width and unarticulated are Frankish in origin ….' Thus he concurred with Kühn's grouping. 30 Most recently, in his assessment of the Gepidic finds of Gallia, M. Kazanski referred to the Szarvas and Bernhardsthal specimens, as well as the La 21 B. TóTh 1999, 265-266. D. Quast even remarked that the foundations of this type were too broad from a typological perspective. QuAsT 1993, 63. 22 Kühn 1940, 74. 23 Åberg 1922. 'Fibeln mit schmalen Tierkopffuss und Kopfplatte'. The brooch dubbed La Rue-Saint-Pierre by A. Koch was also considered to have come from an 'unknown site' by N. Åberg. His source was obviously the same as mine: the information found in the National Archaeology Museum of Saint-Germain-en-Laye. 24  28 csALLány 1961, 268. More specifically, the sites of discovery of four other brooches are listed in addition to Szarvas: Nagyvárad, in the area of Gyöngyös, Szomód and ʻHungary' (Ungarn). With the exception of Nagyvárad, there is total confusion. The specimen from 'Gyöngyös' is in fact the footplate fragment of the Szarvas brooch, which D. Csallány presents here twice under two different sites of discovery. The object from Szomód does not even appear in the entire volume. The one from 'Hungary' -based on the detailed description -is identical to the brooch from Ószőny (the wrong table number is given: it should be Taf. CCVII.4 and not Taf. CCVII.6). On the above, see B. TóTh 1999, 262-264. And as usually happens, the errors and misunderstandings were carried forward. H. F. Müller referred to the Hemmingen pair as an analogy to the brooch from ʻSzomad' (ʻnorth of Tata'), which was 'not published' and thanked J. Werner for the related information. However, the inventory number he refers to in the Hungarian National Musem is 94/1903.21. This is identical to the inventory number of the Szarvas brooch; thus, here too, the citing of 'Szomad' as the site of discovery is incorrect. H. F. Müller  Rue-Saint-Pierre brooch, as belonging to the 'Thuringian tradition'. He was probably influenced by the most recent scholarly consensus (J. Tejral, J. Bemmann). 31 A. Koch collected the Merovingian-period bow brooches from the region of western France. He placed the La Rue-Saint-Pierre brooch, found in Picardia, (and also the Szarvas brooch from the 'Eastern Danube region') among the 'Thuringian bow brooches', in group B of the 'bow brooches with animal heads pointing downwards' (Bügelfibeln mit nach unten beissenden Tierköpfen). 32 In his opinion the La Rue-Saint-Pierre brooch corresponded 'exactly' to the Szarvas artefact, and the two were without question products of the 'same workshop and of the same type'. He also refers to the 'Italian' brooch as a similar specimen. Furthermore, he determined that exact analogies to the two brooches are unknown in the Central German Thuringian region (yet he still classified them in a group originating from this area: 'bow brooches with animal heads pointing downward'); therefore he felt 'only with reservations can we consider these bird head brooches to have derived from Thuringia', and perhaps they only reflect a Thuringian influence. In addition, citing the opinion of I. Bóna, he remarked that the two brooches could, with some justification, be considered Gepidic, 33 as the use of red semi-precious stones and the polychrome style itself was very widespread in Gepidic areas. Nevertheless, he considered another alternative: 'we cannot rule out the possibility that both came from central Germany or Bohemia and travelled in the waves of migration (durch Bevölkerungsbewegungen) to two geographically very disparate regions.' 34 In summary, he did not take a definite stance on his suggestions for the possible origins and places of fabrication of the brooches.
Turning our attention to the other two brooches also possessing 'cicada-like' headplates, we find the following references. The publishers of the Bernhardsthal brooch (S. Allerbauer, F. Jedlicka) described it only briefly and did not analyse it. Concerning the Uppåkra brooch, J. M. Sundberg established that the 'design' of the headplates was most similar to that of the Bernhardsthal brooch, and the two differed in only the minutest details. He described the headplate as 'flower-like' and does not acknowledge or accept the 'cicada' analogy. He does observe, however, the engraved sections ('wings') beneath the panel of stone inlays on the headplate in both specimens, noting the variations and typological differences. Relying on his solid knowledge of the literature, J. M. Sundberg concludes that the Uppåkra brooch (too) belongs to the bird head type, which may be Thuringian (Central German) in origin, but the dispersal of this type transcended regional boundaries ('superregional'). Moreover, the areas under 'Thuringian influence'-Bohemia, Moravia and Lower Austria -can also be posited as possible places where these brooches were created. Today it is impossible to specify the locations more precisely. The find from Uppåkra, however, is without question a 'foreign' import, as the use of the term 'continental' in the title of his article indicates. Thus J. M. Sundberg also analyses the four 'cicada-head' brooches as part of the broader (bird-head) brooch type and not as a unit onto itself.
The similarities between the Narona, Hemmingen, 'Italian' and Collegno brooches are obvious. H. F. Müller wrote about the Hemmingen cemetery approximately forty years ago. In his thorough study, he cited only the brooches from Šaratice, Nagyvárad and Szarvas (mistakenly referred to as Szomad, see above) among the analogies. The 'fanlike' engraving, the almond-shaped settings with garnet inlays and niello pattern are presented as typological features. 35 In describing the Narona artefact, Z. Buljević mentioned as good analogies the 'Italian' brooch and the specimen in the National Archaeology Museum of Saint-Germain-en-Laye (now known as the La Rue-Saint-Pierre brooch), which he also learned of from Åberg's analysis. 36 A. Uglešić, without embarking on a detailed analysis, emphasized the Narona brooches' connection to the 'Gyöngyös', Szarvas and Nagyvárad brooches. 37 G. Tica -familiar with the Collegno brooch pair -posits that the Narona, Hemmingen and Collegno objects all belonged to the same group. 38 31 KAzAnsKi 2010, 103. 32 The author seemed to struggle with uncertainty in determining the material used to make the brooch, or at least his remarks in various places in the book suggest this: 'Supposedly gold, but probably silver gilt', and elsewhere, 'bronze, perhaps silver gilt'. However, an analysis of the material is not necessary, as just a glance at the object in the museum tells us it is silver gilt. Koch 1998, 396, 645. 33 '… it cannot be decided whether it was a Gepidic from the Danube region, but there is a certain likelihood.' 34 Koch 1998, 398. 35 MüLLer 1976, 109. He referred to the Soponya brooch too as a further analogy. 36 buLJević 1999, 240-241. He highlighted the oval, garnet inlaid flush settings, the ribbed footplate and the rectangular setting as typological characteristics. For further analogies, he referred to the Nagyvárad and Eisleben specimens too. In addition, with respect to the 'Italian' brooch, he noted an analogy to the triangular cell at the tip of the animal head: the Brochon brooch (France Discovered in 1876, the Nagyvárad brooch has appeared many times in archaeological publications that highlighted its connections to various other artefacts. 39 As we have seen above, D. Csallány classified it among the 'narrow bodied' brooches 'with stone inlays'. In his monograph presenting the artefacts from the migration period in 'Romania', R. Harhoiu managed to find a place for the Nagyvárad example among the brooches with 'three knobs and bows equal in width to the footplates': without providing details, he referred only to Thuringia, Pannonia, and Yugoslavia as locations where similar sorts of brooches with stone inlaid flush settings have been found. 40 As the Domoszló brooch pair was only published in 2002, naturally the number of references to it are smaller. Margit Nagy discussed it in her book on animal depictions in the Middle Danube region. The author found the forerunner of the five round stone settings projecting from the headplate in a 'pair of gold fittings shaped like peacock heads with stone inlays, originating from an unknown site, which had circulated through the art market before arriving in Washington'. 41 The five cells on the Domoszló headplate therefore must be simplified depictions of peacock tufts. M. Nagy also mentioned the 'Italian' brooch, with its triangular inlaid stone at the end of the footplate, as a good analogy to the Domoszló example, and remarked that more good analogies to this form of stone inlay can be found in the Carpathian Basin.

AN EXAMINATION OF THE 'BROADER ARCHAEOLOGICAL CONTEXT'
The detailed formal-typological examination above and the lessons that can be learned from the literature on these brooches help to define this circle of ten brooches and brooch pairs as a set of subgroups, each consisting of two or four more closely related brooches. This brooch circle is naturally deeply embedded, in terms of typology as well, in its historic, narrower-broader archaeological context. If we look at the broader context of this circle -in accordance with its geographical distribution -two regions require examination: the Carpathian Basin and those areas that have yielded artefacts that are either Thuringian or 'belonging to the Thuringian tradition'.

a) Related brooches from the Middle Danube region
Several specimens of brooches with footplates equal in width to the bows, terminating in animal heads, and semi-circular headplates are known from the various regions of the Carpathian Basin. Three knobs and engraved designs are generally found on the headplates. One example was analysed in detail by M. Nagy when the finds from the cemetery of Hódmezővásárhely-Kishomok were published. The brooch was unearthed from grave no. 105: the very stylised animal head at the end of the footplate terminates in a round flush setting containing a garnet inlay. The footplate has cross ribbing and the bow is longitudinally proportioned. The shape of the headplate is somewhere between a triangle and semicircle and is decorated with radial grooves; the three knobs are positioned along its edge. 42 The author refers to Kühn's Krefeld type as a typological antecedent widespread in the Alemannic-Frankish region, although good analogies to this form from Lombardian, Pannonian Gothic and Gepidic territories have been found. The brooch from Kishomok is a bronze copy of the basic form (but with semi-precious stone inlays) and perhaps acquired its defects during the casting process. In all likelihood, it was made by Gepids in the Tisza region. It is part of a 'small brooch series' in which the animal head was decorated with or replaced by stone inlays. in Szentes-Berekhát, grave no. 4 in Kormadin-Jakovo, grave no. 3 in Bakodpuszta, grave 18 in Hács-Béndekpuszta, grave no. 12 in Šaratice 12: all have three knobs. Grave no. 105 in Kishomok was disturbed. In the grave of the approximately 50-year-old woman, an unadorned jug, a double-sided antler comb, amber beads, a spindle whorl, an iron knife, fragments of broken glass, and coffin clasps were found in addition to the brooch. 43 For comparison: true Krefeld-type brooches are also known from the area of the Carpathian Basin; for example, the specimen from Tác/Gorsium: schiLLing 2011, 385-387, Abb. 1.1.
