Anglo-Saxon Button Brooches from East Kent and the Isle of Wright Typological and Genealogical Perspectives

55 Anglo-Saxon Button Brooches from EaseASt Kent and the Isle of Wight: Typological and Genealogical Perspectives seiichi suzuki Anglo-Saxon button brooches are copper-alloy disc brooches, about 2cm in diameter and decorated with a single human face mask, usually frontal and occasionally in profile. Their geographical distribution is largely limited to southern England, with a concentration in east Kent and a notable outlier in northern France. As at January 2007, the corpus contained 209 instances (Suzuki 2008, 356-81), of which 20 per cent were discovered in east Kent. Since the book manuscript went to press, however, at least nine button brooches (as at April 2008) have been discovered by metal detectorists. As recorded on the PAS (Portable Antiquities Scheme) database, the identification numbers and find places of these new items are given as follows: Serial F ind Code L ocation 1 IOW-8BD381 C albourne, Isle of Wight 2 IOW-B30872 Shalfleet, Isle of Wight 3 IOW-7F2C72 ditto 4 IOW-A35703 ditto 5 IOW-7105F4 ditto 6 IOW-AABE51 ditto 7 KENT-B5DD47 L yminge, Kent 8 HAMP-C17827 Alton, Hampshire 9 HAMP-14F866 O verton, Hampshire By integrating these new finds with the corpus of button brooches presented in Suzuki (2008), this article examines major typological and genealogical properties of the button brooches found in east Kent and the Isle of Wight, and reiterates the well-known close cultural relationship of these two regions in the late fifth and early sixth centuries. Before moving on, some notes may be needed on the typological and SEIICH I SUZUKI 56 genealogical bases of this article. By elaborating on Avent and Evison’s (1982) classification, Suzuki (2008, 8-41) proposes a system of typology that postulates twenty-two classes: A 1 B 1 C D 1 E 1 F G H I 1 J 1 K L A 2 B 2 D 2 E 2 I 2 J 2 A 3 B 3 E 3 J 3 T hese classes are in turn grouped into three series on the basis of three distinct design motifs: series 1 (with a full-face mask): classes A1, A2, A3, B1, B2, B3, C, D1, D2, E1, E2, E3, F, G, H, I1, I2 and K; series 2 (with a profile mask): class L; series 3 (with a mask between animals): classes J1, J2 and J3. C lose examination of class-internal distinctions and class-external similarities in design, with emphasis on idiosyncratic features (‘family resemblances’), then, led to the reconstruction of a genealogy or an evolutionary model of the extant button brooches (Suzuki 2008, 142-207), the relevant part of which is represented in Fig. 1. Typological and genealogical characterisation O f immediate relevance to our present concerns are the six button brooches coming from the Isle of Wight and the one from Kent, i.e. items 1-7 above. In what follows, the writer provides a typological and genealogical characterisation of each item according to the framework presented in Suzuki (2008). Shalfleet A1.2 (IOW-B30872): with the punch-decorated border zone, the rounded cheeks, the curved eye-rings and the round eyes without a sign of angularity, this object counts as a prototypical example of class A1 (Plate I). Among the twenty-five comparable instances known to date, Shalfleet A1.2 seems closest to the first two generations of the Howletts family, namely Bifrons A1.1 (Suzuki 2008, pl. 2), Bifrons A1.2 (pl. 3), Chilham A1 (pl. 7), Howletts A1.2 (pl. 16), Howletts A1.3 (pl. 17), Howletts A1.4 (pl. 18) on the one hand (Suzuki 2008, 150) – and Dover A1.1 (pl. 8) on the other (Suzuki 2008, 151).1 Among the most outstanding similarities are the eyebrows with double peaks in the centre and slight upward curves at the ends, and the well-depicted shape of the mouth and moustache, although extensive corrosion that this item suffers (apparently more damaged than Chilham A1) would make it difficult to Fig. 1 Part of the genealogy of series 1 button brooches relevant to this article (based on Suzuki 2008, fig. 4.24)XBifrons A1.1/2/Chilham A1/Howletts A1.2/3/4/Dover A1.1Calbourne A1.1/Great Mongeham A1/Howletts A1.5/Ripple A1Bifrons A2.1Boxley A2Breamore A2Howletts A2Howletts A3Alfriston A3Dover B3/Lyminge B3Droxford B3Harnham Hill B3Criel-sur-Mer A2.1/2Frénouville A2Lyminge D2Calbourne D2/Herpes D2/Rodmell D2Quidenham D2Unprovenanced D2Calbourne A1.2/Mucking A1.1/Shalfleet A1Fluy A1/Howletts A1.1Lyminge A1.1/2/3/Chatham Lines A1/Elham A1/Herpes A1Alfriston A1/Mucking A1.2/Vron A3.1/2 SEIICH I SUZUKI 58 explore further comparison on fine details. Of paramount importance in determining the genealogical position of Shalfleet A1.2, however, is the trace of double rows of punchmarks on the border. While Dover A1.1 displays a single row of punchmarks, all the six members of the first generation are provided with double-banded punchmarks. On this ground, therefore, we