Upload
oldenglishblog
View
163
Download
0
Embed Size (px)
DESCRIPTION
"A Study of Cross-Hatched Gold Foils in Anglo-Saxon Jewellery" by Richard Avent and David Leigh appearing in Medieval Archaeology, Vol.21 (1977)
Citation preview
A Study of Cross-Hatched Gold Foilsin Anglo-Saxon Jewellery
By RICHARD AVENTInspectorate if Ancient Monuments, Department of the Environment
and
DAVID LEIGHDepartment of Archaeology, University College, CardijJ
I N THE last few years interest has focused increasingly on various aspects ofthe manufacture of migration period and early medieval jewellery. The goldcontent of this jewellery has been the subject of chemical analyses, the produc
tion and use of garnets has been reassessed, and casting techniques are beingreviewed in the light of the workshop discoveries at Helga, Sweden.1 The presentstudy is concerned with the wafer-thin pieces of cross-hatched gold foil used asbackings for garnet and glass settings on jewellery of the 6th and 7th centuries.The foils give life and brilliance to the inlays by virtue of the glittering effectproduced by their many-faceted surfaces; an effect which on jewellery wouldhave been enhanced by the movement of the wearer." It is a technique now replaced by the faceting of diamonds and other precious stones, and yet stillemployed, though to markedly different ends, in the rear reflectors of motor cars.The use of gold, although partly dictated by its technical p roperties'' and itsavailability as foil! also adds a depth of colour to otherwise pale garnets and,
1 On gold analyses, S. C. Hawkes, J. M. Merrick, D. M. Metcalf, 'x-ray fluorescent analysis of somedark age coins and jewellery', Archaeometry, IX (1966), 98-138; P. D. C. Brown and F. Schweizer, 'x-rayfluorescent analysis of Anglo-Saxon jewellery', Archaeometry, xv (1973), 175-92; W. Holmqvist, et al.Excavations at Helgo, IV, Workshop, pt. I (Stockholm, 1972); and for instance K. Larnm, 'The manufactureof jewellery during the migration period at Helga in Sweden', Bull. Historical Metallurgy Group, VII (ii)(1973), 1-7; T. Capelle, H. Vierck, 'Modeln der Merovinger- und Wikingerzeit', Friihmittelalterliche Studien,V (1971), 42-100; T. Capelle, H. Vierck, 'Weitere Modeln der Merowinger- und Wikingerzeit',Friimittelalterliche Studien, IX (1975), 110-42. Of particular relevance to the present study, B. Arrhenius,Granatschmuck und Gemmen aus Nordischen Funden des fruhen Mittelalters (Stockholrn, 1971). See also S. C.Hawkes, 'The Monkton brooch'. Antiq. ]nl., LIV (1975), 244-56.
2 Although most settings are of garnets, there is a small proportion in which pink or red glass has beenused in imitation. For the purposes of the present study we have made no distinction between garnets andother settings. It may be asked what advantage garnets have for jewellery over glass. One answer may bethe somewhat higher refractive index of the former (garnet c. 1.8; glass, c. 1.5). This results in greaterrefraction of incident light rays by the setting and thus a smaller critical angle for internal reflection,causing more light to be bounced back and forth between the foil and the top surface of the garnet; inshort, creating more sparkle.
3 See below, p. 3.• See below, p. 3.
I
2 RICHARD AVENT AND DAVID LEIGH
a
e • f
b
FIG. 1
c
9I
h
d
FOIL PATTERNS (STYLIZED) Sc. c. 10: 1
a, standard; b, boxed; c, d, special boxed; e,f, irregular patterns;g, ring-stamped; h, lozenge pattern
indeed, its absence on certain pieces of jewellery only serves to emphasize this.The inlays are found in settings formed either by raised cells cast in one with theobject, or as the cells of a more elaborate cloisonne mosaic.
The foils have a thickness of about 0.03 mm. The ridges or grooves whichmake up the cross-hatched patterns are of a similar depth and are spaced atbetween 2 and 4.5 lines to the millimetre. These patterns may be divided intotwo main groups which we refer to as standardand boxed. The ridges (or grooves)of the standard foils (PL. I, A, B; FIG. I, a) form a simple uniform pattern ofsquares. Boxed foils (PL. I, c; FIG. I, b) have wider and deeper ridges (or grooves)surrounding a number of standard squares, so that the overall appearance is oflarge squares with smaller squares, usually nine in number, within them. Occasionally, however, sixteen or twenty squares appear within each box, and we callfoils having such a pattern special boxed foils (PL. I, D; FIG. I, C, d).
In this paper we present the results of a survey of over 500 pieces of foil on181 objects and we seek to answer two questions. First, how were the foils produced? Secondly, is there any possibility ofidentifying individual foil patterns and,if so, of relating these to the objects, or groups of objects, which bear them? Wehope that in answering these questions we may begin to approach an understanding of the workshop practice and organization of migration-period jewellers.However, before we can progress to a presentation of the survey evidence and itsanalysis and interpretation, we must attempt to establish the method of manufacture, for unless we can confine and unify our preconceptions on this score itwill be difficult to draw reliable conclusions from the mass of accumulated data.
CROSS-HATCHED GOLD FOILS 3MANUFACTURE
Gold is an easily worked metal and from the earliest times must have presented little difficulty to smiths, who could have hammered it, "Without annealing,into a foil." There are to our knowledge no published chemical analyses of thefoils studied here, nor more than the occasional analysis of other types of gold foilsuch as brocading strips forming gold braids in head-dresses and on the bordersof garments. 6 However, we may speculate that, at least in the later part of ourperiod, the composition of the alloy used was similar to that used for the manufacture of solid gold jewellery of the same date, for which there is now an extensive body of analyses which indicates a silver content of generally about IO to 40per cent and a copper content of a few per cent. 7 This in turn is thought to reflectthe composition of continental gold coinage of the late 6th and the 7th centuries."Alloys of this sort, although intended primarily for coinage, not for jewellery,would have possessed the necessary properties for being beaten out into a thinfoil and its subsequent working in one of the ways we postulate. That thesemechanical properties were highly prized may perhaps be concluded from theobservation that such gold foil was being used at a time (the 6th century) whengold was not readily available in England." In fact, in the 6th century the onlyother principal use of gold, apart from the braids already mentioned, was forgilding silver or copper alloy objects, which involved only a very sparing use of themetal.
It may be asked whether the gold foil used in the braids could have been thesame as that used for our cell inlays. A conclusive answer to this must awaitchemical analyses of the two sorts offoil, but it is worth noting that the thicknessesof a few measurable examples are within one order of magnitude: three loosefoils from Taplow have mean thicknesses of 0.03 I mm. each (at points where thepattern no longer exists); thicknesses of some braids from Faversham are 0.065and 0.053 mm. (no. ii), and 0.060 mm. (no. iv).lO These thicknesses are far inexcess of the thickness of modern gold leaf (0.0001 mm.) or even Roman goldleaf (0.0002 mm.),l1 but similar to that of, say, modern aluminium (cooking) foil(0.025 mm.).
We consider there to be three possible ways in which these foils could havebeen manufactured:
i) A sheet of hammered gold foil was placed on a supporting but slightlyplastic surface and the evenly spaced grooves were ruled on to it with a suitable
6 R.]. Forbes, Studies in Ancient Technology, III (Leiden, 1964), 154.6 E. Salin, 'Les tombes gallo-romaines et mcrovingicnnes de la basilique de Saint-Derris (fouilles de
]anvier-Fevrier 1957)', Memoires de I'Academie des Inscriptions et Belles-Lettres, XLIV (1958), 48, cited in S. C.Hawkes and E. Crowfoot, 'Early Anglo-Saxon gold braids', Medieval Archaeol., XI (196 7) 44.