We know of a similar brooch from grave no. 46 in Viminacium-Burdelj. It was found on the left side of the chest and was the only artefact from this grave. It is so strongly reminiscent of the Kishomok example (which is also made of bronze with a stylized animal head, but without a stone inlay) that it may have been produced in the same (local?) workshop. 44 The authors refer to the brooches of Šaratice and Schletz as possible direct prototypes of these specimens. They too consider the Krefeld type to be a more distant antecedent.
The cast-engraved radial decoration was widespread in a brooch group whose specimens have turned up most frequently in sites in the Carpathian Basin: they are known as the Bakodpuszta type. This type also has a threeknobbed head plate but the footplate is different: it is rhombus-shaped and likewise has simple line decorations. 45 The silver-gilt specimens in this group are between 4.8 and 5.4 cm long: in other words they are similar in size to a portion of the Szarvas-type brooches.
Given all this, it is not surprising that the radial pattern on the headplates of the Narona and Hemmingen brooches can be considered a typological element. 46 The three settings studding the edge of the headplate, however, obviously fulfil the role of the three knobs on the other brooches while preserving something of the three, but differently-shaped settings on the Szarvas specimen. We have concrete proof of this: on three brooches (Szarvas, La Rue-Saint-Pierre and Uppåkra), a chip-carved band appears beneath the panel of settings with inlaid stones on each side of the bow. The bands have a 'wing-like' form that terminates in a downward facing tip, just as the cicada wing-ends do: this kind of formal solution can be seen in the corresponding parts on the Narona and Hemmingen headplates. In the headplates of these latter two examples, the Bakodpuszta-type radial decoration meets a 'relic' of the cicada form. This cannot be seen on the 'Italian' and Collegno brooches: on these, the lower edge of the headplate is straight and the triangular niello border decoration and engraved scrolls on the interior panel reflect another kind of brooch tradition (but one that is also from the Danube region). Only the use of three settings (but here oval or, in the case of the Collegno pair, perhaps semi-circular?) in place of knobs is reminiscent of the Narona and Hemmingen headplates. These different kinds of workshop traditions were not necessarily far-removed from the Carpathian Basin: in fact, there are so many brooches with engraved scrolls or volutes bordered by niello triangles that it would be impossible to list them all here. The Narona and Hemmingen brooches were thus the products of the same workshop, but for some reason the rectangular setting was omitted from the footplate of the latter, or perhaps the Hemmingen brooches were copies of the Narona pair. The two brooch pairs are not very distant from one another chronologically either. It is not surprising that H. F. Müller had referred emphatically to the Hemmingen specimens' analogies within the Carpathian Basin. 47 The brooches discovered in 'Italy' and the Collegno thus may have been made in the Carpathian Basin (in Pannonia?), but in different workshops and by different goldsmiths.
It has already been discussed above that while the Nagyvárad and Domoszló brooches are related, but their differences are greater; moreover their typological distance from the other members of the group is also the greatest. I have already mentioned the heightened role played by polychromy in these two works as well as the presence of the animal style in the Domoszló pair in the form of the 'peacock tuft' motif. These are all features that may have been characteristic of workshops active in the Carpathian Basin in the 5 th century A.D. Moreover, M. Nagy was justified in noting that the engraved S-shaped spirals along both sides of the Nagyvárad brooch's bow are typical elements in the Gepidic repertoire of motifs in the Tisza region and thus point to a goldsmith workshop in the Great Hungarian Plain (Alföld). 48 Therefore, it is not impossible that the two artefacts were made somewhere in the eastern half of the Carpathian Basin and that the craftsmen were aware of other examples within this circle of brooches. 44 ivAnišević- KAzAnsKi-MAsTyKovA 2006, 16, 154, Pl. 7. 45 Mészáros 2015, 63-66, Figs 1-3. In addition to the specimens from the Carpathian Basin, it is striking that others have been found in Moravia (Kyjov) and the Czech Basin (Horny Ksely, Czech Republic): these are made of bronze. 46 The 'radial' pattern elsewhere is also described as 'fanlike'. The origin of the pattern can be found, perhaps, on a larger, finely executed brooch from Belgrád-Zimony (Zemun

b) Prospects of the brooch group having 'Thuringian' origins 49
A. Koch clearly placed the La Rue Saint-Pierre specimen (and its analogy, the Szarvas brooch) in the B type of 'bow brooches with animal heads pointing downwards'. 50 It says a lot that according to the author this small group consisted of just two objects: the La Rue-Saint Pierre brooch and another from Villey-Saint-Etienne. 51 The latter brooch, however, is a true Thuringian specimen with 'animal head pointing downwards', whose bow and footplate do not even resemble those of the La Rue-Saint-Pierre brooch: the Villey brooch has a wide, articulated bow and engraved, rhombus-shaped footplate. It is thus questionable whether this is the best analogy. According to A. Koch, the centre of gravity for this brooch group is in Central Germany and this is probably where the Thuringian specimen published here originated. In any case, it says a lot that the author determined the Szarvas brooch was thus far the La Rue-Saint-Pierre specimen's only exact match from the 'eastern Danube region' and was clearly made in the 'same workshop' and using the 'same model'. Finally, he finishes his argument by stating that the 'the Danube region may justifiably be identified as the brooch's place of origin, but the possibility cannot be excluded that both brooches originally came from Central Germany or Bohemia and were swept away to two such distant regions in the migration of peoples.' 52 J. Tejral dealt with this subset of brooches three times in the past decade or more. The first time, he referred to the Lower-Austrian Bernhardsthal brooch among the Lombard finds in the area north of the Danube. Familiar with the Szarvas and La Rue-Saint-Pierre brooches, he asserted that it was difficult to determine the origin of this type because of the widely dispersed sites. In any case, he felt this group of three had no exact analogy in Thuringia. He linked the Nagyvárad brooch to this group too, emphasizing that it was an early specimen. Finally, he stated that these few brooches were not special Thuringian forms but rather the products of a workshop or workshops whose location could not be precisely determined but had presumably operated in the Danube region. The cicada-like modelling of the headplates point to this, but these might be 'downward-pointing birds': in that case, it might be the result of a 'wide-ranging exchange of ideas' (weiträumiger Ideenaustausch) among the various Merovingian cultural regions (Kulturbereich). In any case, this impulse may have contributed to the development of Thuringian brooch types. 53 In the author's next study, which deals with early-Merovingian development in the Middle Danube region, again this subset of brooches arises. He reiterates his assertions about the three brooches in his previous work but expands the circle of analogies (for example, Naumburg, grave no. 26 in Altenerding-Klettham, grave no. 51 in Mörstadt): these all contain the popular 'cicada motif'. He too discerns this motif in the Narona and Hemmingen brooches. The finds that show the development of the Thuringian type, in his opinion, were otherwise strongly impacted by influences from the 'southeast': in addition to several other characteristics, many, originally eastern German brooch forms were also borrowed and further developed. 54 An exchange of models and ideas between the Merovingian territories and the Danube region was characteristic of the second half of the 5 th century; this led to the rich variation in bow brooch types and the adoption of foreign brooch forms in certain regions. Most recently, J. Tejral again presented these brooches together, as the proto-Merovingian and Duna-region prototypes/antecedents of Thuringian brooches. 55 49 For an earlier summary, see B. TóTh 1999, 266. 50 Koch 1998, type VII.2.2. Bügelfibel der Variante mit nach unten beissenden Tierköpfen (Gruppe B), Taf. 50. 16. It is confusing that while only the two sites referred to above are mentioned in the text, he includes the analogies to the type in the list of finds (Fundliste 25B). The circle of analogies is wide: from Frankish, Alemannic, and Thuringian regions, but several Hungarian specimens also appear ('Szolnok-Szanda', 'Mohács'), as well as objects from Kranj and Cividale. 51 Koch 1998, Taf. 50,15. 52 Ibid., 398. He referred to I. Bóna's opinion that the almandine or garnet inlay, polychrome style was frequently used on bow brooches and this extensive use was characteristic among the Gepids. 53 TeJrAL 2002, 331. The author emphasized that the connection between this brooch type and the Eisleben/Stössen type cannot be ignored. M. Kazanski  J. Bemmann undertook a thorough overview of 5 th -century Central Germany in which he deals at length with the various brooch types occurring in this region at the time. One of these was the circle of 'brooches with downward-pointing bird heads': he considers this a relatively uniform type with many variations. An 'unusual variation' were the small brooches that had headplates with stone inlaid settings, bows with zigzag engravings, footplates with cross ribbing and a rectangular, inlaid stone. This description in the article was based on the Bernhardsthal brooch: the other brooches with bird heads pointing downwards that appear on Fig. 3 are indeed similar in terms of one or another typological feature, but the Bernhardsthal specimen clearly does not 'fit in' with this group as a whole. The list of elements defining this type includes variations and special forms (Sonderform). One of these is the 'cloisonnéd headplate variation' (Variante mit cloisonnierter Kopfplatte): here the author refers to the Szarvas and La Rue-Saint-Pierre specimens as well as the Bernhardsthal brooch. 56 The author does not discuss further these brooches, since his focus was elsewhere: his aim was an archaeological examination of the 5 th -century finds from Central Germany and the people who migrated to Moravia and Lower-Austria, later called Lombards. In any case, in a later work, J. Bemmann determined that such brooch types labelled Thuringian, such as the birdhead brooches, occurred in large concentrations in Central Germany. However the numbers outside this region are significant enough to presume that these were not made in the Central German region, but rather elsewhere in the Thuringian style. 57

SUMMARY: THE ORIGIN AND TYPOLOGICAL CONNECTIONS OF THE BROOCH GROUP
Perhaps we can summarize the information above as follows: this brooch group, which rests on more or less identical traditions and is represented in this analysis by ten brooches and brooch pairs, may indeed have derived from the earlier Krefeld basic form, but was strongly influenced by other cultural traditions too: the richness of the stone inlays may have its roots in polychrome art, while the proliferation of animal depictions (cicada and peacock) during the period also certainly played a role in the group's development. In this looser group, perhaps the 'cicada' headplates were the 'trend-setters' (subgroups 1-2). We might think this in part because of their superb quality, reflected in the more complicated method of inlaying stones in cast recesses and the occasional use of chip carving and niello decoration. Furthermore, as expressed in the opinions above, the other two subgroups (3-4), although no longer with 'cicada' headplates, still preserve numerous typological features of the subgroups 1-2.