may be led to conclude that Shalfleet A1.2 constituted one of the earliest known examples of class A1 in evolutionary terms (see Fig. 1). Shalfleet A1.3 (IOW-7F2C72): in terms of overall appearance and representation of individual mask elements (Plate II), this item is closely paralleled by the quartet consisting of Calbourne (Isle of Wight) A1.1 (Suzuki 2008: pl. 4; to be distinguished from Calbourne A1.2, see below), Great Mongeham A1 (pl. 13), Howletts A1.5 (pl. 19) and Ripple A1 (pl. 25; see Fig. 1). Shalfleet displays the slightly curved eyebrows with Shalfleet A1.2 (diameter 18.5mm). C ourtesy of Frank Basford and the Portable Antiquities Scheme plate I. Shalfleet A1.3 (diameter 17.5mm). C ourtesy of Frank Basford and the Portable Antiquities Scheme plate II. ANGLO -SAXON BUTTONutton BROOCHE S FROM KENT AND THE ISLE OF WIGHT 59 double peaks and the less than fully rounded cheeks, two major mask constituents that are characteristic of the quartet in question (cf. Suzuki 2008, 149). The upper part of the mouth/moustache complex that forms a nearly straight line closely matches that on Great Mongeham and Ripple. Furthermore, Shalfleet’s spread mouth with a moustache is comparable to that of Great Mongeham. These specific resemblances as well as the overall common appearance may therefore lead us to assign Shalfleet to class A1. T here is, however, an outstanding feature that sharply differentiates this object from the quartet, that is the absence of punchmarks (beading). While extensive use of punchmark decoration is generally characteristic of class A1, it can be lacking, as on Alfriston (East Sussex) A1 (Suzuki 2008, pl. 1) and Mucking (Essex) A1.2 (pl. 24). In fact, in terms of the composition of distinctive features, Shalfleet A1.3 proves to be no different from the latter two objects: these three are all characterised by the clustering of properties that comprise the round eyes, the rounded cheeks and the absence of punchmarks on the border zone (more technically [+round eyes, -banded cheeks, -border 1 punchmark]; Suzuki 2008, 75). Of related interest, the addition of Shalfleet A1.3 confirms the generalisation presented in Suzuki (2008, 87) that absence of punchmarking from class A1 items is limited to those found outside Kent. Despite the absence of punchmark decoration, Shalfleet A1.3 cannot be regarded as more degenerate than the quartet in question. On the contrary, the prominent, rounded cheeks and the well-defined mouth with a definite depiction of a moustache characterise this object as closer to the earlier generations in distinction from the quartet. On balance, then, rather than forcefully differentiate the five items in evolutionary terms, we should best view them as genealogically equivalent, belonging to the same stage of artistic development (Figs 1 and 5). Of the five closely related objects under consideration here, the two coming from the Isle of Wight, namely Calbourne A1.1 and Shalfleet A1.3, prove to be unique in their own ways with respect to punchmark decoration. While Shalfleet A1.3 displays no sign of punchmarking as noted above, Calbourne A1.1 deviates markedly from the conventional manner of punchmarking: as observed in Suzuki (2008, 44), this object receives punchmarking on recessed parts, rather than on ridges as conventionally practised. More specifically, punchmarks are applied to the underside of the helmet, the perimeters of the eye-rings, of the upper lip and of the mask in itself (rather than on the outer border), and most strikingly on the opening of the mouth. In this light, it seems warranted to generalise that the norm of punchmarking in full force in east Kent was compromised in the Isle of Wight in two divergent ways: on the one hand, it was abolished completely through absolute simplification, and on the other hand it was applied extravagantly through hypercorrection. SEIICH I SUZUKI 60 Shalfleet A1.4 (IOW-A35703): this item is almost identical to Chatham Lines A1 (Suzuki 2008: pl. 5) and Mucking A1.2 (pl. 24) in overall appearance as well as in individual elements (Plate III): these three have in common a steep, straight pointed helmet, similarly shaped, single-peaked eyebrows, round eyes, curved eye-rings, a tall, subtriangular nose, full and rounded cheeks, a moustache and a small lower lip. No trace of punchmarks is in evidence. Due to extensive corrosion, however, it cannot be determined whether punchmark decoration was lacking from the beginning, as with Mucking A1.2, or it was originally provided, as with Chatham Lines A1. Because of the uncertainty on punchmark decoration, we cannot make an exact identification of Shalfleet A1.4 in genealogical terms. In the absence of positive evidence for punchmark decoration, we may tentatively assign this item to the genealogical node that includes