7 Hawkes, et al., op. cit. in note I; Brown and Schweizer, op. cit. ibid.• Brown and Schweizer, op. cit. in note I;]. P. C. Kent, 'Gold standards of the Merovingian coinage,
A.D. 580-700' in E. T. Hall and D. M. Metcalf (eds), Methods of Chemical and Metallurgical Investigation ofAncient Coinage (Royal Numismatic Society Special Publication, London, 1972),69-74.
9 R. Avent, Anglo-Saxon Garnet Inlaid Disc and Composite Brooches (British Archaeological Reports, XI, Oxford,1975), i, 8.
10 Those pieces of braid described and numbered as here in Hawkes and Crowfoot, op. cit. in note 6,69.11 Forbes, op. cit. in note 5, 181, using Pliny the Elder, Natural History, XXXIII, xix, 61.
2
4 RICHARD AVENT AND DAVID LEIGH
tracing tool. The foil was later cut up to fit the cells. There are a number ofobjections to this theory. Such a technique would place considerable strain on thefoil at the point of contact with the tool and would almost certainly result insome pulling, crinkling or even tearing of the foil. Secondly, it is difficult to suggesta supporting medium which would have the necessary mechanical properties topermit the tracing of the sharp-contoured groove that we see without bending thefoil downwards on either side of the groove or, at the other extreme of hardness,risking cutting the groove through to the fax side of the foil - an accident nevernoticed in our observations of foils. Moreover, it seems unlikely that such atechnique could result in so even-depthed a groove as seems generally to be thecase. Finally, and convincingly, to the authors at least, the patterns do not lookas if they were traced.
ii) Maryon suggested that the rows of squares forming the patternwere made by impressing a specially notched chisel-like tool and that the patternwas built up by repeated blows of this tool at a regular spacing all over the surface.l" This would seem to us to be a most laborious and, at the required scale,almost impossible process, but in any case it is plain to see, under even low-powermagnification (e.g. PL. I, B) that the ridges (or grooves) are smooth unbrokenlines, that the pyramids are there by default, as it were, and that the lines are thedominant feature.
iii) The pattern could have been stamped on to foil laid on a firm but slightlyplastic surface; sheets made in this way could have been cut up later to fit thecells. This is a technique which tallies with the visual appearance of the foils,and which we suggest was the method used.
Convincing evidence comes from the occasional foil where it has been possibleto identify one stamped impression abutting the next, suggesting that a largesheet offoil had been stamped more than once. Examples of this can be found on aClass 3.2 keystone garnet disc brooch from Faversham (Catalogue no. 38) andon a square-headed brooch from Gilton (Catalogue no. 84). On another squareheaded brooch from Bifrons (Catalogue no. 80) the grooves of the two separateimpressions are not properly aligned and this seems to reflect a careless stampingof the original sheet of foil. On a Class 4 keystone garnet disc brooch from Sarre(Catalogue no. 44) the two impressions are set at completely different angles toone another and this provides further, and more or less conclusive, evidence thatthe pattern must have been stamped and not ruled on to the sheet of foil.
If this explanation is accepted, it remains to debate the material of which thestamp was made. The most likely candidates seem to be copper alloy, lead, woodor bone. Hardened iron tools would have been available for engraving the patternin copper alloy, but there seems to be little other evidence at this time for metalengraving. Had this method been used, it seems probable that some metal burrwould have remained on the stamp and that the impressions of this could havebeen, but were not, microscopically detectable on the surviving stamps. In addition, it would have been extremely difficult to cut fine lines, of the type found
12 H. Maryon, Metalwork and Enamelling (5th edn. London, 1971),79.
CROSS-HATCHED GOLD FOILS 5on these foils, with such smoothness and regularity into this type of material.There is, however, the possibility of a cast copper alloy having been used, thepattern or model for the casting having been made in some more plastic materialsuch as wood or wax. We have evidence for the use ofcopper alloy matrices for theproduction of repousse decoration.!"
Lead might have provided a suitable medium in which to cut the pattern~certainly softer than copper alloy - but there might still have been some tracesof metal burr and, ifan engraving tool was used, it would remain difficult to cut asufficiently even line. On the other hand, if the lines had been traced, the cuttingtool would probably have blunted rather quickly, although this may not be asubstantial objection. The chief objection must be to the very softness of leadwhich would make it deform and quickly lose its pattern after relatively few blowson to the foil.
Far more likely and more obvious materials must surely be wood and bone(or antler), which seem to us to possess the properties needed to allow fine, evendepthed lines to be cut into them with ease. We should not wish to choose betweenthe two materials. Bone and antler are known to have been used for fine decorativework at this period (e.g. bone trial pieces and decorated bone combs). Wood,though rarely surviving, is likely to have been used for similar purposes. We knowthat fine-grained woods were available: such as yew, for making bucket stavesand bows; and no doubt also box, used extensively by the Romans for fine woodwork such as combs, and likely to have been available in this country throughoutour period.l!
There is some evidence which inclines us to think that a fine-grained woodwas used, at least for the manufacture ofsome foils. Close inspection ofa number offoils shows there to be, superimposed on the intended pattern, a much finer andmore faint pattern of roughly parallel, but not always straight lines which seem tocorrespond with the grain of a fine-grained wood (PL. II, D). In the course ofmaking our survey we were not looking for this feature and it may be that it ismore common than might be suggested by the four or five examples whieh wehave noted, although it is by no means universal. One possible alternativeexplanation for the grain effect might be ivory, but, since it is much harder thanwood, it is unlikely that its much finer grain could be transferred to the gold foil.Furthermore, the spacing of the grain lines on the foils of about fifty lines to themillimetre, corresponds most closely with the fine spacing of the grain on a finetextured wood (the lines within the growth rings, rather than the growth ringsthemselves). However, a third possibility is that the lines are striations associatedwith the finishing stage of foil manufacture.
In order to test this theory some experiments were performed; these aredescribed in the Appendix and some of the results illustrated in PL. III, A, B. Theexperiments showed that it is possible to stamp metal foil with the desired patternby using a stamp of fine-grained wood. We do not claim to prove that this was the
13 T. Capelle, H. Vierck (1971), op. cit. in note 1,66 If.; idem (1975), op. cit. ibid., 123 f.14 H. Godwin, History of the British Flora (Cambridge, 1956), 181.
6 RICHARD AVENT AND DAVID LEIGH
method used, but the possibility would seem to be strong. Other materials mayequally well have been used: bone, as already mentioned, and copper alloy whichcould have been made from a wooden model. A metal stamp would have had alonger life than a wooden one. However, it is debatable whether a casting couldreproduce such a fineness of detail as was necessary, and indeed our observationsof foils have not revealed any minor blemishes such as one would have expectedalmost certainly to survive.
We have also shown that it is possible to rule by eye thin parallel lines withlines spaced evenly at about three per millimetre; that using this method stampedpatterns were unlikely to be more than about 20 mm. across; and that boxedpatterns could be created by cutting extra deep lines on to a standard pattern.
We have no way of knowing how large a foil was before it was cut up formounting into cells. Commonsense and our observations, however, conspire tosuggest that sheets of foil of some size, say 50 mm. square at least, were stampedall over by repeated blows of the stamp, and that the sheet was cut as requiredto the desired shape formed by the cell walls, generally discarding edges, junctions and other blemishes. To cut plain foil to such small sizes (rarely more than10 mm. across, and usually less), and then to attempt to stamp it would have beena singularly tiresome process; the safe handling of tiny pieces offoil without loss ordamage is clearly extremely difficult. What has become plain from our observations is that the cutting of the sheet foil into pieces was not always carefullyperformed. In instances too numerous to mention, we have seen the shapes of thepieces only roughly corresponding with the shapes of the garnets above.