Most of the prominent features -as we have seen above -characterise the Szarvas and La Rue-Saint-Pierre brooches. These two objects can still be viewed as a pair: many tiny details persuasively indicate that they were made at the same time by the same goldsmith. In theory we can posit that their maker created not just one but many such brooch pairs at the same time, but we have scarcely any archaeological evidence of this from the period. 58 At the time, 'mass produced' items of this quality (silver-gilt cast brooches) were not made; good quality brooch pairs that are very similar in type still differ in smaller or larger details. 59 Naturally, examinations of the material used (the provenance of the semi-precious stone and the composition of the metal) could prove important in answering this question, but even these might not lead to a definitive result: information in the literature shows that even brooches made in pairs do not always contain exactly the same composition of metals. 60 Nevertheless, there is no question that these brooches were made and used in pairs (see below): thus it is unlikely that the Szarvas and La Rue-Saint-Pierre brooches had been placed in separate graves, despite cropping up later in separate and remote archaeological collections. Perhaps we are not far from the truth if we state that these two brooches -based on 56 beMMAnn 2008, 176-177, Abb. 30.8, Liste 9: 'Bügelfibeln der Variante mit nach aussen blickenden Vogelköpfen '. beMMAnn 2008, 207-208. 57 beMMAnn 2009 An exception may be, for example, the two brooch pairs found in Hungary, in the cemetery discovered in Balatonszemes-Szemesi berek: the silver-gilt, chip-carved brooch pairs unearthed from the child's grave no. 268 and the adult woman's grave no. 269 were described by the authors as being 'completely identical '. bondár et al. 2007, 130, ill. 118 and 121. Most recently on these brooches: Miháczi-PáLfi 2018b, 84: '…besonders die Stücke aus Grab 268 und 269 zeigen eine auffallende Ähnlichkeit miteinander'. 59 But even the brooches made in pairs showed minor differences. I. Bóna established with respect to the Gepids that 'even those brooch pairs cast from the same mould showed variations in size and details '. bónA 1978, 140. 60 The reason for the difference in the precious metal content could be that the silver coins used as raw material also had variations in their metal content. See Koch 1998, 504. If the goldsmith had only one casting mould, then in all likelihood, he made the two members of a brooch pair in succession using different material. roTh 1986, 50.
formal-typological features -were made in one workshop by one goldsmith somewhere in the 'Middle Danube region', although the exact location cannot be determined. Rather than 'fellow group members', these two may have been a source of inspiration for the brooches referred to as Thuringian (bird heads pointed outwards or downwards) and linked to them in the literature. Thus, it is not surprising that no examples from this small brooch circle have been found in any regions linked to the Thuringians, although the sites containing these finds are scattered over a wide geographic area.
The Bernhardsthal and Uppåkra brooches may have been produced in the same workshop, by the same master, if we accept their 'shared' feature, a headplate reminiscent of a cicada, as sufficient proof. Moreover, the cicada headplates of these two brooches are almost identical in terms of execution and have smaller dimensions and proportions than the headplates of the other members of this subgroup; this makes it likely that they originated in the same workshop, although their smaller-larger technical and formal differences suggest they were made on different occasions (by different goldsmiths?) Here, we need to digress and mention another possible, simpler specimen belonging to this group (?). This brooch also came from Bernhardsthal and was found in the immediate vicinity of the specimen discussed at greater length in this study ( , Parzelle 1953: the headplate of this cast bronze specimen is also cicada-like and appears to be a (local?) copy of the silver-gilt brooch with stone inlay. Its surface is unadorned, the bow and footplate are of equal width and the foot terminates in a sphere, obviously a replacement for the animal head. The picture suggests it was about 4 cm long. These finds clearly came from former graves of a (destroyed?) cemetery. 61 In any case, this brooch was published by J. Tejral, and classified with the earlier 'Thuringian' type objects among the Lombard finds from Moravia. 62 The above four brooches (Szarvas, La Rue-Saint-Pierre, Bernhardsthal, Uppåkra) certainly belong to the same brooch type (Fig. 3.1,2,3,4) with respect to typology. In addition, important technical differences can also be observed among these. For example, the clasp mechanism: in a unique fashion, the spring on the Szarvas and La Rue-Saint-Pierre brooches was mounted on three small lugs. This means that the spring chord was run through the "central" lug too. In contrast, the intact spring and pin mechanism visible on the Uppåkra brooch shows the (silver) spring affixed to just two lugs. The back plate of the Bernhardsthal brooch is unknown, so we are unable to draw similar conclusions. The technical differences of the above brooches are difficult to explain: would the same goldsmith have created two brooches with such different mechanisms? Perhaps the differences in size justified the technical variations (two or three lugs?) Why were different materials used (a silver or an iron spring?): was it merely to suit the demands and financial means of the customer? At present we have no satisfactory answers to these questions and are therefore unable to decide whether the same goldsmith/workshop was responsible.
Numerous features link the Narona, Hemmingen, 'Italian' and Collegno brooches (Fig. 3.5,6,7,8) to each other and to the Danube region: common characteristics suggest they may have been made in the same workshop, although the 'Italian' and Collegno specimens display other kinds of technological solutions (too), but these are all 'at home' in this region. The similarities between the Narona and Hemmingen brooches are so strong that they must have been the work of just one goldsmith (although for some reason the stone inlaid setting was omitted from the footplate of the latter), or perhaps the Hemmingen pair was a copy of the Narona pair. The headplates engraved with scrolls on the 'Italian' and Collegno brooches also point to close connections. (Perhaps it is not coincidental that their places of discovery are close to each other -assuming we accept the location posited for the former is indeed 'Italy'.) The place of fabrication of the Domoszló and Nagyvárad brooches (Fig. 3.9,10) remains linked to the eastern half of the Carpathian Basin and it cannot be a coincidence that they were discovered there. Of the two, the Domoszló is of lower quality, although this thinly cast and roughly wrought specimen abounds in stone inlays, which is a feature of this region. Furthermore, as we have seen, the engraved S-shaped spirals also links the Nagyvárad brooch (among others) to this region.
When I published the Szarvas brooch more than fifteen years ago, the modifier 'Gepidic' appeared in the title, and at the time, I also hypothesized that the Szarvas and Nagyvárad brooches were made in a Gepidic goldsmith workshop in the Körös region. 63 As seen above, the significant expansion in the circle of brooches since then, as well as the discussion in recent literature, has led me to modify my opinion: I now consider the broader term 'Middle Danube region' to be a more apt description of the place of origin of this 'loose' brooch circle. Furthermore, we cannot state with any certainty that the women who wore the brooches and were buried in the cemeteries of Szarvas, Nagyvárad and Domoszló during the period of the Gepidic kingdom were indeed ethnic Gepids.

THE QUESTION OF THE BROOCH CIRCLE'S DATE OF ORIGIN AND THE ETHNICITY OF THE USERS
Perhaps we can better determine the relationship between the various subgroups of the brooch circle discussed above, consisting of ten brooches and brooch pairs, if we can manage to date them. Among the ten finds, altogether four came from graves whose other finds and their characteristic features are well known: the Narona, Hemmingen, Domoszló, and more recently the Collegno brooches (for the Nagyvárad specimen, see the earlier summary 64 ).
The Narona brooch pair came from a grave covered in late Roman imbrex and tegula roofing. 65 Alongside the brooch pair, amber and glass beads, a small ring and a silver, conical object (of 'unknown function') were recorded. 66 Although A. Uglešić dated the grave to the first third of the 6 th century, in the case of the brooch pair and the silver conical pin holder or brush handle (?), we might better think of the second half, or perhaps the last third, of the 5 th century. 67 The other objects (the beads and ring) are difficult to attribute to a narrower time period. In any case, A. Uglešić assigned the grave to a woman from Gepidic kingdom married to a Gepidic mercenary fighting in the Eastern Gothic army and stationed in Dalmatia.