Mucking A1.2 (Figs 1 and 5). If punchmarking was absent from Shalfleet A1.4 in the original, this would lend further support to the generalisation that obligatory punchmarking is a privilege of class A1 items of Kentish provenance. Calbourne A1.2 (IOW-8BD381): the heavily corroded surface of this item would make its precise description hardly feasible, particularly in regard to the presence vs. absence of punchmarks and the characterisation of eye-rings (Plate IV). Yet the eyebrows with a single peak without a conspicuous curling-up at both ends, the fully rounded cheeks, the relatively short nose and the outline of the mouth/moustache complex may justify us in attributing it to the genealogical group (Suzuki 2008, 149-50) that contains Mucking A1.1 (Suzuki 2008, pl. 23) and Shalfleet A1.1 (pl. 26), as represented in Fig. 1 above. This brooch should therefore be viewed as belonging to class A1, rather than class A2 as is tentatively classified on the PAS record. Shalfleet A1.4 (diameter 18.1mm). C ourtesy of Frank Basford and the Portable Antiquities Scheme plate III. ANGLO -SAXON BUTONutton BROCHE S FROM KENT AND THE ISLE OF WIGHT 61 Shalfleet B3 (IOW-7105F4): the vertical lines (hair) on the head, the almost straight eyebrows, the short, widened nose and the resultant overall appearance (see Plate V) indicate that this item is very close to Harnham Hill (Wiltshire) B3 (Suzuki 2008, pl. 99) in typological and genealogical terms (Suzuki 2008, 176).2 The more round finish of Shalfleet in comparison with Harnham Hill – as clearly shown by the eyes, the eye-rings, the cheeks and the mouth in particular – furthermore would assign the Shalfleet object to a stage earlier than Harnham Hill’s (Figs 1 and 5). In this light, Shalfleet’s distinct shape of the mouth (and moustache?) – consisting of a volute-like element (moustache and/or upper lip) and a small drop-like element (closed mouth or lower lip) – is matched by the angular and simpler counterpart on Harnham Hill. The stage immediately preceding that of Harnham Hill, however, is occupied by Dover B3 (Suzuki 2008, pl. 97) and Lyminge B3 (pl. 106) according to the current genealogical tree (Fig. 1; Suzuki 2008, 176, 242). A Shalfleet B3 (diameter 18.1mm). C ourtesy of Frank Basford and the Portable Antiquities Scheme Calbourne A1.2 (diameter 17mm). C ourtesy of Frank Basford and the Portable Antiquities Scheme plate IV. plate V. SEIICH I SUZUKI 62 comparison between Shalfleet and Dover B3/Lyminge B3 would then suggest that Shalfleet is more likely to have followed the latter through simplification and abbreviation of design than the other way round. First, Dover and Lyminge’s complex depiction of the head – a combination of vertical and horizontal lines – was replaced with the across-the-board application of vertical lines by generalisation. This generalisation was accomplished concomitantly at the expense of the double-banded, thick eyebrows on Dover and Lyminge. Second, the horizontal lines seen on Dover and Lyminge’s nose were eliminated. Third, the broad border area comprising two zones on Dover and Lyminge was substituted by a narrow strip of border on Shalfleet. Incidentally, the pervasive vertical lines characteristic of Shalfleet B3 might ultimately be traced back to the striation applied on the border of Howletts A3 (Suzuki 2008, pl. 62), a predecessor of Dover B3/Lyminge B3 (Fig. 1; Suzuki 2008, 177): the bundle of lines seen on the border of Howletts A3, a border decoration unparalleled throughout the corpus, would have been transposed onto the head, with a concomitant minimisation of the originally wide border area. More specifically, while Dover B3 and Lyminge B3 display vertical lines only at the centre and inherit the horizontal line of the head from Howletts A3, Shalfleet generalised the vertical lines to the entire head at the cost of the traditional horizontal counterpart, as remarked above. At this point, a new find from Alton (HAMP-C17827; Plate VI) brings a revealing picture on the genealogical status of Shalfleet B3. Apart from the representation of the head, the Alton brooch closely resembles Shalfleet B3, particularly in regard to the double-lined eye-rings and the peculiar Alton B3������������������3(diameter21mm). (diameter 21mm). C ourtesy of the Portable Antiquities Scheme plate VI. ANGLO -SAXON BUTONutton BROCHE S FROM KENT AND THE ISLE OF WIGHT 63 mouth/moustache configuration. Given this combination of idiosyncratic traits, it seems reasonable to classify the Alton find as class B3 and assign it to the same genealogical node as Shalfleet B3 (Fig. 5). In some other respects, however, Alton B3 differs from Shalfleet B3. Most notable are the horizontal line on the head, the relatively thick eyebrows and the almost straight substantial nose. These distinct features of Alton