FOIL SURVEYS
Our findings are based on two sets of measurements of the spacings of theridges (or grooves) on the foils, expressed in lines per millimetre (1. Imm.). Thefigures were derived by counting the number of lines viewed over a convenientnumber of millimetres, either directly through a microscope (Group I) or indirectly, on photographs (Group 2). Group r (TABLE r) is the larger set ofobservations made on every visible piece of foil on each of r69 objects from sixmajor collections." The objects include round and square-headed brooches,pendants, buckles, spoons, etc. The smaller set of readings, Group 2 (TABLE III) isbased on accurate and repeated measurements on every visible piece of foil onhigh-quality photographs of thirty-one square-headed brooches, including nineteen from the Group r survey. The two methods have produced slightly different,though still comparable, results for the same set of brooches, and they thusprovide a valuable cross-check. Slight differences are to be expected when usingdifferent techniques of measurement and must be due to small systematic errorsin one or both of the techniques which cannot readily be detected. As will be
15 The six collections are in the Ashmolean Museum, Oxford; Merseyside County Museums, Liverpool;Maidstone Museums and Art Gallery; Royal Museum and Public Library, Canterbury; Department ofMedieval and Later Antiquities, British Museum; and the Maison Dieu Collection which is temporarilyhoused in the Ancient Monuments Laboratory, Department of the Environment, Fortress House, London.
CROSS-HATCHED GOLD FOILS 7seen, however, they do not stand in the way of general observations on groupingsand distributions."
It may assist the reader to be informed of the terminology to be used in thefollowing discussion. Where a piece of foil is so placed in its setting that its ridgesappear to be upstanding we denote its position as being positive (P). Where it isplaced so that it displays grooves, rather than ridges, we call its position negative(N). Where it is immaterial whether we speak of grooves or ridges we use theword line to stand for either.F When we write ofsheet foil we refer to an (imagined)uncut sheet of cross-hatched foil, having an assumed constant pattern, and fromwhich cellfoils have been cut to fit the cells.
SURVEY GROUP I
TABLE I contains the results of the larger survey of foils using direct microscopic measurements. In FIGS. 3 to 7 these results are summarized in the form ofhistograms.
The method of measurement used was to view the foils through their garnetor glass inlays using a binocular microscope which had a graticule fitted into oneeyepiece." The graticule had at its centre two crossed hairlines, set at rightangles, and graduated in millimetres (FIG. 2). Working at a magnification of X 20
with suitable illumination it was possible to count the number of lines to a givennumber of millimetres. The microscope was set so that one foil line coincidedwith a fixed point on a hairline and lines were counted (counting the first line as I)up to the 2 mm. mark and then, if the foil was large enough, up to the 4 mm.mark and, where possible, up to the 6 mm. mark. Although it was sometimespossible to infer an exact half spacing, in which case 0.5 was added to the lastwhole number of lines, it was usual to count to the line nearest to the chosengraduation. The process was then repeated in the direction at right angles to thefirst. One set of such measurements was made on all visible foils on each object.Sometimes this might mean one foil; more often several foils. The spacing in eachdirection was simply calculated by dividing the line count by the relevant numberof millimetres. The method provided measurements with an accuracy of between2 and 12.5 per cent, depending on the size of the cell and hence the distanceover which measurements could be made.l"
16 See below, p. 42.17 It appears to have been a purely arbitrary decision whether the foil was placed in its setting in a
negative or positiveposition. The Group I Survey shows that slightly more foils were in a positive rather than anegativeposition. However, in general, all the foils on a particular object tend to be the same way up.
18 A Myacope 1000 binocular microscope was used for this work. The accuracy of the method was checkedby observing a separate microscope-stage graticule through the microscope. Furthermore, any possibleerrors caused by refraction oflight by the garnets could be discounted by carrying out a few measurementson foils which, by accident or loss, were not overlain by garnets, and then comparing these with measurements on foils which were so overlain, on the same object.
19 The major source of experimental error is the accuracy of line counts which was rarely closer than tothe nearest line. Over a distance of 2 mm. this represents an error of ± 0.25 l./mrn.; over 4 mm. an error of± o.r j l.jmm.; and over 6 mrn., an error of ± o.od Ljrnm, These equate errors ranging from, at worst,± 12.5% (on the low spacing of z l.jrnm. read over a distance of 2 mm.) to, at best, ± 2% (on the highspacing of 4l./mm. read over a distance of 6 mm.).
8 RICHARD AVENT AND DAVID LEI GH
FIG. 2
VIEW THROUGH MICROSCOPE FOR READINGS OF SURVEY GROUP I
In the TABLE I we have condensed the readings somewhat. For each foil onlythe most accurate reading is recorded (i.e. if there were readings at 2 and 4 mm.,only the latter reading is used). The error stated is that associated with the mostaccurate reading. The readings in each direction on a foil were usually identical,and we have not therefore presented both sets, except where they differ (italics).
For the purposes of graphical representation, i.e. in the histograms, we havetaken a number ofliberties. Firstly, we represent each object by one reading only;where counts differed in each direction on a foil we have averaged them. Thecounts rarely differ by more than one line in 6 mm., so the loss in informationis minimal, but, in any case, is retained in the Table. Where several foils on anobject show similar, but not quite identical, sets of readings, and appear to havebeen produced by the same stamp, we have averaged them, for the purposes of thehistogram, working on the assumption that, being on one object, they are likely tobe the same foil. This happens only a few times, and the small risk of confusion isoverriden by the increase in simplicity of presentation. The more accuratemeasurements from Survey Group 2 confirm that such an assumption ispermissible.
As in all histograms the height of each bar indicates a frequency, in this casethe number of objects bearing foils with particular spacings (lines per millimetre).In slight contrast to other histograms, however, the widths of the bars are used toindicate the degree of error associated with each reading. Thus the wider a barthe greater the error associated with its central value and the more uncertain we
CROSS-HATCHED GOLD FOILS
NUMBER 35
of OBJECTS
9
30
25
20
15
10
45 5 SPACINGlines! mm.
~ combined with another standard foil ~ combined with a boxed foil
~ combined with a special boxed foil
FIG. 3
FOIL SURVEY GROUP I
Histogram for objects bearing standard foils
are of the true spacing represented. The overlapping bars thus convey the imprecision of our measurements as well as the variation of line spacing across thecell foils in an object. Furthermore, we have distinguished on the histograms thoseobjects on which more than one sheet foil has been used. Lastly, in assessing thehistograms we must bear in mind the predominance of objects from Faversham inthis survey group. They represent 36 per cent ofall the provenanced objects studied.
From FIG. 3, showing the overall distribution of standard foils, we observethat within the spread of spacings from 1.25 to 5 l./mm. most foils fall in the range3.25 to 4.0 l./mm., with a tendency towards the coarser end of that range (about3.5 l./mm.). The evidence for boxed foils is summarized in FIG. 4 and, allowingfor the smaller sample size, we see a much narrower spread of spacings (2.75 to41./mm .) with a much more pronounced peak at 3.25 l./mm.
The information can be analysed in more detail by dividing the readings forstandard foils according to the category of object on which they are found. InFIG. 5 keystone garnet disc brooches are presented by Class, while the otherobjects are grouped in a more general way. All the disc and composite broochhistograms arc set out in an approximately chronological order, starting with the
10
"NUMBERof
OBJECTS
20
"
'0
RICHARD AVENT AND DAVID LEIGH
~ combined with standard foil
'-5 zs 3-5 4-5
SPACINGlines Imm
FIG. 4FOIL SURVEY GROUP I
Histogram for objects bearing boxed foils
Class I keystone garnet disc brooches and ending with composite brooches, anorder which obscures the overlapping periods of many of the brooches but whichis fairly reliable.s" In general terms all the objects represented in FIG. 5 fit into adate-range beginning early in the 6th century and ending around the middle ofthe 7th. By so dividing the evidence we are left with a relatively small sample foreach group and consequently we can derive only somewhat tentative conclusions.Nevertheless, certain tendencies are observable from these histograms:
i) Earlier items employ more foils in the 4 l.(mm. region, but as time goes onthere is a movement towards coarser spacings of about 3.25l.(mm.
ii) For all objects other than buckles, there is a notable paucity of foils in the2.75l.(mm. region.
iii) Among the earliest keystone garnet disc brooches there are no foils withspacings over 4l.(mm. (± 0.25)·
iv) Foils on composite brooches show little variation in their spacings, nearly alloccurring at 3.25 l.(mm.
v) The square-headed brooch foils are concentrated in the 3.75-4 1.(mm.region, and have an overall distribution which places them in line withkeystone garnet disc brooch Classes I and 2.
vi) Apart from the progression implied in (i) above, there is no significantchange in the range of standard foil spacings throughout the period underdiscussion.