The Hemmingen bow brooch pair was unearthed from grave no. 14 in an Alemannic cemetery along with a wealth of other grave goods: two small bird brooches, a silver bracelet, a string of glass beads, a rock crystal pendant, a fragment of a small bronze filter spoon, a bronze coin, an iron knife, bronze roll (perhaps pin holder?) and a Frankish glass cup (Fig. 5). After a detailed analysis of the artefacts, the author concluded that the grave was dug prior to 500 A.D., and the brooches are late 5 th -century specimens. 68 According to M. Nagy, the Domoszló brooch pair 'belong to the earliest Gepidic brooch group, dating to the end of the Hun period, the middle third of the 5 th century; this chronology is corroborated by the other objects from this set of graves' (Fig. 4). Elsewhere, M. Nagy states: the Domoszló brooch pair 'came from an assemblage of artefacts that can be dated to the middle or second half of the 5 th century.' 69 Other finds from this set of graves include: gold earrings with polyhedral end; necklace of glass, chalcedony and amber beads; a Cypraea shell; a spindle whorl; a double-sided antler comb; a mirror with radial design on its back; a bronze cone; a bent bronze rod. According to I. Bóna, the assemblage 'points to the single burial of a noblewoman.' 70 The assemblage of artefacts from grave no. 6 of Collegno (Fig. 6) can be securely dated primarily based on a belt buckle (presumably worn originally at the waist). Caterina Giostra considers the silver-gilt pentagonal buckle with oval loop, niello decoration, settings inlaid with garnet, four-legged animals in the Germanic animal style I along the edges and other animal depictions to be an Italian product, although she thinks the type can be observed in the Danube region, too. 71 The author dates the assemblage to the last third of the 5 th century or first half of the 6 th , based primarily on the brooch pair's Central European or 'Middle Danube regional' analogies. M. Nagy collected similar buckles in her analysis of the one from Gyula and analysed and dated them. She posited that the Gyula buckle was a Gepidic (local) product that may have been made in the middle or last third of the 5 th century, before the northern style I brooches. 72 Overall, it can be said that the buckle found among the Collegno finds was a worn, used specimen (and the brooches perhaps even more so): therefore it is possible these goods were buried with a young adult woman ('adulto-giovanile') in the first half of the 6 th century. 64 On the circumstances of the find and the dating, see B.
TóTh 1999, 262, 268. 65 The photograph of the find in the publication was of such poor quality that we could not reproduce it here. 66  In summarizing the above, we can state that, based on the assemblage of grave goods, these brooches were buried in graves dug in the last decades of the 5 th century, at the turn of the century (or in the case of Collegno, in the first half of the 6 th century). The dating of the Nagyvárad brooch does not contradict this. This information suggests that the rest of the brooches, too, can best be dated to this period in terms of their burial (with their date of fabrication somewhat earlier), allowing that, in some cases, the time of burial was in the early 6 th century. 73 However, this information is too scant for us to date the individual brooch subgroups more precisely.
Among the brooches, approximately half were found in the Carpathian Basin, and three individual or pairs of brooches can be dated to the last decades of the 5 th century or the early 6 th century, in the territory of the Gepidic kingdom (Szarvas, Nagyvárad) or in its borderlands (Domoszló). Another site is in today's Lower Austria and since we are unable to narrow that date of fabrication to a smaller period, it is not possible to know what people held political authority at the time of burial. The Herules have been suggested, but research has also pointed to the presence of the Lombards, too, during this period. 74 Here we need to stress again that political boundaries do not necessarily coincide with ethnic boundaries: we must be extremely cautious when addressing the ethnicity of the buried individuals. H. F. Müller adopted this caution in his approach to the Hemmingen grave and brooches. Although he emphasized that the brooches' origins could be found in the Danube region, the other goods in the graves, the apparel and the burial customs all clearly indicated an Alemannic environment. In his opinion, still valid today, the origins of the brooches cannot be crucial factors in determining the ethnicity of the interred. 75 This social stratum may have had distant connections, the nature of which we cannot accurately determine today. Researchers, however, had other opinions about the Narona grave and its brooches: they viewed the possibilities of identifying ethnicities differently. Z. Buljević believed the cemetery was Gothic, but he considered it an open question whether the woman buried there was, based on the brooches, Gothic, Lombardic, or Alemannic. 76 A. Uglešić, however, judged that although the grave did indeed date to Narona's Ostrogothic period, the brooch and other objects of the interred woman indicated she was Gepidic. 77 As the Collegno cemetery can be dated to the first half of the 6 th century, its publishers clearly considered it 'Gothic'. 78 It is another question to what degree this assertion relates to the northern Italian historical reality (that this was the period of the Gothic kingdom) or whether it refers specifically to the ethnicity of the people (as it was a small 'German' cemetery).
A. Koch noted that even with the Szarvas analogy, it still cannot be decided if the La Rue-Saint-Pierre brooch originated in the Danube region, but there is some likelihood that, if the brooch were Gepidic, then a strong Thuringian influence could be detected. He determined that the number of foreign brooches of Gepidic origin (Vicq, Artres) in the region he examined was negligible. He suggested the reason for this was that exogamy was rare among the Franks and Gepids; meanwhile there is no archaeological evidence of the deliberate settling of Gepidic groups in the region. 79 The most recently discovered Uppåkra brooch was also found in an area very far from the Middle Danube region, making its appearance there quite surprising. In his analysis, J. M. Sundberg cites J. Tejral and A. Koch. He does not think the brooch's place of fabrication can be more precisely identified, but because Thuringian influences felt in Bohemia, Moravia and Lower Austria may have played a role in the creation of this item, then is might have been made there. 80 According to the author the brooch is not the legacy of a woman who had travelled from a great distance, but rather of one who had arrived in the southern Swedish area of Skåne by other means. He believes that 73 According to J. M. Sundberg, the Uppåkra brooch can be dated to the end of the 5 th century or first third of the 6 th based on analogies.  Longobardi 2017, 9. 79 Koch 1998, 555. He remarks with respect only to the brooch from Artres that the marriage of a Gepidic woman who belonged to an upper stratum of society was responsible for its arrival there. According to M. Kazanski, in Gallia, only eagle buckles and the Vicq-type brooches can be found among Gepidic objects. These have analogies in the Tisza region and in Transylvania. Consequently it is not possible to verify archaeologically the hypothesis that Gepids from Sirmium were settled in 523 A.D. KAzAnsKi 2010, 130. 80 sundberg 2013, 54. L. Larsson and B. Söderberg also believed the brooch was a continental work. Closely related brooches originated in Thuringia and Bohemia, where, because of its execution, this brooch most likely would have belonged to an 'aristocratic costume'. The object dates to c. 500. LArsson-söderberg 2013, 241. because there is no proof that foreign craftsmen were active in Uppåkra, it is far more likely that this object had been a commercial product or gift.
Thus, there is no uniform opinion about how these brooches were separated from each other, in some cases by great distances (and how they travelled from their place of fabrication?): researchers have proposed that for some brooches (Narona, perhaps Hemmingen) the owner's individual mobility was the explanation, while for others it was commerce or the owner's other personal connections (Uppåkra). In any case, H. Steuer offers an interesting contribution to the debate: he posits that the wide territorial range of Merovingian-period brooches more likely reflects the activities of military retinues rather than the movement of goods through commerce or exchange. 81

HOW WERE THE BROOCHES WORN? WHAT MODES OF WEAR SHOULD WE RECKON WITH?