B3 are in turn counted among the major characteristics of Dover B3/Lyminge B3. Of particular interest is the representation of the head on Alton. Shalfleet B3 and Alton B3 turn out to be complementary in this respect. While Shalfleet generalised vertical lines at the expense of horizontal ones, Alton adopted the opposite way of simplification by eliminating vertical lines altogether in favour of horizontal counterparts. Thus, the addition of Alton B3 situates the genealogical position of Shalfleet B3 in its relation to the preceding generation on firmer empirical grounds. Lyminge D2 (KENT-B5DD47): while having an eye-ring only on the left side (as viewed, see Plate VII ), its integration with the double-peaked eyebrows and its separation from the nose forms a configuration that closely resembles the lozenge eye-mask characteristic of class D2 (Suzuki 2008, 60). The partly rounded helmet and cheeks count as further similarities to this class. Among the extant class D2 examples, Herpes (Charente, France) D2 (Suzuki 2008, pl. 130) seems closest to the Lyminge item, particularly in view of the overall round finish and the well-defined shape of the eyebrows with double peaks. What differentiates between the two, however, is the punchmark decoration that is provided on the helmet and on the border of Lyminge; in fact, none of the class D2 instances known so far (the corpus contains five examples; Suzuki 2008, 371) is decorated with punchmarks. T he existence of punchmarking, however, yields to an explanatory account when we take into consideration the genealogical status of Lyminge D2 (diameter 17.5mm). C ourtesy of the Portable Antiquities Scheme plate VII. SEIICH I SUZUKI 64 class D2, as reconstructed in Suzuki (2008, 206). As illustrated in Fig. 1, the earliest group of examples belonging to class D2 – Calbourne D2 (Suzuki 2008, pl. 129), Herpes D2 (pl. 130) and Rodmell (East Sussex) D2 (pl. 132) – occupies one of the four daughter nodes that branch off from their immediate predecessors, namely Calbourne A1.1 (pl. 4), Great Mongeham A1 (pl. 13), Howletts A1.5 (pl. 19), Ripple A1 (pl. 25) and Shalfleet A1.3 (Plate II; for the ascription of the last item to the group in question, see above). The three other daughter nodes are all represented by class A2 items – Bifrons A2.1 (Suzuki 2008, pl. 30), the pair of Criel-sur-Mer (Seine-Maritime, France) A2.1 and A2.2 (pls 37-8) and Frénouville (Calvados, France) A2 (pl. 40), respectively. Moreover, these instances of class A2 all display punchmark decoration. In this light, it would appear no less plausible to characterise the Lyminge piece under consideration as class A2 than to regard it as a protoform of class D2. Whichever classification we may adopt, however, there is little doubt that the Lyminge object must be regarded as unique in one way or another.3 If assigned to class D2, it would constitute the only class D2 instance with punchmark decoration. On the other hand, if characterised as class A2, it would make the only example of this class that has the feature composition of punchmark decoration, (partly) rounded cheeks and straight eye-rings without a sign of curving (see Suzuki 2008, 75-6). Most significant of all, whether categorised as class A2 or class D2, the representation of an eye-ring only on one side is totally unprecedented in either class. In this connection, we may as well call attention to the fact that east Kent (along with the Lower Thames) is distinguished from other regions by preponderant occurrences of punchmark decoration (Suzuki 2008, 87): only in these two regions can punchmarking be occasionally seen in the classes that are excluded from this embellishment elsewhere. At issue are Ringwould with Kingsdown C (Suzuki 2008, pl. 119), Mucking D1.1 (pl. 125) and Darenth E2 (pl. 148). These three items are the only examples of classes C, D1 and E2 that receive punchmark decoration. In this light, it seems hardly surprising that the Lyminge item, if identified as class D2, constitutes the sole example of this class decorated with punchmarks. Given the independently confirmed conspicuous appearance of punchmarks in east Kent, it seems to make better sense to categorise the Lyminge item as a marked instance of class D2 than to characterise it as an unparalleled, idiosyncratic example of class A2. Consequently, Lyminge D2 can be viewed as the earliest instance of this class known to this date (Fig. 5). Shalfleet L (IOW-AABE51): with a profile mask represented on the surface (see Plate VIII), this item belongs to class L, which constitutes a separate tradition on its own (series 2) in distinction from the classes of button brooches with a frontal mask representation (see introductory section). Overall round and smooth finish as well as the well-articulated ANGLO -SAXON BUTONutton BROCHE