20 Avent, op. cit. in note 9, Table 8.
7squareheadedbrooches
buckles
,0 composite
K7
Keystone (K)
Ks
K.
'·5 3·S
,'"'
~ combined with another standard fad
SPACINGlinea] mm
~ combined with a boxed foil
~ combined with a special boxed foil
FIG. 5FOIL SURVEY GROUP I
Histograms for objects bearing standard foils, according to category of object
12 RICHARD AVENT AND DAVID LEIGH
In interpreting these observations it is tempting to equate peaks in thehistograms with identifiable sheet foils; to say that if there is a large number offoils at, for instance, 3.25 l./mm., and a slight spread on either side of the peak,then we have identified a sheet foil with roughly this spacing. Unfortunately, it isimpossible, using this set of readings, to determine whether each occurrence at apeak value represents a number of different stamps with closely similar spacing,or one stamp used repeatedly. It may well be that certain stamps continued in usethrough many decades, a view to which we subscribe, or, on the other hand, itmay simply be that fashion, tradition or even the use of certain units of measurement (see below p. 29) ensured the repeated appearance of certain spacings orgroups of spacings. The gap in the readings at about 2.75l./mm. and the existenceof definite peaks which appear to vary with time, however, imply not that we aredealing with a more-or-less even spread of spacings about some general mean (animplication we might have guessed at from the overall histogram, FIG. 3), but thata limited number of spacings was being aimed at. It is less easy to infer the precisemeasurements, but we might suggest one foil or group offoils produced at 3.75 to4l./mm., and others at about 3.5,3.25 and 2.75l./mm.
The fact that the distribution offoils on the square-headed brooches parallelsso closely those on keystone garnet disc brooch Classes I and 2 is welcome confirmation of the potential usefulness of this line of enquiry, since it is readilyaccepted that there was a chronological overlap between these types of objects;and indeed the occurrence of Class 2. I discs affixed to the bows of square-headedbrooches from Howletts and Dover (TABLE I, Catalogue nos. 77, 162) serves toconfirm this.
When we turn to the boxed foils, with histograms similarly apportioned(FIG. 6) we find somewhat more precise evidence. As has already emerged, thereis a marked preponderance of boxed foils in the region of3.25l./mm. We now seethat they occur at this spacing throughout our chronological range, with littlevariation between one type of object and another. The contrast between thispicture and that provided by the standard foils is so marked that we must considerthe possibility that in many cases we are seeing repeated use of just one or twostamps which had spacings of 3.25 l./mm. ± 0.2 mm. If these had been thehistograms of standard foils, we might be less confident, but the boxed foil,containing nine little pyramids, is so characteristic that it seems unlikely that verymany stamps would have been created with this particular spacing. The uniqueness of this foil is further testified by evidence discussed below.
In FIG. 7 we have taken those cemeteries where the incidence of foiledobjects is statistically sufficient and have plotted their histograms for standard andboxed foils site by site. The sites are: Howletts, with no object later than the 6thcentury; Kings Field, Faversham, which was in use from at least early in the6th century until well into the 7th century, wi th its floruit in the late 6th andearly 7th centuries; Sarre, with a long date-range from early 6th to mid 7thcentury; and Gilton, Ash, dating probably from the late 5th to the mid 7thcentury. The approximate correspondence of the histograms with these generaldate ranges is thus seen to provide confirmation of the chronology implied by our
CROSS-HATCHED GOLD FOILS
scareheadedbrooch
pNUMBER ']
of OBJECTS ~I -r-'- +-_,-- -,-__~ --.
Keystone (K)
composite
buckles
pendants
plated
o ,
~ll----~---r---~--,--JD'----.-----,---~---,
:l [b~_:1 Q'----~,----,-------,
~ll__-~_,_-~-,--'----'-~-,--~_____,oj-~-----.---~----,---.-----~----.-~---,
fl~----,.---,-----"---L,----DL-,---------,-----,------,K5
ol-_-~-_-----,-~----.-_~_---.
'·5 '·5qCJ
3 3·5 ~ 4·5
SPACING - hnes Imrn.
~ combined with standard foil
FIG. 6
FOIL SURVEY GROUP I
Histograms for objects bearing boxed foils, according to category of object
earlier histograms (FIG. 5) where we were able to show an apparent progressionfrom finer to coarser spacings with time. Howletts, the one undoubtedly earlycemetery, shows a preponderance of standard foils of the finer variety, 3.75-41./mm. Faversham, with the cemetery continuing rather later, shows a preponderance offoils with coarser spacings, especially around 3.5 l./mm., as well as arespectable proportion of finer foils. The histograms for Sarre and Gilton are notconclusive in this respect, but correspond roughly with the suggested development.
RICHARD AVENT AND DAVID LEIGH
Gillonboxed
CJNUMBER ~l
ofOBJECTS 0 f---------.----~---_._.L--.L.-~---r_---~---r
Gillonstandard
Favershamboxed
15
10
Favershamstandard
B combined wilh standard toil
'-5 4
~ combined with boxed fGil
5 5-5
SPACINGJinesImm
FIG. 7FOIL SURVEY GROUP I
Histograms for objects bearing standardfoils and for objects bearing boxed foils,according to provenance: Faversham, Howletts, Sarre, Gilton
CROSS-HATCHED GOLD FOILS
The site histograms for boxed foils (FIG. 7) need occasion little furthercomment, since they display the regular appearance of the 3.25 I.jmm. foil towhich we have already pointed. However, the high incidence of boxed foils atFaversham is worth noting. They occur there eighteen times out of a total ofthirty provenanced occurrences, i.e., 60 per cent, which should be compared withthe overall incidence of Faversham objects in relation to all provenanced objectsin this survey, namely, 36 per cent. This clearly signifies a close relationshipbetween this foil and the Faversham cemetery.
One of the most distinctive types of foil is the special boxed foil (PL. I, D;FIG. I, C, d) on which the pattern consists of sixteen or twenty small squareswithin each box. Examples of this foil are found on seven objects'" and wereevidently produced by a stamp (or stamps) having a rather irregular pattern ofrf-unit boxes (4 X 4) interspersed with occasional eo-unit boxes (5 X 4). Wherewe can see the transition (PL. I, D) it becomes apparent that the craftsman'soriginal intention was to produce a tfi-unit box, but that in some places his boxforming lines diverged from their correct paths, thus producing zo-unit boxes.This highly characteristic pattern has a line spacing centred in the region of 3.5to 3.6 l.jmm., and variations in the readings may be attributed to the irregularityof the pattern. We may have here identified a unique foil.
It seems likely to us that individual stamps would have been used repeatedlyover a fairly long period of time, and this view appears to be confirmed by thedate-range of those objects bearing the special boxed foil. The silver spoon fromBifrons (Catalogue no. 163) and the square-headed brooch from Sarre (Catalogueno. 164) should be dated to around the middle of the 6th century, while the Class6.1 keystone-garnet brooches (Catalogue nos. 158, 159) and the silver pendantfrom Dover (Catalogue no. 161) should be dated to the latter part of the 6th andthe early 7th centuries."