We have far more information today than a decade and a half ago about how small brooches were worn. In an earlier overview, M. Nagy determined that, because of their size, small, cast brooches excavated from Gepidic cemeteries were used alongside the main decorative brooch as complementary accessories. 82 Her opinion about the use of a brooch from grave no. 105 in Kishomok was similar. The grave had been disturbed, and the author surmised that grave robbers had stolen the large, valuable brooches and left in place the smaller bronze brooch, which had been pinned at the height of the lower ribs. This placement suggests this artefact was a 'third' brooch (Drittfibel) -that is, a jewellery supplement to the main brooches. 83 The question, therefore, is what function did these brooches in the Szarvas group have in the female costume of the time: do we have any reliable information from which we can draw conclusions? As we have seen above, in four cases the brooches originate from authentic graves that can be analysed for the most part (Hemmingen, Narona, Domoszló, Collegno). In these cases it is extremely important to note that all of these small bow brooches were excavated in pairs. Perhaps we can conclude from this that all the other specimens in this group had also originally been worn in pairs. We only know the positioning of the brooches in the grave (and thus on what part of the garment they were worn) for the Hemmingen and Collegno specimens. The former pair was worn on the lower part of the upper body and at the pelvis, with the two items aligned vertically, one above the other. It is important to observe that, despite their size, these bow brooches were in the position usually occupied by the 'large brooch' in the waist area. This is reinforced by the placement in the grave of two 'small brooches' (bird-shaped brooches) under the neck and on the chest. The presence of four brooches in these positions corresponds to the Alemannic manner in that period of wearing four brooches. This way of dressing, the placement of the brooches, had contemporary analogies in the Carpathian Basin, mainly in Transdanubia. M. Nagy also mentioned the positioning of the brooches in Hács-Béndekpuszta: in graves no. 18 and no. 20 the brooches are similarly arranged, in pairs, at the waist level (aslant, and one above the other). But in contrast to the Alemannic custom, here they were not accompanied by two additional small brooches on the upper body. In grave no. 150 in Balatonszemes-Szemesi berek, the positioning of the interred woman's bow brooches was similar: she had high-quality accessories (gold beads and a belt with amulets) but did not wear any additional brooches on her chest. 84 We can also refer to grave no. 4 in Szurdokpüspöki: two brooches were placed on the right, vertically aligned between the pelvis (iliac bone) and the ribs, but again no brooches were found in the chest area. 85 Thus, in the western and northern regions of the Carpathian Basin, a new style of dress appeared in the third 81 sTeuer 1994, 139. 82 nAgy 1993, 72. 83 bónA-nAgy 2002, 120. In addition, M. Nagy does not consider it impossible that the brooch -because of the disturbancewas in a secondary position. The length of the brooch is 4.7 cm, and it had a flat bow. For a brief summary on the typological differences between bow brooches and small brooches, see B. TóTh 1999, 269. 84 Miháczi-PáLfi 2018c, 131, 146-147. 85 bácsMegi-gubA 2007, 16, 26. 86 sTrAub 2008. The author notes that this group lacks certain jewellery or unique accessories (such as small brooch pairs) that were characteristically found in Alemannic cemeteries. Objects typical of the Carpathian Basin, however, were found. But, as with every rule, here too there are exceptions. The Krefeld-type brooch of Tác/Gorsium (from grave no. 154) was found between the woman's elbow and spine; thus it was only a single brooch, not part of a pair. It was accompanied by a belt buckle, glass beads, a comb, a spindle whorl and iron implements. There is no mention that the grave was disturbed; therefore, this grave, dating to the second half of the 5 th century, was the resting place of a woman whose style of dress was characteristic of women found in later, mostly 6 th -century, cemeteries in Gepidic regions. On the brooch and the grave, see fiTz- Lányi-bánKi 1975, 292;schiLLing 2011, 385-387, Abb. 1.1. quarter of the 5 th century: bow brooches were worn in pairs, one above the other in a vertical line or slightly inclined, but, in contrast to the fashion of the analogous upper Danube region, without additional small brooches. 86 In grave no. 6 in Collegno, the partially preserved skeleton of the young woman has half of a broken brooch on the left shoulder, while the other half of the brooch was on the right side of the torso, next to the more intact brooch. 87 Unfortunately, we cannot definitively conclude that, if the grave had been completely undisturbed, both brooches would have been located at the shoulders, one on each side (meaning one had slipped lower down) or that the deceased had worn clothing that closed diagonally. 88 Both dress styles belonged to the 5 th -century female fashion of the Carpathian Basin, although to differing degrees.
Sadly, we have no information about the position in which the Narona and Domoszló brooches were found in the graves, on the skeletons. As both authors primarily presented East-German (Gothic or Gepidic) analogies in the case of the Narona grave, we need to consider that these pairs of brooches may have been worn on the shoulders in the Gothic style of dress ('peplos'). In the middle third of the 5 th century or in the decades following, this type of costume was typically accompanied not only by so-called silver plate large brooches and, later, larger cast-engraved silver brooches, but by another type also: most recently, Zs. Rácz published an approximately 5-cm copper alloy brooch pair from Hajdúnánás. Both members of this pair have a footplate terminating in a bird head and headplate with three knobs and were discovered in situ, at the shoulders of the deceased. 89 This clearly demonstrates that such a small bow brooch pair could also have been worn with a 'peplos'. The author has collected brooch pairs found in similar positions in the eastern half of the Carpathian Basin. 90 A few bow brooch type specimens widespread in the western and northern parts of the Carpathian Basin in the second half of the 5 th century (Léva-Perse/Levice-Prša, Bakodpuszta, Letkés-Hács-Béndekpuszta) were similarly small, especially the last two types, which were mostly made of gilt silver. 91 Those artefacts that were not stray finds, but rather were found in graves, were discovered chiefly in pairs and typically represented every style in which brooches were worn in the region in that period. The positioning of some pointed to the wearing of a peplos (Letkés, Keszthely-Fenékpuszta). 92 In graves no. 18 and no. 20 in Hács-Béndekpuszta, however, (as discussed above) the vertical, but slanted positioning of the two brooches strongly suggests the garment was a tunic. In any case, it is certain that the brooch type had no bearing on its positioning, since small bow brooch pairs of the same type could be used to fasten both peploi and tunics, depending on the customs of the wearer. Almost without exception, however, brooches were won in pairs during this period. 93 In summary, we can conclude, based on some certain facts, that the small bow brooches in the Szarvas circle were worn in pairs according to the fashion of the period and in the mode adopted by the deceased woman 87 C. Giostra kindly provided this information in a letter on 4 January 2019: ʻGrave 6. was intact, the skeleton partially preserved, the broken brooch was on the left shoulder, but the fragment was on the right side of the torso, near the whole fibula.' 88 Most recently, see Miháczi-PáLfi 2018a, 24-26, 33-34, 2, 5, Fig. 9. The author searches for the antecedents to the 'clothing fastened on a diagonal' in the Chernyakhov-Sântana de Mureș culture and dates this fashion to middle third of the 5 th century in the Carpathian Basin. 89 rácz 2014, 205-206. In addition to the brooch pair, the grave contained two small dark-coloured beads, a double-sided comb and an animal bone. This grave of a 45-to 55-year old woman was to a smaller degree disturbed by animals. The author dated this cemetery consisting of a few graves to the second half of the 5 th century. 90 On the various arrangements of the brooches in the graves in the Danube region in the 5 th century, see: rácz 2016, 315-317. 91 Mészáros 2015, 63-86. The lengths of the silver-gilt specimens were 4.8 to 5.4 cm for the Bakodpuszta type and 4.4 to 5 cm for Letkés type. The bronze specimens were typically longer (as much as 9.2 cm). For information on the Léva-Perse/Levice-Prša type, see Mészáros 2014: brooches of this type were made of bronze, and their lengths were accordingly between 5.8 and 8.3 cm. I am grateful to the author, who allowed me to use the figures in her manuscript dissertation.
92 For more on the Léva-Perse/Levice-Prša type brooches, see, for example grave no. 133 in Csongrád-Kaszárnya and grave no. 1 in Jászberény-Szőlő-dülő. Mészáros 2014. The 'peplos' was still the likely costume when two small iron brooches with inverted foot were found at the collar bone: see grave no. 1979/2/17 in Keszthely-Fenékpuszta, Südmauernekropole, sTrAub 2011, 337 or grave no. 489 in Fonyód-Vasút-dülő, Mérnöki telep, sTrAub 2014, 218-220. The length of the iron brooches is between 4 and 5.5 cm. The author dates the site of discovery to the 'Hun period', the middle third of the 5 th century. 93 Although during this time, in albeit narrow circles, the custom of using true 'small brooches' as part of women's costume was known in the Carpathian Basin. The so-called third brooches (Drittfibel) were largely used as supplements to the two brooch pairs that fastened the peplos. A portion of the bird-shaped brooches, made from various materials using a variety of techniques, are examples of these. Most recently Zs. Rácz collected bird-shaped brooches, but few of them could be definitively dated to the 5 th century (for example, Beregszász/Beregovo) rácz 2011, 171. On the brooches and the different variations in clothing in the eastern half of the Carpathian Basin, see this paper. during her lifetime. In the second half of the 5 th century, this may have been a peplos or tunic, or even (in the case of the deceased in grave no. 6 in Collegno, dating to the second half of the 6 th century) a 'garment that closes diagonally' ('asymmetrical'): however, we cannot know this for certain without further information. 94 Here we should note that the silver pin mechanism of the Uppåkra brooch (not only the catch plate but the spring and pin mechanisms were also made of silver) does not have a lot of analogies from the period (since the spring and pin mechanisms were generally made of iron): the material alone, therefore, clearly demonstrates the exceptional nature of this brooch. D. Quast observed similar characteristics in a unique, high quality, gold-plated brooch with cloisonné decoration from an Alemannic cemetery in Gültlingen. He determined that the silver pin mechanism was impractical, since silver is a soft metal ill-suited for supporting a heavy brooch. In fact, this item was not used for a long periods of time: it would not have held up under more intensive use. 95 Several decades earlier, however, in workshops that produced the highest-quality brooches, it may have been the practice to use the same metal in the spring and pin mechanisms as in the body of brooches made of precious metals and richly decorated with stone inlays; see, for example, several brooches from the second treasure of Szilágysomlyó. 96 Some superb-quality, silver plate brooches also display similar practices. A few examples include: the spring and pin of specimens from Castelbolognese (Italy), Újlak (Ilok, Croatia), and Untersiebenbrunn (Austria) which were also made of silver wire (whether the brooches themselves were silver-or gold-plated). 97 The same is true for the spring and pin of the cast-engraved, gilt, silver brooch with vine motifs from Zsibót-Domolospuszta. 98 Thus practicality was not the primary consideration in the creation of the Uppåkra brooch either: perhaps here too quality and prestige played a greater role.