S FROM KENT AND THE ISLE OF WIGHT 65 plate VIII. depiction of each mask element – the helmet, eye, double eye-rings, cheek, mouth and chin – would no doubt assign this item to the same genealogical node that contains Bifrons L (Suzuki 2008, pl. 202) and Mill Hill L (pl. 206). The absence of punchmarking, however, makes Shalfleet closest to being identical to Mill Hill among the eight extant examples of class L. Accordingly, drawing on Suzuki (2008, 204), we may postulate the following genealogical tree representation for class L (Fig. 2). Shalfleet L (diameter 15.6mm). C ourtesy of Frank Basford and the Portable Antiquities Scheme Bifrons L/Mill Hill L/Shalfleet LChessell Down L/Unprovenanced L.1Frénouville LLyminge LUnprovenanced L.2Unprovenanced L.3 Fig. 2 Genealogy of class L Of particular interest, Shalfleet L is regarded as more archaic in gen-ealogical terms than Chessell Down L (Suzuki 2008, pl. 203), the only class L example previously known from the Isle of Wight. The latter is SEIICH I SUZUKI 66 subject to simplification and abbreviation of design, as evidenced most clearly in the shorter helmet, the single (as opposed to double) eye-ring and the replacement of the mouth/chin complex by a simple band. In other words, Shalfleet L constitutes the earliest instance of class L found in the Isle of Wight and occupies the same initial node as the two Kentish finds (Bifrons L and Mill Hill L). Significance of the new finds: a typological perspective T his section examines the regional distribution of button brooch classes in England and the class composition of individual regions, with a focus on east Kent and the Isle of Wight.4 Following Suzuki (2008, 130-4), the entire set of locations of button brooches in England is grouped into eight regions: (1) East Kent; (2) Lower Thames (Essex and west Kent); (3) Sussex; (4) Wessex coast; (5) Isle of Wight; (6) Wessex hinterland; (7) Upper Thames; (8) Anglia (Midlands and East Anglia). The addition of the four new A1 items (Shalfleet A1.2, A1.3, A1.4 and Calbourne A1.2) to the corpus leads the Isle of Wight to figure most prominently next to east Kent concerning the number of class A1 finds in England, as indicated in Table 1.5 Incorporating the recent finds to the corpus presented in Suzuki (2008), the table gives the occurrences of button brooches according to class in the eight regions in England.6 Although far behind east Kent in absolute quantitative terms, the Isle of Wight proves to be indistinguishable from it as far as the relative concentration of class A1 is concerned. This is brought to light when we make a cross-regional comparison of the varying frequencies of class A1 finds relative to the whole assemblage of button brooch classes involved in each region (see under %A1 in Table 1). In order to make sure that the percentage figures are statistically significant, we perform two-tailed Fisher’s exact test by comparing the occurrences of class A1 as against those of other classes, between east Kent and each of the remaining seven regions in England. Table 2 provides the p-values obtained for the seven English regions in question through comparison with east Kent. As shown in Table 2, the Isle of Wight proves to be unique by virtue of its maximal p-value, 1.000: this region displays no significant difference from east Kent with respect to the relatively high proportion of class ANGLO -SAXON BUTONutton BROCHE S FROM KENT AND THE ISLE OF WIGHT 67 TableABLE 1. REGIONAL DISTRIBUTION OF BUTTON BROOCHES IN EnglandNGLAND ACCORDING TO CLASS (AS AT APRIL 2008) E ast Kent IoW L wr Thames Sussex Wx coast Wx hinter U pper Thames Anglia T otal A1 18 6 3 1 1 - 1 - 30 % A1* 38.3 35.3 13.6 3.9 3.0 0 3.8 0 A2 10 3 3 1 1 - - - 18 A3 5 - - 1 - 3 - - 9 B1 2 1 1 - 9 - 1 - 14 B2 1 - - 12 1 - - - 14 B3 2 1 2 - 4 - - 2 11 C 1 - - 5 3 - 1 - 10 D1 2 - 3 - - - 2 - 7 D 2 1 1 - 1 - - - 1 4 E1 1 - 4 - - 2 2 - 9 E 2 - - 1 0 4 1 1 - 7 E3 - - - 3 - - - - 3 F - - - - 4 - - - 4 G - - - 1 2 2 2 - 7 H 1 3 1 - 1 - - - 6 I1 - - 2 - - - 1 - 3 I2 - - - - - 3 5 - 8 J1 - - - 2 3 2 - 1 8 J2 - - 2 - - - 1 - 3 J3 - - - - - - - - - K - - - - - - 9 - 9 L 3 2 - - - - - - 5 T otal 47 17 22 27 33 13 26 4 189 * A1 examples as a percentage of all button brooches found in region. TableABLE 2. P-VALUES OF FISHER’s exACT TEST RESULTING FROM comparisonCOMPARISON OF OCCURRENCES OF CLASS A1 AND OTHER CLASSES betweenBETWEEN EAST KENT AND OTHER REGIONS. R egion P -value Isle of Wight 1.000 L ower Thames 0.050 Sussex < 0.001 Wessex coast < 0.001 Wessex hinterland 0.013 U pper Thames 0.002 Anglia 0.284 SEIICH I SUZUKI 68 A1 to the other classes found. Taken in isolation, the p-value of 0.284 obtained for Anglia might appear somewhat comparable. However, because not a single example of class A1 comes from this region and because the total number of