Of the six provenanced occurrences of this foil, we note that four are onobjects from Dover. It has been possible for us to examine the foils on a total often objects from Dover, all of which, with one exception (Catalogue no. 154),came from the same site in the town.s" Three of these have standard boxed foils:one, a mixture of standard and ordinary boxed foils; four, the special boxed foil;and the remaining two have unusual and distinctive foils which have beenentered in the miscellaneous section of TABLE 1. Amongst these the four keystonegarnet and the two plated disc brooches are characterized by a particularly finestandard of workmanship, in comparison with other brooches in their respectiveclasses, and it may be that there is some connexion between this and the use ofunusual foils at Dover. Lastly, one point should be made about the square-headedbrooch from Dover (Catalogue no. 162). The fact that the special boxed foil isfound on both the main part of the brooch and on the disc attached to its bow
21 Catalogue nos. 158--64.22 Avent, op. cit. in note 9, 36, 59, 62.23 The exception, a composite brooch, is thought to have come from the Priory Hill cemetery: S. E.
Rigold and L. E. Webster, 'Three Anglo-Saxon disc brooches', Archaeologia Cantiana, LXXXV (1970),13-17.The other objects come from the Old Park cemetery.
TA
BL
EI
FO
ILS
UR
VE
YG
RO
UP
I
Foi
lN
um
ber
ofl
ines
Set
ting
spo
siti
onp
erm
m.
Err
or
Cat
alo
gu
eC
orp
us
Mu
seu
m/C
oll
ecti
on
----
nu
mb
erS
ite
Cla
ssn
um
ber
acce
ssio
nn
um
ber
Pos
itiv
eB
oxed
orP
osit
ionj
Nu
mb
er(P
)S
tan
dar
dsp
ecia
l±
line
sS
hap
ere
adab
leN
egat
ive
foils
bo
xed
per
mm
.(N
)fo
ils
ST
AN
DA
RD
FO
ILS
Key
ston
eG
arne
tD
isc
Bro
oche
sI
Bea
kes
bo
urn
e,G
rav
e22
1.1
IC
ante
rbu
ry75
17K
eyst
on
eI
P4
.25
2F
aver
sham
1.1
7B
M10
50'7
0K
eyst
on
e3
P3·
75.1
3
3S
tow
tin
g1.
113
Mai
dst
on
eA
S30
1K
eyst
on
eI
P4
.25
Cen
tral
N3·
75.1
3
4G
ilto
n,
Gra
ve
411.
215
Mer
seys
ide
M60
58K
eyst
on
e1
P4
.25
5H
ow
lett
s,G
rav
e10
1.2
16B
M19
365
-11
40K
eyst
on
eI
P4
.25
Cen
tral
P4
.25
6M
ersh
am,
Gra
ve
1.2
17C
ante
rbu
ry26
32K
eyst
on
eI
P2
.25
Cen
tral
P2
.25
7S
arre
,G
rav
eC
LV
III
1.2
19M
aid
sto
ne
40
4K
eyst
on
eI
P3·
75.1
38
Sar
re1.
218
BM
93
6-1
233
Key
sto
ne
2P
3·5
.25
9U
np
rov
enan
eed
1.2
20B
MO
A2
98
5K
eyst
on
eI
P3
.25
10U
np
rov
enan
ced
1.2
21C
ante
rbu
ry21
64K
eyst
on
eI
P2.
25.2
5II
Bif
rons
,G
rav
e29
2.1
23M
aid
sto
ne
B29
592
Key
sto
ne
IP
2.2
5C
entr
alN
2.2
512
Do
ver
,G
rav
e14
2.1
25B
M19
631
1-8
78
Key
sto
ne
2P
3.25
.13
13L
ym
ing
e,G
rav
e44
2.1
31
Mai
dst
on
eK
AS
1954
Key
sto
ne
3p
4.2
5P
air
to14
Cen
tral
P3·
75.1
3
14L
ym
ing
e,G
rav
e44
2.1
32M
aid
sto
ne
KA
S19
54K
eyst
on
e3
P4
.25
Pai
rto
13C
entr
alP
4.2
5
15P
ostl
ing
2.1
33M
erse
ysid
eM
735
IK
eyst
on
e2
P3·
75.1
3P
air
to16
16P
ost
lin
g2.
134
Mer
seys
ide
M73
75K
eyst
on
e3
P3·
75.1
3P
air
to15
I7S
arre
,G
rav
e4
2.2
42
Mai
dst
on
e40
2K
eyst
on
e2
p3
.25
18F
aver
sham
2·3
44B
M10
45'7
0K
eyst
on
e2
P2
.2.')
Cen
tral
P1.
67.0
8
19G
ilto
n,
Gra
ve
622·
345
Mer
sey
sid
eM
6062
Key
sto
ne
2N
2.25
.13
20B
ifro
ns2-
447
Mai
dst
on
eA
S82
019
5ac
Key
sto
ne
IN
3.25
.13
Key
sto
ne
2P
3.25
.13
Oi
::0 H o ::r: ;p- ::0 tJ ;p <: M Z >-3 ;p Z tJ tJ ;p- <: H tJ r M H Q ::r:
21F
aver
sham
2-4
49E
M10
43'7
0K
eyst
on
e3
N3·
75.1
322
Fav
ersh
am2·
451
Ash
mo
lean
19
°9.2
°3K
eyst
on
e1
p3·
5.2
5
23F
aver
sham
2·4
52A
shm
ole
an1
9°9
.19
3C
entr
alN
3·75
.13
24U
np
rov
enan
eed
2-4
55D
.o.E
.6
82
42
9K
eyst
on
e2
~4
.25
25D
ov
er,
Gra
ve
382·
556
EM
1963
11
-819
2K
eyst
on
e3
N3·
75.1
326
Ho
wle
tts,
Gra
ve
182·
557
EM
19
36
5-1
162
Key
sto
ne
3p
4.2
5
27W
estb
ere
2.6
60C
ante
rbu
ry64
23K
eyst
on
eI
p4
.25
28F
aver
sham
?2.
659
EM
OA
389
Key
sto
ne
IN
3.25
.13
29G
ilto
n2·
969
Ash
mo
lean
1782
Key
sto
ne
3N
3.1
3D
ou
gla
sC
oll
ecti
on
No
.I
I
30C
hat
ham
Lin
es,
3.1
72A
shm
ole
an17
79.1
2K
eyst
on
e1
p3·
75.1
3"T
umul
usX
II'
Inte
rmed
iate
IP
4.2
5In
term
edia
teI
N4
.25
31
Fav
ersh
am3.
177
EM
1041
'70
Key
sto
ne
3N
3·5
.13
0In
term
edia
teI
N3·
75.2
5id
Cen
tral
N3·
5.1
30
32
Fav
ersh
am3.
179
Ash
mo
lean
1942
.238
Key
sto
ne
1N
2.25
.13
00
33G
ilto
n,
Gra
ve
193.
180
Mer
sey
sid
eM
60
09
Key
sto
ne
IP
4·5
.25
00
Inte
rmed
iate
PI
14·
5.2
5::r:
34K
ing
sto
n,
Gra
ve
161
3.1
81M
crsc
ysi
dc
M63
47K
eyst
on
e2
N4
.25
;I>In
term
edia
teI
N4
.25
>-335
Un
pro
ven
anee
d3.
186
D.o
.E.6
82
01
2K
eyst
on
eI
N3·
75.1
30
Inte
rmed
iate
2N
4.2
5::r:
Cen
tral
N4
.25
l'j
36F
aver
sham
3.2
87E
M10
39'7
0K
eyst
on
e3
N3·
5.1
3tj
Rim
I~
4.2
5
37F
aver
sham
3.2
88E
M10
40'7
0K
eyst
on
e3
N2.
25.1
3o
38F
aver
sham
3.2
91
Ash
mo
lean
1909
.209
Key
sto
ne
Ip
3.67
.08
0
39K
ing
sto
n,
Gra
ve
299
3.2
93M
erse
ysi
de
M6
24
3K
eyst
on
e2
N3.