IN THE SECOND HALF OF THE 5 TH AND THE EARLY 6 TH CENTURIES
These questions have clearly best been researched in the western German regions: in addition to the numerous cemetery publications, independent volumes have also been published by researchers involved in collecting and classifying brooches for more than a century. The professional literature could soon fill a library; the goal here is not to analyse this material or acquaint the reader with it. All that can be said briefly is that in the middle of the 5 th century in the western and eastern German regions, a process began during which various bow brooch types were created and developed, many of these arising in part from different traditions but still displaying very close connections. In the west, brooches arose from basic forms found in the late Roman Empire: generally crossbow brooches are cited as the prototypes for the Krefeld-type brooches.
In the Carpathian Basin, sheet brooches (either of silver or copper alloy) and the (mainly iron) brooches with inverted foot, both types known from the first half of the 5 th century, may have been the starting points for the development. In this region, two further characteristics contributed to the shaping of these brooches: the use of animal depictions and lavish stone inlays. Furthermore, in the earlier Roman regions (workshops?) chip-carved, cast clothing accessories (mainly belt fixtures) were most likely made: in these cases, the changes entailed a shift in these 94 This fashion of wearing the brooches on a 'diagonal' meant one brooch was worn on one side of the upper torso, while the other was worn lower down, near the waist, on the opposite side. This perhaps derived from one of the modes of clothing among the people of the Chernyakhov-Sântana de Mureș culture. For the most recent discussion, see: rácz 2011, 171: 'brooches worn in pairs, asymmetrically.' An example of this comes from grave no. 18 in the cemetery of Szolnok-Zagyvapart-Alcsi: at least, the Viminacium-type iron brooches worn beneath the right shoulder and next to the left elbow allow for this conclusion. According to J. Cseh, who excavated and wrote about the cemetery and the grave, it can be dated to the late 5 th -early 6 th centuries, cseh 2005, 22. Similar discoveries were made in Transdanubia: for example, the Léva-Perse/Levice-Prša type brooches in grave no. 210 in Szekszárd-Palánk. Mészáros 2014, 10. 95 QuAsT 1993, 68, 75. 96 Barbarenschmuck 1999. See, for example, the gold pin mechanism of the gold brooches with seahorse decoration, the silver brooches with enamel decoration, or the examples no. 56-57 and 58-59 in the catalogue and the silver spring remains on their reverses. The pin mechanism on the rest of the specimens did not survive. D. Quast listed the analogies to the Gültlingen brooch, those with silver spring and pin mechanisms (London, Berlin, Imola, Untersiebenbrunn and the two Szilágysomlyó brooches cited above). items from male to female costume (women's belts and later brooches in place of military belt fittings) or the replacement of copper alloy with silver as a raw material. The earlier method of gold plating the silver body of betterquality brooches was supplanted by the fire gilding technique for coating the item. (This was the method that made it possible to 'coat' chip-carved decoration.) Stone inlay decoration became somewhat less popular in the second half of the 5 th century: the inlaying of stones was reduced to just a few elements, but it remained characteristic to place a few inlaid stones on the edges of buckles and brooches or in areas requiring emphasis. In addition, the already known practice of niello decoration was revived. This formal and technical richness resulted in great variety and many exceptional creations in the middle decades of the 5 th century: these elements were often uniquely combined in many different ways. 99 Among brooch typologists, it is a widely accepted view that objects of outstanding quality were likely trendsetters: these generally unique works served as models for the simpler, smaller (?) series of brooches, which can often be traced to just one design. 100 We have only two examples of this process from the second half of the 5 th century. Perhaps we are not mistaken if we consider the Zimony (Belgrade/Zemun) brooch the typological prototype of the Bakodpuszta-type brooches, at least as far as the characteristic radial engraving on the headplate is concerned. Similarly, the very good quality brooch with a round headplate, chip-carved decorations and a straight footplate from grave no. 150 of Balatonszemes-Szemesi berek may be (one of?) the possible prototypes of the Letkés/Hács-Béndekpuszta-type brooches. 101 In the last decades of the 5 th century and into the next, brooch types emerged in Western Europe and the Carpathian Basin that later, in the 6 th century, would be produced in larger series and for a longer period of time. 102 Crucial elements in this process were certainly the individual goldsmiths and workshops as well as the customers. For some time, researchers have endeavoured to identify the works of various craftsmen and workshops based on smaller or larger formal or technical features: although this effort rarely garners much success, recently significant progress has been made in this respect in the assessment of finds from the Carpathian Basin. 103 In part, other masters or workshops may have copied the original or used the same techniques. At the same time, it would be a mistake to assume that workshops or goldsmiths had just one brooch type in their repertoire. We even lack true consensus about whether single objects can be attributed to an individual (itinerant?) master, or whether the oftentimes high-quality objects were products of workshops employing craftsmen specialized in performing only certain processes. 104 Also a subject of debate is the social and legal standing of the (gold)smiths, although most likely this was not uniform, but varied from case to case. 105 Perhaps in some instances, we will manage to recognize the works of certain 'hands' by observing the minor details of metal workshops. 106 We have already discussed how the differences or uniformity of the raw materials provided no clear answers in the case of another pair of brooches. Nevertheless, in an interesting development, an electron microprobe has shown that not only were two, very similarly executed brooch pairs from the Hács-Béndekpuszta cemetery (from graves no. 19 and no. 20) composed of almost exactly the same materials, but yet another pair belonging to a different brooch type, but found in the same cemetery (in grave no. 18), also had the same composition. We can therefore postulate that the Hács/ Béndekpuszta type brooches were cast in the same workshop as the Bakodpuszta types from the same cemetery, and this workshop may have been somewhere in Pannonia. 107 It can therefore be proved that this goldsmith/workshop did not produce just one kind of brooch, but several kinds. 99 A similar process occurred in the western German regions too. According to A. Koch, in the mid-5 th century, all kinds of distinct brooches were developed, a large portion of which were experiments: the various forms and decorative elements were combined almost indiscriminately, and thus new formal groups emerged. Koch 1998. 470. For a summary of the Carpathian Basin: nAgy 2007; horváTh 2012. 100 Koch 1998, 488. 101 Mészáros 2015. There were also good-quality, uniquely-fashioned brooches that had no typological continuation: for example, the chip-carved design on the brooch from grave no. 4 of Szurdokpüspöki. The decoration of the semi-circular headplate with quadrilateral elements may have been a 'clumsy' solution, but the symmetrical 'wave-bundle' pattern similarly has (thus far) no known analogies or imitations (bácsMegi-gubA 2007, 16, 6). 102 The two phenomena, the larger number of objects and the formal-technical similarities or uniformity, are obviously related. 103

Possible connections to the Uppåkra brooch: how did it wind up in the distant north?
One of the 'central places' (central plats) lies 5 km from Lund, in Skåne, in Uppåkra, in Southern Sweden. What these settlement centres had in common was that all of them may have been centres of (smaller or larger) environments for longer periods of time and their role as centres may have encompassed multiple functions. 108 Traces of their artisan, commercial or sacred functions in some cases can be demonstrated through archaeological methods, but their importance may have differed and, over the course of the settlement's life, may have changed from time to time. Obviously these settlements were home to the local elite and were thus simultaneously 'centres of power', part of an extensive commercial/trade network, but also centres of distribution for their own products. 109 Uppåkra was continuously inhabited from 100 B.C. to 1000 A.D. Systematic excavations began in 1996, and until just recently 40 hectares were examined (in addition to the excavations, site ground surveys and metal detecting methods were also employed). In the centre of the settlement (the residential area of the elite), a cult-type building 110 and several 'longhouses' were unearthed: between the 5 th and 10 th centuries, four buildings stood in the same place, some of which fell victim to fire. 111 The artefacts that have come to light date the so-called 'B' house to c. 500 A.D. The brooch discussed above was excavated in the vicinity of the foundation walls, but its presence had already been discovered through metal detection. As mentioned, the brooch was thought to be 'Central European' and attempts are being made to find its place within the settlement's network of connections, which have been steadily emerging. Already in the Roman period, Southern Sweden had extensive connections with the various regions of the 'continent': not only with the Roman Empire but with the people of the Chernyakhov culture too. The many import goods found in the graves verify this connection. At the beginning of the migration period, the structure of society underwent change: the new military elite took power and the previous community built on kinship was replaced by one that relied on personal connections. The nature of the goods arriving from afar also changed: high quality weapons and jewellery were no longer typical trade goods. Among them is a group consisting of gold or gilt jewellery with garnet inlays: these were acquired through direct, personal connections. However, the usual trade goods were also present: for example glass vessels. 112 Bow brooches with garnet inlays thus belong to a group of objects that represent connections with other regions of the continent. 113 According to J. M. Sundberg, there is no evidence that foreign goldsmiths, who used designs from distant lands to create objects locally, were active in Uppåkra. It is more likely that the Uppåkra brooch was a trade item or gift, but it is possible that it arrived to the region with the wearer (and was one member of a pair). Its final function may have been as an 'offering' a sacrificial object put in the foundation trench. An object of this quality would scarcely have been lost, and its role as an object of sacrifice is underscored by the presence of a pin and pin holder discovered nearby. 114 Taking into consideration the wider context, we see that the brooch in question -because of its uniqueness in the north-most likely belonged to the 'personal mobility' or 'gift' categories of artefacts. 115 108 'A place where one or more social functions are concentrated in such a way that the place has become the dominating room of the region' and therefore, 'archaeologically, the site should have a rich and varied find material indicating various functions '. hÅrdh 2002b, 27. 109 The local craftsmen were involved in working metal, bone and horn, producing these objects in such quantities that they surpassed demand among the residents of the settlement. Local goldsmith production also must have been significant, as the small figural patterns stamped in gold foil (ʻguldgubbar') seem, at least, to indicate. hÅrdh 2002a, 42-43. 110 A smaller building (cult house, ceremonial building). It was built in the 3 rd century and restructured in seven phases and was used for more than 700 years. An abundance of sacrificial finds and animal bones indicate the purposes of this structure. LArssonsöderberg 2013. 111 LArsson-söderberg 2013. 112 LJungKvisT 2009 Archaeological finds suggest that during the 5 th century, the connection between the two regions (Southern Scandinavia and Central Europe) worked in the opposite direction too. J. Tejral mentions a northern type with cross-shaped headplate and rhombus-shaped footplate from Vrchoslavice, Moravia, whose closest analogy is known from Uppåkra. (The object is dated to 'the middle or second half of the 5 th century'). The author considers this as evidence of a relationship between the two regions; in any case, with respect to culture, Moravia was a part of the Middle Danube region during this time. 114 The author notes the brooches found recently and not yet published may include several that were continental in origin. sundberg 2013, 55.