finds there is negligibly small, the figure may as well be dismissed. Finally, the Lower Thames barely exceeds the 95% threshold, and yet the degree of concentration of class A1 is obviously less impressive both in absolute and relative terms. Thus, we may conclude that the Isle of Wight is the only region in England that is indistinguishable from east Kent on account of the heavy concentration of class A1 in relative terms.7 Shalfleet B3 is the first find of class B3 on the Isle of Wight. The regional distribution of this class is now represented as in Table 1. As noted in Suzuki (2008, 109-10), class B3 is distinguished by its widespread distribution without a marked area of concentration. The discovery of Shalfleet B3 reinforces this pattern. Lyminge D2 counts as the first example of class D2 that has been discovered in east Kent. Disregarding a single find from Herpes and another without provenance, we now have four single instances of class D2 coming from England, distributed as in Table 1. With the addition of Shalfleet L, we now have two examples of class L from the Isle of Wight and three from east Kent (Table 1), excluding the four unprovenanced items (three continental and one insular; Suzuki 2008, 380-1). Notable about the distribution of this class is that it is limited to only two regions in England, east Kent and the Isle of Wight. As it turns out, the discovery of Shalfleet L strengthens this characteristic pattern. As with class A1, the regional distribution of class L groups east Kent and the Isle of Wight closely together in sharp distinction from other regions. The spectrum of classes represented in east Kent and the Isle of Wight Suzuki (2008, 140) pointed out that the assemblage of classes coming from the Isle of Wight largely constituted a subset of that for east Kent, except for class D2. In other words, all the classes attested in the Isle of Wight were subsumed under the whole spectrum of classes found in east Kent, with the exception of class D2, which was lacking in the latter region. Now that we have Lyminge D2, class D2 no longer counts as an exception: as it turns out, every single class that is represented in the Isle of Wight is attested in east Kent without exception, as illustrated in Fig. 3. It seems also worthy of note that the addition of Shalfleet B3 makes the class composition of the Isle of Wight still closer to that of east Kent: the number of classes represented in east Kent but not in the Isle of Wight reduces to five, namely classes A3, B2, C, D1 and E1 (Fig. 3). In this connection, it will be interesting to compare the composition ANGLO -SAXON BUTONutton BROCHE S FROM KENT AND THE ISLE OF WIGHT 69 012345678910111213141516171819A1A2A3B1B2B3CD1D2E1E2E3FGHI1I2J1J2J3KLEast KentIsle of Wight Fig. 3 Occurrences of button brooch classes in east Kent and the Isle of Wight (as at April 2008) SEIICH I SUZUKI 70 012345678910111213141516171819A1A2A3B1B2B3CD1D2E1E2E3FGHI1I2J1J2J3KLEast KentLower Thames Fig. 4 Occurrences of button brooch classes in east Kent and the Lower Thames (as at April 2008) ANGLO -SAXON BUTONutton BROCHE S FROM KENT AND THE ISLE OF WIGHT 71 of classes between east Kent and the Lower Thames. As may be recalled (see Table 1), this is the only region (besides the Isle of Wight) that would appear remotely comparable to east Kent by virtue of the relatively high incidence of class A1 in both absolute and relative terms. As Fig. 4 indicates, however, the spectrum of classes represented in the Lower Thames markedly differs from that of east Kent. The Lower Thames cannot be subsumed under east Kent with respect to the range of classes attested: as many as three classes represented in the Lower Thames – E2, I1 and J2 – are unknown to east Kent; moreover, there are five classes – A3, B2, C, D2 and L – that are attested in east Kent but not in the Lower Thames. Taken together and in conjunction with the far lower incidence of class A1 examples (13.6% as against 38.3%; Table 1), the Lower Thames can be regarded as more divergent from east Kent than is the Isle of Wight in terms of the spectrum of classes represented. Significance of the new finds: a genealogical perspective On art-historical grounds, Suzuki (2008, 142-204) reconstructed a genealogy or derivational history of button brooches with a frontal mask design (series 1), that is, for classes A1 through I2 and class K.8 The seven new finds that fall under this category have been incorporated into the proposed genealogical framework, as treated in the foregoing discussion. In recapitulation, the relevant part of the expanded genealogy may be represented as in Fig. 5. T he genealogy is of paramount interest for determining the relative chronological depth of the Isle of Wight. By far the most significant item is Shalfleet A1.2: belonging to the same