25.1
3t"
"
40G
ilto
n3·
398
Ash
mo
lean
1782
Key
sto
ne
3N
3·75
.13
tj
Do
ug
las
Co
llec
tio
nN
o.
10"T
I4
1H
ow
lett
s,G
rav
e5
3·5
101
EM
1936
5-1
129
Key
sto
ne
IN
3·5
.25
04
2U
np
rov
enan
eed
3·5
102
EM
OA
2984
Key
sto
ne
4p
2·5
.13
.....
43U
np
rov
enan
eed
3·5
103
D.o
.E.
6824
09K
eyst
on
e1
N3·
5.2
5r-
44S
arr
c4
106
Mai
dst
on
e4
07
Key
sto
ne
1p
4·5
.25
00
45U
np
rov
enan
eed
410
8D
.o.E
.6
82
40
4K
eyst
on
e2
N3·
5.1
3
46
Fav
ersh
am5
109
EM
1042
'70
Key
sto
ne
3N
4.2
5
47F
aver
sham
5III
Ash
mo
lean
1909
.205
Key
sto
ne
4P
3·5
.08
Rim
3p
3·5
.25
48F
aver
sham
511
2A
shm
ole
an19
09.1
97K
eyst
on
e2
N3
.13
49F
aver
sham
6.1
115
BM
1031
'70
Key
sto
ne
2N
3·75
.13
Rim
IN
3·5
.13
Rim
IP
3·5
.13
5°IF
aver
sham
I6.
1
I11
6IA
shm
ole
an19
42.
239
IKey
sto
ne
IN
1.2
5.1
3C
entr
alN
3·5
.13
51F
aver
shar
n6.
1Il
SA
shm
ole
an19
09.2
02K
ey
sto
ne
4N
3·5
.13
Inte
rmed
iate
Ip
3·5
.25
-J
Inte
rmed
iate
Ip
2·5
.25
TA
BL
EI
(con
tinu
ed)
FO
ILS
UR
VE
YG
RO
UP
I
.... CO
Foil
Nu
mb
erof
lines
Set
ting
spo
siti
onp
erm
m.
Err
or
Cat
alo
gu
eC
orpu
sM
useu
mjC
olle
ctio
nn
um
ber
Site
Cla
ssn
um
ber
acce
ssio
nn
um
ber
Pos
itiv
eB
oxed
orP
osit
ionj
Nu
mb
er(P
)S
tan
dar
dsp
ecia
l±
lines
Sh
ape
read
able
Neg
ativ
efo
ilsbo
xed
per
mrn
.(N
)fo
ils
52G
ilto
n6.
111
9A
shm
olea
n17
82K
eyst
one
2P
3.25
.13
Dou
glas
Col
lect
ion
53F
aver
sham
7.1
128
Ash
mol
ean
1909
.I83
Key
ston
eI
P3·
75.1
354
Un
pro
ven
ance
d7.
113
0D
.o.E
.68
24
3I
Key
ston
eI
N3·
5.1
355
Fav
ersh
am7·
213
1A
shm
olea
n19
42.2
37K
eyst
one
IN
4.2
556
Wro
tham
7·3
132
BM
1927
5.12
2K
eyst
one
2P
2·75
.13
57O
zing
ell
U14
1M
erse
ysid
eM
7276
Key
ston
eI
P3·
5.2
558
Sar
re,
Gra
ve
CX
VU
142
Mai
dst
on
e39
9K
eyst
one
IN
3.25
.13
Key
ston
eI
P3·
5.2
5R
imI
P4.
25.1
359
Un
pro
ven
ance
dU
144
D.o
.E.
6824
30K
eyst
one
IN
3·5
.25
Pla
ted
Dis
cB
rooc
hes
60F
aver
sham
I14
7B
M3
53
.20
4A
Few
Foils
N3·
5.2
561
Fav
ersh
am1
148
Ash
mol
ean
1909
.184
On
eFo
ilP
4.2
562
Un
pro
ven
ance
dI
151
Can
terb
ury
2614
All
Foils
N3.
25.1
363
Fav
ersh
am2
155
Ash
mol
ean
1909
.199
Cen
tral
N2·
5.2
5O
ne
Oth
erN
5.2
5Fo
il64
Gil
ton,
"Tum
ulus
XV
'2
153
Ash
mol
ean
1836
,A
llR
ead
able
P3·
25.1
3p.
129,
216
Foils
65W
ing
ham
,G
rav
e2
157
BM
79
5-2
43
4A
Few
Foils
N&
P3·
5.1
366
Fav
ersh
am3
158
BM
84
12
-21
4A
llFo
ilsN
3.25
.13
67F
aver
sham
315
9B
M19
162-
1I
IA
llR
ead
able
N3·
75.1
3Fo
ils68
Un
pro
ven
ance
d4
161
D.o
.E.
6824
10T
wo
Foils
N3·
75.2
569
Dov
er,
Gra
ve
F7
164
BM
19
63
11
-87
58
Cen
tral
N3·
5.2
5M
ain
IN
4.25
.13
Ma
in4·
75.1
3M
ain
IN
4·5
.25
Mai
nI
N4·
75.2
5
~ .... o :::r:: > ~ tj > < t:I:1 z .., > z tj tj > < .... tj r M .... Q :::r::
Com
posi
teB
rooc
hes
~&P
7°F
aver
sham
1.1
17°
Ash
mo
lean
1909
.204
All
Rea
dab
le3.
25.1
3Fo
ils7
1M
on
kto
n,
Gra
ve
31.
217
2A
shm
ole
an19
73.1
401
All
Rea
dab
leN
&P
3.25
.13
Foils
72G
ilto
n,G
rav
e42
217
5M
erse
ysid
eM
6006
Cen
tral
P3.
25
.13
Oth
erR
ead
-N
3·5
.25
able
Foils
73IS
arre
,G
rav
e2
177
BM
60
10
-24
IA
llFo
ilsN
3.25
.13
74K
ing
sto
n,
Gra
ve
205
3·2
179
Mer
seys
ide
M62
26A
llFo
ilsP
2.67
.08
75M
ilto
nN
ort
hF
ield
,4·
218
2A
shm
ole
an33
6A
llFo
ilsN
&P
4·5
.25
Ab
ing
do
n,
Ox
ford
shir
ep.
123
.59
Squa
re-H
eade
dB
rooc
hes
076
IChe
ssel
Do
wn
,B
M6
77
.29
8A
llFo
ilsN
&P
2·5
:;0Is
leof
Wig
ht
.13
0 rfl
77H
ow
lett
s(K
eyst
one
2.1
29B
M19
187-
1I
1H
ead
-All
Foils
P4
.25
rfl
Dis
cB
rooc
hon
bow
)F
oo
t-O
ne
Foil
P3·
75I
.13
::cK
eyst
one
IP
4.2
5>
78L
ym
ing
e,G
rav
e44
Mai
dst
on
eK
AS
1954
All
Foils
P3.
25.1
3..,
Pai
rto
79()
79L
ym
ing
e,G
rav
e44
Mai
dst
on
eK
AS
1954
All
Foils
N&
P3·
75.1
3::c
Pai
rto
78i:!
j80
Bif
rons
,G
rav
e42
Mai
dst
on
e58
7B
42H
ead
IP
2·5
.25
tjP
air
to81
Hea
d2.
25.2
5H
ead
IP
2.25
.25
0F
oo
t2
P2.
25.2
50
Fo
ot
IP
2·5
.25
r-81
IB
ifro
ns,
Gra
ve
42I
II
Mai
dst
on
e57
4B
42
IHea
dI
P2
.25
tjP
air
to80
Hea
d2
P2.
25.2
5"'l
Fo
ot
IP
2·5
.25
0F
oo
t2
P2.
25.2
5....