SUMMARY
The group of ten brooches and brooch pairs analysed above -based on very detailed typological and technical comparisons -despite striking common features, do not constitute a type. Instead they form a more loosely circumscribed brooch circle that can be broken up into subgroups (which could be considered separate types if there were a larger number of specimens). This group cannot be classified among the far more uniform brooches belonging to the Thuringian bird-head type; rather they can be considered antecedents, precursors, prototypes. 116 They may have been made in the last third or final decades of the 5 th century, or perhaps around the turn of the century, but a more specific date cannot be determined. The Szarvas and La Rue-Saint-Pierre brooches were surely made by the same goldsmith ('hand') and perhaps originally formed a pair. Presumably the Bernhardsthal and Uppåkra brooches were also made in the same workshop, but on different occasions and for different clients. The 'Italian' and Collegno specimens, and to a lesser degree the Narona brooches, also show strong typological similarities, but are distinguished from each other by the variations in goldsmith methods and modes of decoration that were fashionable in the region at that time. The Hemmingen brooch pair was either made in the same workshop as the Narona pair, or the former is a copy of the latter. All we can conclude about the place of fabrication of the brooches listed is that they were made somewhere in the more broadly interpreted 'Danube region', perhaps in the Carpathian Basin (western, southern areas?). The Nagyvárad and Domoszló brooches may be the products of another workshop: given the location of their sites of discovery, this place may be in the eastern half of the Carpathian Basin. The Narona, Hemmingen, Collegno and Domoszló finds suggest that these brooches -despite their small size -were used in pairs, serving the same function as a 'large brooch'. We have evidence that during this period, similarly sized brooches were used on both 'peploi' and tunics (and in the case of the Collegno brooches, on a garment 'closing diagonally'?) The sites in which the individual members of this brooch group were found are extremely far from one another (Southern Germany, 'Italy', Collegno, Southern Sweden, Dalmatia), with the greatest concentration in the Carpathian Basin (Bernhardsthal, Szarvas [for both the Szarvas brooch and its original companion, the 'La Rue-Saint-Pierre' specimen], Domoszló, Nagyvárad). We do not know the reason for the extreme 'scattering' of this small group of brooches. Thus far, expert opinions vary: some believe it was a consequence of exogamy, while others contend that it was because families moved across great distances. Researchers have also suggested commerce or exchange may have been responsible: in elite circles, these finely executed brooches may have been considered appropriate gifts. An exception may be the brooch named for its place of discovery, La Rue-Saint-Pierre: it still seems likely that the circuitous paths of late 19 th -century art market are the reason this specimen from Szarvas wound up in the museum in St. Germain-en-Laye, where it was recorded as having an 'unknown site of discovery'.

CATALOGUE 117
Szarvas, Békés County ( Fig. 1.1, Fig. 3.1 118 ) Gilded silver, cast. The footplate terminates in an animal head: the almond-shaped eyes contain dark red garnets inlaid in a recessed collar. In line with the eyebrows and nose is a Y-shaped strip accentuated with a row of punched niello triangles. Above the animal's nose is a transverse rib. The jowls are formed by two small, rounded bulges. The cross-section of the upper part of the footplate is 'rooflike', surmounted by a long, rectangular setting with garnet inlay (the stone rises slightly above the walls of the setting and is damaged). The stone is backed by gold foil with a crosshatch pattern. The two strips flanking the setting both contain a row of ten transverse ribs. The bow and footplate have the same width, with the bow width increasing almost imperceptibly towards the headplate. The bow is arched: the middle, longitudinal rib contains two rows of opposing, punched triangles inlaid with niello. This protruding part of the bow, towards the centre, is worn. The two side bands of the bow are adorned with a deep, chip-carved pattern consisting of 116 The Altenerding specimen (grave no. 532) may be an example of the 'Thuringian' brooches. The bow and footplate -based on the photograph -are nearly identical to those of the Szarvas brooch, but it lacks the rectangular setting and the headplate is 'pincer-like' (Zangenfibel). Although this differs from the others, it does have dec-oration consisting of two scrolls. sAge 1984, I. 153, II. Taf. 73.12, Taf. 188.1. 117 The catalogue lists the sites not in alphabetical order, but in the order in which they (or their subgroups) appear in the text.
118 I am grateful to Anita Korom and Viktor Szinyei for their assistance in editing the Figures. a slightly curved zigzag. The headplate has tulip-like contours and is adorned with three, with stone inlaid flush settings in cast recesses. The central setting is pentagonal and is flanked by two settings that correspond in shape to approximately a 120 degree circle sector. The upper 'radii' of the sectors are slightly curved. Beneath the garnets is corrugated gold foil with a crosshatch pattern. The stone in the right setting is cracked. The upper edge of the headplate towards the bow -on both sides of the bow -is slightly wider and has a deep groove running along it. At one time the entire brooch had been fire gilt, but this has worn away on the protruding parts. The bow is broken in two and today is glued together. On the reverse, three integrally cast lugs supported the spring, two on each side and the third at the bottom of the headplate. Nothing of the (iron) spring and pin has survived. The reverse of the footplate has an integrally cast, folded catchplate: it is worn, damaged and part is broken off. The length of the brooch is 4.95 cm.

.3)
This description is based on the published drawing and description of this brooch. Silver cast, gilt. The footplate of the brooch terminates in an animal head. The eyes are almond-shaped with no stone inlays. An unadorned band runs along the nose line. Next to it are two long, protruding shapes forming the jowls. The rest of the footplate is composed of a centrally positioned rectangular setting with red 'glass inlay'. The two bevelled side bands both have five pronounced engraved, transverse ribs. The bow is strongly arched, and the central, longitudinal strip is decorated with a zigzag and inlaid niello (?) The two side bands have an engraved zigzag design. The headplate is adorned with three stone inlaid flush settings in cast recesses: the centre one is four-sided while the two side recesses form circle sectors. The upper edges are straight and the settings contain 'glass inlays' (?) Beneath this panel, on each side of the bow are rectangular fields each with two grooves ('mit Strahlenbündeln') On the reverse, the spring may have been affixed to a single lug (based on the photo); the remains of an iron spring can be seen. The catchplate is rectangular and appears intact in the drawing. The length of the brooch is 3.7 cm.
The brooch was discovered in lots in Wehrlehen field along with finds from a similar period (bronze and silver-gilt brooches with engraved decoration and a bronze dagger handle 119 According to A. Koch, the length of the brooch is 5.3 cm, but the photograph suggests this figure is inaccurate. Koch 1998, 645. 120 A. Koch changed the name of the site of discovery for this brooch from 'unknown' to La Rue-Saint-Pierre without offering any explanation (perhaps based on documents he had recently discovered?). In the article he cites (dAnJou 1858), there is just one description of a grave rich in grave goods, including weapons, but there are no references to a possible grave (of a women) or the goods it contained. Therefore, I do not feel this unexplained change in the place of discovery is justified, and rather, it only increases the number of misunderstandings. Nevertheless, I have no choice but to use the name La Rue-Saint-Pierre since this is how the brooch appears in the latest literature.
Uppåkra, Skåne, Southern Sweden ( Fig. 1.3, Fig. 3.4) This description is based on the photo and a brief, published description. Gilded silver, cast. The material has a high silver content. The end of the footplate terminates in an animal head with almond-shaped eyes framed by deep grooves. A Y-shaped strip runs along the nose and eyebrow lines and is accentuated with dark, linear inlays. The jowls are formed by two distinct, convex forms. The lower part of the footplate is surmounted by a rectangular setting inlaid with garnet that is backed by corrugated gold foil. The two bands flanking the setting contain pronounced, transverse ribs (the photo shows five on each side). The bow and footplate are of equal width, and the bow is strongly arched. On both sides of the central, longitudinal band are two unadorned strips separated by a rib. The central band is filled with a dark, paste-like material. The headplate is decorated with three settings in cast recesses: the central one is four-sided, while the others are shaped like circle sectors. Garnets backed by corrugated foil fill the settings. The upper garnet is damaged and cracked. Although the upper edges of the side settings are straight, the fields above them are curved with peaked ends and are grooved. The reverse has a completely intact spring and pin mechanism. The silver spring is mounted on two lugs: the spring axis (not visible) must also have been made of silver. The spring chord loops upward. The sharp pin is intact and rests in the rectangular, folded catchplate, which likewise shows no signs of wear. The brooch was probably very seldom used. Its length is 4.1 cm and its weight is 8.71 g.