generation as Bifrons A1.1/2, Chilham A1, Howletts A1.2/3/4, this object counts as one of the most archaic examples of class A1 in art-historical terms. It is still earlier than Calbourne A1.1, which constituted so far the earliest instance of class A1 coming from the Isle of Wight. Therefore, as far as the genealogical depth of button brooches is concerned, the Isle of Wight starts with the first generation, along with east Kent and the Lower Thames, as represented in Fig. 6.9 The discovery of Shalfleet B3 shifts forward the terminal of the evolutionary trajectory to a considerable extent. In the earlier account (Suzuki 2008, 240), the last generation attested in the Isle of Wight was the fifth one, corresponding to Bowcombe Down A2 (Suzuki 2008, pl. 44),10 Calbourne A2 (pl. 35), Calbourne D2 (pl. 129), Chessell Down B1 (pl. 68) and Shalfleet A2 (pl. 52). In the revised version, the genealogical depth for the Isle of Wight turns out to be the same as that for the other two regions that started with the first generation, that is east Kent and the Lower Thames, as illustrated in Fig. 6. XBifrons A1.1/2/Chilham A1/Howletts A1.2/3/4/Shalfleet A1.2Dover A1.1Calbourne A1.1/Great Mongeham A1/Howletts A1.5/Ripple A1/Shalfleet A1.3Bifrons A2.1Boxley A2Breamore A2Howletts A2Howletts A3Alfriston A3Dover B3/Lyminge B3Droxford B3Alton B3/Shalfleet B3Harnham Hill B3Criel-sur-Mer A2.1/2Frénouville A2Lyminge D2Calbourne D2/Herpes D2/Rodmell D2Quidenham D2Unprovenanced D2Calbourne A1.2/Mucking A1.1/Shalfleet A1.1Fluy A1/Howletts A1.1Lyminge A1.1/2/3/Chatham Lines A1/Elham A1/Herpes A1Alfriston A1/Mucking A1.2/Shalfleet A1.4/Vron A3.1/2 Fig. 5 Genealogical positions of Calbourne A1.2, Shalfleet A1.2, Shalfleet A1.3, Shalfleet A1.4, Alton B3, Shalfleet B3 and Lyminge D2 ANGLO -SAXON BUTONutton BROCHE S FROM KENT AND THE ISLE OF WIGHT 73 conclusion E ast Kent and the Isle of Wight share outstanding distribution patterns of button brooches in contrast to other regions in England. These common patterns are predicated on the typology and genealogy of button brooches involved. On the typological dimension, the specific individual classes as well as the range of classes represented unify east Kent and the Isle of Wight in distinction from the other regions. First, class A1 is strongly represented only in east Kent and the Isle of Wight and the relative incidences of class A1 do not show statistically significant difference between the two regions. Second, class L is attested exclusively in these two regions in England. Third, the spectrum of classes represented in the Isle of Wight is exhaustively contained in the counterpart of east Kent. Such a perfect part-and-whole relation is unknown to any other pairs consisting of east Kent and another English region. Fig. 6. The genealogical depth of English regions 0 1 2 3 4 5 6 7 8 9 10 11 generation SEIICH I SUZUKI 74 Map. 1 Distribution of button brooch finds in east Kent. 1. Bifrons, 2. Capel le Ferne, 3. Chartham, 4. Chilham, 5. Dover, 6. Eastry, 7. Elham, 8. Great Mongeham, 9. Howletts, 10. Lyminge, 11. Mill Hill, 12. Northbourne, 13. Ringwould/Kingsdown, 14. Ripple, 15. Saltwood, 16. Sarre, 17. Staple, 18. Stowting, 19. Wingham. The significance of the first two typology-based patterns could not be fully appreciated, however, before we would take into account the ANGLO -SAXON BUTONutton BROCHE S FROM KENT AND THE ISLE OF WIGHT 75 derivational history of button brooch classes. As shown in great details in Suzuki (2008), class A1 constituted the earliest class of button brooches in general and of those with a full-face mask depicted alone (series 1) in particular. Furthermore, both classes A1 and L began to be manufactured in east Kent, heavily drawing on artistic sources largely derived from Jutlandic Group bow brooches (Suzuki 2008, 245-63). We may accordingly be led to surmise that the common distribution patterning resulted from whatever cultural stimulus the Isle of Wight received from east Kent, rather than vice versa. Such a reasoning gains in plausibility when we recall that not a single example of Jutlandic brooch is located in the Isle of Wight, in contrast to their derivatives of varying removes, namely Kentish square-headed brooches. Furthermore, in view of the formative role that the Quoit Brooch Style presumably played in the genesis of button brooches (Suzuki 2008, 269-74), the absence in the Isle of Wight of objects belonging to the latest phase of the Quoit Brooch Style (Suzuki 2000, 109) – as exemplified by a large quoit brooch from Sarre (Suzuki 2008, fig. 5.3) – lends further support to the conjecture. A finer-detailed genealogy brings us a still more revealing picture. Within class A1, Shalfleet A1.2, one of the new finds under investigation here, counts as one of the earliest examples of class A1, along