82B
ifro
nsM
aid
sto
ne
KA
S62
0H
ead
IN
4.2
5r-
Pai
rto
8319
54C
rfl
83B
ifro
nsM
aid
sto
ne
KA
S62
0F
oo
tI
N4
.25
Pai
rto
8219
54C
84G
ilto
nE
M62
7-[
13A
llFo
ilsN
1.88
.13
85H
owle
tts,
Gra
ve
7B
M19
365-
1I
28A
llFo
ilsP
3·75
.13
86H
owle
tts,
Gra
ve
7E
M19
365-
1127
Hea
dI
P4
.25
Fo
ot
IP
4.2
587
Ho
wle
tts
BM
961
1-2
62
Hea
dI
P3·
5.2
588
Ho
wle
tts
BM
961
1-2
6I
Hea
d2
N&
P3.
83.0
8F
oo
t2
N&
P4
.13
89M
ersh
arn
Can
terb
ury
2636
Hea
dI
N3
.25
Fo
ot
IN
2·5
.25
goS
tow
tin
g,
Gra
ve
9C
ante
rbu
ry26
77A
llFo
ilsN
&P
3·75
.13
<.0
g!
Sto
wti
ng
BM
OA
275
Fo
ot
!P
2.2
5
TA
BL
EI
(con
tinu
ed)
FO
ILS
UR
VE
YG
RO
UP
I
J\) o
Foil
Nu
mb
erof
lines
Set
ting
spo
siti
onp
erm
m.
Err
or
Cat
alo
gu
eC
orp
us
Mu
seu
mIC
olle
ctio
nn
um
ber
Sit
eC
lass
nu
mb
erac
cess
ion
nu
mb
erP
osit
ive
Box
edor
Pos
itio
n!N
um
ber
(P)
Sta
nd
ard
spec
ial
±lin
esS
hap
ere
adab
leN
egat
ive
foils
per
mm
.(N
)b
ox
edfo
ils
Tri
angu
lar
Buc
kles
wit
hIn
terl
aced
An
ima
lO
rnam
ent
92
Fav
ersh
amB
MI0
98
-70
Buc
kle
Pla
teN
3·5
.13
To
ng
ue
Pla
teI
P3
.25
To
ng
ue
Pla
te5
P&
N2·
5.1
393
Gil
ton,
Gra
ve
23M
erse
ysid
eM
60I8
To
ng
ue
Pla
teN
3·75
.13
94T
aplo
w,
Buc
ks.,
Gra
ve
BM
83
12
-14
IA
llR
ead
able
P2.
67.0
8Fo
ils95
Wic
kh
amb
reu
xB
M19
054
-18
14T
on
gu
eP
late
P3·
5.1
3
Mis
cell
aneo
usB
uckl
es
96B
arri
ngto
n,C
amb
rid
ge-
Ash
mo
lean
1909
.273
Cen
tral
P3·
5.2
5sh
ire
97B
room
fiel
d,E
ssex
BM
94
12
-16
IA
llFo
ilsP
3.25
.13
98C
run
dal
eB
M93
6-1
204
All
Foils
P2·
75.1
3
99G
ilto
nM
erse
ysid
eM
6652
All
Foils
P2·
75.1
3
Gol
dC
lois
onne
and
Fil
igre
eP
enda
nts
100
Fav
ersh
amB
M11
37'7
0O
ne
Foil
N5
.25
101
Fav
ersh
amB
MII
45
'70
On
eFo
ilN
3·5
.25
102
Fav
ersh
amB
M8
41
2-2
15
All
Foils
N3·
5.1
3I0
3F
aver
sham
BM
84
12
-21
6O
ne
Foil
P3·
5.2
5
Oth
erP
enda
nts
I04
Kem
pst
on
,B
edfo
rdsh
ire
BM
91
5-2
43
Sq
uar
eN
3·5
.25
I05
Kin
gsto
n,G
rav
e14
2M
erse
ysid
eM
6273
Foil
P3.
17.0
810
6S
tan
dla
ke,
Oxf
ords
hire
Ash
mo
lean
1971
.448
Foil
P3.
25.1
3I0
7U
np
rov
enan
ced
BM
358
-3I
All
Foils
P3·
5.2
5
?J ...... o ::r: > ?J tj > <: trJ z >-3 > :z: tj tj > <: ......
tj r trJ .... o ::r:
Oth
erO
bjec
ts10
8B
ifro
ns,
Gra
ve
71IM
aid
sto
ne
615
B7
1IC
entr
alI
IP
2·5
I.0
8P
air
to10
9S
mal
lsa
uce
r-sh
aped
bro
och
109
Bif
rons
,G
rav
e71
IM
aid
sto
ne
615
B7
1IC
entr
alI
IP
3I
.25
Pai
rto
108
Sm
all
sau
cer-
shap
edb
roo
ch11
0I
Sar
re,
Gra
ve
94
II
IMai
dst
on
e40
0C
entr
al
II
P
I3·
75
I.1
3B
utt
on
-ty
pe
bro
och
III
IG
ilto
nB
M6
27
-115
Pin
-hea
dN
4.2
5G
ild
edsi
lver
pin
frag
men
t0
112
Wig
ber
Lo
w,
Der
by
shir
e.B
M73
6-2
96
All
Foils
N4
.25
it!P
air
to11
30
Cro
ss-s
hap
edp
in-h
ead
00
113
IW
igb
erL
ow,
Der
by
shir
e.B
M73
6-2
97
On
eF
oil
N4
.25
00 I
Pai
rto
112
::r:C
ross
-sha
ped
pin
-hea
d>
114
Ixw
ort
h,
Suf
folk
Ash
mo
lean
1909
.453
On
eF
oil
N3
.25
>-3C
hri
stia
nC
ross
011
5W
ilto
n,
Nor
folk
BM
59
5-1
2I
All
Foils
P3·
5.2
5::r:
Ch
rist
ian
Cro
sstt:I
116
Bif
rons
,G
rav
e42
Mai
dst
on
e56
6B
42F
oil
P3.
670.
8t:J
Gil
ded
silv
erri
ng
117
IF
aver
sham
BM
1144
'70
Sq
uar
eN
3·5
.13
0G
old
hin
gei
ng
Bos
sN
3·5
.25
0O
bje
ctr-
II8
IS
tod
mar
shB
M5
41
2.2
6F
oil
N2·
75.0
8t:J
Sil
ver
spo
on
":I
119
Wic
kh
amb
reu
xB
M1
90
54
-18
16T
wo
Foils
P3
.25
0In
laid
go
ldst
ud
...... r- 0
0
II
II
IB
OX
ED
FO
ILS
Keys
tone
Garn
etDi
scB
rooc
hes
120
IH
ow
lett
sI
I.I
I9
IEM
1918
7-8
41
Key
sto
ne
1P
3.25
.13
Key
sto
ne
2P
3.25
.13
Key
ston
e3
.13
121
IS
arre
II.I
I12
IB
M9
36
-123
2IK
eyst
on
e1
P2·
75.1
3K
eyst
one
3·5
.25
Key
sto
ne
1P
3.1
3K
eyst
one
3·5
.25
NK
eyst
on
eI
N2.
88.1
3Ke
ysto
ne3·
5.2
5
TA
BL
EI
(co
nti
nu
ed)
FO
ILS
UR
VE
YG
RO
UP
I
Jo,)
Jo,)
Foi
lN
urn
ber
ofli
nes
Set
ting
sp
osi
tio
np
erm
m.
Err
or
Cat
alo
gu
eC
orp
us
Mu
seu
m!C
oll
ecti
on
nu
mb
erS
ite
Cla
ssn
um
ber
acce
ssio
nn
um
ber
Pos
itiv
eB
oxed
orP
osit
ion!
Nu
mb
er(P
)S
tan
dar
dsp
ecia
l±
line
sS
hap
ere
adab
leN
egat
ive
foils
bo
xed
per
mm
.(N
)fo
ils
122
How
lct.
ts,
Gra
ve
92.