5)
The two Narona brooches appear identical based on the photo and the description. Gilded silver, cast, engraved. The footplate terminates in an animal head, the oval eyes are positioned almost vertically, and between them runs a protruding rib decorated with niello triangles. The animal head is separated from the rest of the footplate by a horizontal rib. The other part of the footplate is surmounted by an empty rectangular setting that had obviously been inlaid with stone. On each longitudinal side of the setting runs a rib decorated with horizontal lines. The bow is equal in width to the footplate and is arched. The protruding central band is occupied by two rows of opposing, punched triangles filled with niello. In the two bands that flank it, five double semi-circles can be seen. The headplate is semi-circular, and its upper edge, where it meets the two side bands of the bow, is curved. Radial ('fanlike') ribs fill the headplate, and each rib is adorned with transverse engravings (pseudo-filigree). At the pinnacle of the headplate and laterally, lower down, are three settings with garnets (although in the photograph, the settings are empty). The lateral settings are slightly irregular ovals and the centre one is a pentagon with the upper two sides smoothed and rounded. The pin mechanism is made of iron. The brooch's length is 5.8 cm.
The brooch pair was excavated from a grave that had imbrex and tegula roofing. Also found in the grave were six amber beads, a ring and conical, silver object of 'unknown function' (pin holder or brush handle?) The placement of the goods in the grave is unknown.
Arheološki Muzej, Split, inv. br. 1088-1089; S1, G2. Literature: buLJević 1999, 240-241, T. XVII.1-2;ugLešić 2003, 206, Sl. 6,7;TeJrAL 2008, 274, Abb. 17.13-14;TicA 2017, 241-246. Hemmingen, grave no. 14, Kreis Ludwigsburg, Baden-Württemberg, Germany (Fig. 2.2, Fig.3.6, Fig. 5) Brooch pair. Gilded silver, cast. The footplate terminates in an animal head, the almond-shaped eyes are slanted and a Y-shaped strip runs along the line of the nose. The part of the footplate towards the bow is flat, with a semi-circular cross-section, and is covered in horizontal ribs (with no setting). The bow is equal in width to the footplate and arched. The central band protrudes and contains two rows of punched decoration. The headplate is semi-circular and the upper edge on each side of the bow is curved. The headplate has a 'fanlike', ribbed pattern. At the pinnacle is a setting shaped like a pentagon with rounded top, while two more, oval settings containing flat ('almandine') garnets are laterally positioned. The reverse is undecorated and has two lugs close together, but the pin is missing. The pair is in bad condition: the headplate of one brooch is fragmented and the drawing also suggests it has suffered more abrasion than the other. Their length is 5.8 cm.
The brooches were discovered in a grave containing the poorly preserved skeleton of a 30-year-old woman. The dimensions of the grave are: 1.75×.45×1.5 m. One of the brooches was found between the left elbow and the ribs, in the lower area of the chest, with the headplate pointed downwards. The other brooch was on the left pelvic bone (ilium), along the same axis as its companion. The other grave goods: two bird-shaped brooches beneath the chin and on the chest, a silver bracelet on the left wrist, glass beads and a piece of bone with a drilled hole beneath the chin and on the chest, a rock crystal pendant between the knees, amber and glass beads, a small bronze filter spoon, iron knife, a bronze coin on the outer side of the right knee, a bronze tube next to the left foot (perhaps a pin holder) and a glass cup next to the skull.
Württembergisches Landesmuseum, Stuttgart, Inv. Nr. F 67/11 Literature: MüLLer 1976, 30-34;LoserT-PLeTersKi 2003, 181, Abb. 26.6;TeJrAL 2008, 274, Abb. 17.12;TicA 2017, 243. 'Italy' (Fig. 3.7) The description is based on the drawing. The footplate of the silver brooch terminates in an animal head with tiny, round eyes. Along the eyebrow and nose line is a Y-shaped band containing two rows of punched triangles. The mouth of the animal is a triangular setting inlaid with a stone backed by crosshatch foil. The other part of the footplate has a long, rectangular setting in the protruding central band. It is inlaid with a stone backed by crosshatch foil. The two side bands are bevelled contain transverse ribs. The bow and footplate are the same width. The bow is arched and only the central longitudinal band is decorated with two rows of opposing, punched triangles. The headplate is semi-circular with a raised border that has similar punched decoration. The inner panel is adorned with two S-shaped volutes. Three, elongated oval recesses stud the edges of the headplate. The settings are inlaid with stone backed by crosshatch-patterned foil. The drawing only shows the front view of the brooch, so the reverse, pin mechanism, etc. are unknown. The length is 5.7 cm.
'Torino museum', the circumstances in which it was found are unknown. Literature: sALin 1904, 194, Nr. 467;Åberg 1922, 109, Abb. 153;buLJević 1999, 241;nAgy 2007, 23, Abb. 15.3. Collegno (Torino, Piemonte, Italy) ( Fig. 3.8, Fig. 6) The description is based on the photograph and information in a preliminary report. The footplate of this silver-gilt brooch pair terminates in an oval animal head with a semi-circular setting at the end (the animal's 'mouth'). The eyes are round settings. Along the eyebrow and nose line runs a Y-shaped band with punched (niello) decoration. The rest of the footplate has a centrally positioned, rectangular setting. One member of the brooch pair has an inlay of green stone (Fig. 6.1) and the other of red stone (Fig. 6.3). 121 The two outer bands have transverse ribs. The bow is equal in width to the footplate and arched. The central, longitudinal band has two rows of opposing, punched triangles (probably filled with niello). The headplate is semi-circular and has a silver-coloured, prominent border with punched decoration. The inner panel has engraved, perhaps symmetrical acanthus scrolls. Three settings stud the headplate: the centre one at the top is semi-circular, and perhaps the two lateral ones were also, but because of their fragmented state, it is not possible to determine their original shape. The more intact member of the pair has a red stone inlay, while the more damaged brooch has a 'colourless glass' inlay in one of the side settings. One of the brooches is broken in two and both are heavily abraded and damaged. The photograph shows only the front view, so the spring and pin mechanism is unknown. The length of both brooches is 6.6 cm.
The brooch pair was found in grave 6 of a 'Gothic' cemetery with eight graves. The grave contained the skeleton of a 'young, adult female' and was undisturbed; the skeleton, however, was in very bad condition. One part of the broken brooch was found at the left shoulder, while the other fragment was on the right side of the skeleton, next to its more intact companion. 122 A necklace of 21 amber beads and a 15-cm-long, silver-gilt, engraved, buckle adorned with inlays and niello were also found in the same grave.
Musei di Torino -Museo di Antichità, Torino. Inv. No. 93925-93929. Literature: PeJrAni bAricco et al. PeJrAni bAricco et al., in prep.;Longobardi 2017, 8-11;TicA 2017, 245-247. Nagyvárad/Oradea, Guttman brick factory, Romania (Fig. 2.3, Fig. 3.9) Silver gilt, cast. The footplate of the brooch terminates in a round animal head with round eyes enclosed in grooves and inlaid with convex garnets. The entire head is slightly 'turtle-like'. The other part of the footplate has a rectangular setting with garnet inlay and is flanked by two bevelled bands, each containing five horizontal ribs. The bow is equal in width to the footplate, arched, and decorated with two rows of punched triangles, with niello (this last feature is not visible in the photo or drawing published, but D. Csallány includes it in his description of the Fig. 6. Collegno, grave no. 6, grave goods (Longobardi 2017, based on I. 6, p. 197) object). The two outer bands of the bow each have a row of vertically aligned, engraved S-scrolls. The headplate is an incomplete semi-circle (circle sector) and is occupied completely by three settings, each with four slightly curved sides and inlaid with garnets. The headplate has three integrally cast knobs in the shape of animal heads. These knobs were mould cast and greatly resemble the animal head at the end of the footplate. Their eyes contain convex garnets. The drawing published by R. Harhoiu suggests the reverse had just one lug holding the now lost spring. An unusual technical solution was used to make the catchplate, again based on the drawing: the catchplate was cast along with the edge of the footplate and was then folded over afterwards. The length of the brooch is 5.8 cm.
This brooch was excavated in 1876, probably from a grave. However, it cannot be proved that it belonged to a set of grave goods (glass beads, and a silver-gilt, engraved buckle with stone inlays) found in the same year. The brooch was the gift of A. Bölönyi to the museum.
Museul Domoszló, Víztároló, Heves County (Fig. 2.1, Fig. 3.10, Fig. 4) 'Carefully gilt' surface, silver cast brooch pair. The footplate ends in a rounded ('turtle-like') animal head, with two prominent settings, each containing a convex garnet inlay. The ears are indicated by engraved ovals. The animal head is separated from the rest of the footplate by a Y-shaped strip. This other part is decorated with a rectangular setting with high walls containing a flat garnet. On each side is a bevelled band with horizontal ribs. The bow is strongly arched and has a thin, flat cross-section. A central, unadorned rib runs the length of the bow, and both sides of the rib are decorated with pseudo-filigree. The headplate is semi-circular and has three settings with prominent contours in cast recesses decorating the semi-circular panel. In the centre is a four-sided setting, while the two side ones are both in the shape of curved triangles; all have flat garnet inlays. Five integrally cast shafts project from the edges of the headplate and have round, high-walled settings at the end containing convex garnets. The reverse of the headplate has two lugs to support the spring axis: on both brooches this feature is strongly abraded and fragmented. The reverse of the footplate also has a much-used, worn and fragmented catchplate. The brooches are 5.4 and 5.5 cm in length.