with Bifrons A1.1, A1.2, Chilham A1, Howletts A1.2, A1.3 and A1.4, all coming from east Kent (Fig. 5). On the other hand, Shalfleet L, on a par with Bifrons L and Mill Hill L, belongs to the first generation of class L, which forms a separate genealogy of its own (Fig. 2). Thus, the Isle of Wight attests to the initial generation of both classes A1 (series 1) and L (series 2). Since no other artefact-type of distinctively Kentish origin dated earlier than class A1 button brooches seems to have been found in substantial numbers in the Isle of Wight, we may hence speculate that the Isle of Wight became susceptible to cultural influx from east Kent with a minimal time lag after the inception of button-brooch production there. More specifically, given the initial production date of class A1 brooches approximately at ad 480 (Suzuki 2008, 257-9), the Isle of Wight became incorporated into the Kentish cultural sphere in the last two decades of the fifth century. acknowledgements T he writer owes a special debt of gratitude to Frank Basford, who keeps him updated on the discovery of button brooches in the Isle of Wight. He is thankful to Mark Stedman for allowing access to his forthcoming work on several metalwork finds from Bowcombe Down. Frank Basford and the Portable Antiquities Scheme (PAS) generously gave permission to reproduce the images of the brooches treated in this article, as indicated under the Plates. SEIICH I SUZUKI 76 bibliography Avent, Richard and Vera I. Evison, 1982, ‘Anglo-Saxon Button Brooches’, Archaeologia, 107, 77-124. Stedman, Mark, forthcoming, ‘Seven Early Anglo-Saxon Metalwork Finds from Bowcombe Down, Carisbrooke and Medina Parish, Isle of Wight’. Suzuki, Seiichi, 2000, The Quoit Brooch Style and Anglo-Saxon Settlement : A Cast-ing and Recasting of Cultural Identity Symbols, Woodbridge: Boydell Press. Suzuki, Seiichi, 2007, ‘Les button brooches de Picardie et du Pas-de-Calais: Aperçus typologique, «généalogique» et chronologique’, Revue archéologique de Picardie, 2007, 3/4, 91-110. Suzuki, Seiichi, 2008, Anglo-Saxon Button Brooches: Typology, Genealogy, Chronology, Anglo-Saxon Studies 10, Woodbridge: Boydell Press. endnotes 1 Each individual button brooch is designated by its find place (e.g., Shalfleet) followed by the class to which it belongs (e.g., A1) and, if more than one example of the same class membership come from the same location, an identification number (e.g., .2). For conversion between Avent and Evison’s (1982) system of designation and the writer’s, see Suzuki (2008, 345-55). 2 In this regard, the writer would not follow the categorisation of this brooch as class I2 on the PAS record. The depiction of the hair by vertical lines is indeed an almost ubiquitous feature of class I2. And Long Wittenham (Berkshire) I2 could be adduced as an interesting parallel to Shalfleet. As far as the diameter is concerned, however, the brooch in question is measured 18.1mm. By contrast, class I2 is much larger with a statistical significance, with a mean of 25.375mm and a standard deviation of 1.768mm (Suzuki 2008, 90). This physical property, along with the design features mentioned in the text, would therefore favour the classification proposed in the text. 3 The PAS analyst’s hesitation in categorizing this item between classes A3 and B (without further specification) can be ascribed in large part to the absence of its close parallels in the existing corpus. 4 For a comparable account of the regional distribution in France, see Suzuki (2007). 5 It is interesting to note that not a single example of class A1 was known to Avent and Evison (1982). In fact, it was only in 2005 that the first instance of class A1 to be found in the Isle of Wight came to light, namely Shalfleet A1.1 (Suzuki 2008, pl. 26). 6 The new find from Overton (HAMP-14F866; see the introductory section) should better be identified as class E2, rather than class E1 as on the PAS record. Overton is subsumed under the Wessex hinterland. In this connection, figs. 3.24a-h in Suzuki (2008) need to be updated and corrected. 7 Outside England, Charente (Herpes) shows a fairly high proportion of class A1 (one example of class A1 as against four others; p = 0.641). 8 For class L (with a profile mask; series 2), see Fig. 2 above. A third genealogy is postulated for classes J1, J2 and J3 (series 3), namely brooches with a motif of a mask between animals (Suzuki 2008, 198-200). 9 Charente (France) also began with the first generation (Suzuki 2008, 240). 10 In Suzuki (2008) this item was imprecisely referred to as Isle of Wight A2 without further provenance. It should now be renamed Bowcombe Down A2. For details on the context this brooch comes from, see Stedman (forthcoming). The writer is profoundly grateful to Mark Stedman for giving access to his forthcoming work.