13
°B
M19
365
-11
39K
eyst
on
e2
P3.
25.1
3K
eyst
one
3.25
.25
123
Fav
ersh
am2.
24
°B
M10
48'7
0K
eyst
on
e2
P3
.13
124
Sar
re,
Gra
ve
42.
243
Mai
dst
on
e40
1K
eyst
on
e3
P3·
75.1
3C
entr
alP
3·75
.13
125
Fav
ersh
am2.
863
BM
1033
'70
Key
sto
ne
2P
3.25
.13
126
Fav
ersh
am2.
866
BM
704
-27
99
Key
sto
ne
IP
3·5
.25
127
Un
pro
ven
ance
d2.
868
D.o
.E.
6824
08K
eyst
on
e4
P3.
25.1
312
8F
aver
sham
3.1
75C
ante
rbu
ry26
59K
eyst
on
e2
p3·
5.1
312
9F
aver
sham
3.1
76B
M10
35'7
0K
eyst
on
e2
P3·
5.2
5
13°
Fav
ersh
am3.
178
Ash
mo
lean
19°9
.200
Key
sto
ne
3N
3.25
.13
Inte
rmed
iate
2N
3.25
.13
131
Sto
wti
ng
3.1
83M
aid
sto
ne
AS
189
Key
sto
ne
IP
3.25
.13
132
Un
pro
ven
ance
d3.
294
D.o
.E.
6824
06K
eyst
on
e3
P3.
17.0
8
133
Mai
dst
on
e6.
112
1M
aid
sto
ne
AS
283
Key
sto
ne
2p
2·75
.13
Key
sto
ne
1p
2.83
.08
Key
ston
e2·
75.1
3
134
Un
pro
ven
ance
d6.
112
2D
.o.E
.68
2407
Key
sto
ne
4P
3.25
.13
Rim
2P
3·5
.25
135
Un
pro
ven
ance
d6.
112
4D
.o.E
.6
82
4°3
Key
sto
ne
2P
3.2 5
.13
Pla
ted
Dis
cB
rooc
hes
136
Fav
ersh
am1
145
BM
1030
'70
All
Rea
dab
leP
3.25
.13
Foils
Com
posi
teB
rooc
hes
All
Foils
N13
7S
arre
3.1
178
Ash
mo
lean
1934
,20
23
.13
Squa
re-h
eade
dB
rooc
hes
138
Sar
re,
Gra
ve
159
Mai
dst
on
e40
9A
llFo
ilsP
3.2
5.1
3A
llF
oils
3.1
3
id >-< o ~ > id t:l > -< ttl Z ""l > Z t:l t:l > -< >-< t:l t'" t'1 .... o ~
Tri
angu
lar
Buc
kles
wit
hIn
terl
aced
An
ima
lO
rnam
ent
139
IGil
ton
IIB
M6
27.
110
IT
on
gu
eP
late
I5
NI
I3.
25I
.13
Mis
cell
aneo
usB
uckl
es14
°IF
aver
sham
II
I Mer
seys
ide
2.4.
85.4
IFoi
lI
IP
II
3.25
I.1
3
Gol
dC
lois
onne
and
Fil
igre
eP
enda
nts
141
\F
aver
sham
\\
\A
shm
ole
anIg
og.1
88\
Tri
ang
ula
r2
N&
P
II
3.25
I.1
3P
air
to15
5C
ircu
lar
3N
&P
3·5
.25
o14
2I
Un
pro
ven
anee
dD
.o.E
.68
241
IA
llR
ead
able
P3.
25.1
3~
Foils
0 en en I
Oth
erP
enda
nts
::r:14
3F
aver
sham
Ash
mo
lean
Igog
.I95
Foil
N3.
25.1
3;J>
144
Fav
ersh
amB
M11
39'7
0A
llFo
ilsP
3·5
.13
...,P
air
to14
5Q
145
Fav
ersh
amB
MI1
3gA
'70
All
Foils
P3·
5.1
3::r:
Pai
rto
144
trI
146
Un
pro
ven
ance
dC
ante
rbu
ryR
M21
61A
llFo
ilsP
4.2
5t:::
l
Q 0 r- t:::lS
TA
ND
AR
DA
ND
BO
XE
DFO
ILS
(on
the
sam
eob
ject
)>'J
'j
Key
ston
eG
arne
tD
isc
0 .....B
rooc
hes
r-14
7F
avcr
sham
2.8
64E
M10
34'7
0C
entr
alP
3·75
.13
en
Key
ston
e2
P3.
25.1
314
8F
aver
sham
2.8
65A
shm
olea
nIg
09.1
87K
eyst
one
2N
3·5
.13
Key
ston
eI
N14
9F
aver
sham
3.2
8gA
shm
olea
nIg
og.2
06C
entr
alN
3.25
.13
Rim
6N
3·5
.25
Key
ston
e3
N3.
25.1
315
°IF
aver
sham
I3.2
Igo
IAsh
mo
lean
Igog
.201
IRim
3N
Key
ston
e2
N3.
25.1
315
1F
aver
sham
511
3A
shm
olea
nIg
og.1
86R
im2
P3·
75.1
3K
eyst
one
4P
3·75
.13
Cen
tral
P3·
5.2
515
2I
Fav
ersh
amI
6.1
I11
7IA
shm
ole
anIg
og.I
g6IR
im4
N4
.25
IV coK
eyst
one
3N
~25
.13
TA
BL
EI
(co
nti
nu
ed)
FO
ILS
UR
VE
YG
RO
UP
I
Kl
>-f:>.
Foil
Nu
mb
erof
lines
Set
ting
spo
siti
onp
erm
rn.
Err
or
Cat
alo
gu
eC
orp
us
Mu
seu
m/C
oll
ecti
on
nu
mb
erS
ite
Cla
ssn
um
ber
acce
ssio
nn
um
ber
Pos
itiv
eB
oxed
orP
osit
ion/
Nu
mb
er(P
)S
tan
dar
d±
lines
Sh
ape
read
able
Neg
ativ
efo
ilssp
ecia
lp
erm
m.
(N)
bo
xed
foils
Pla
ted
Dis
cB
rooc
hes
153
Sib
erts
wol
d-B
arfr
esto
n,I
149
Mer
seys
ide
M64
9I
On
eFo
ilP
3.2
5G
rav
e10
1O
ne
Foil
P-
-
Com
posi
teB
rooc
hes
154
Do
ver
,?P
rio
ryH
ill,
217
4B
M7
91
0.1
3I
So
me
Rin
g3
N3
.13
Gra
ve
Oth
erFo
ilsN
3.25
.13
Gol
dC
lois
onne
and
Fil
igre
eP
enda
nts
155
Fav
ersh
amA
shm
ole
an19
09.I
89C
ircu
lar
2N
&P
3·5
.25
Pai
rto
141
Cir
cula
rI
N&
P4
.25
Tri
ang
ula
r3
N&
P3.
25.1
3
Oth
erP
enda
nts
156
Fav
ersh
amB
M11
40'7
0C
entr
alP
3.25
.13
Oth
erFo
ilsP
3.25
.13
Oth
erO
bjec
ts
157
Bro
omfi
eld,
Ess
exB
M94
12
-16
2Fo
ilon
To
pP
3.25
1.3
Go
ldg
arn
et-i
nla
idFo
ilson
Sid
eP
3.0
.08
py
ram
idst
ud
Foi
lson
Side
3·17
.08
ST
AN
DA
RD
AN
DS
PE
CIA
L
BO
xED
FOIL
S(1
6SQ
UA
RES)
(on
the
sam
eob
ject
)
Key
ston
eG
arne
tD
isc
Bro
oche
s15
8D
over
,G
rav
e29
6.1
114
BM
1963
11
-814
1C
entr
alN
3·5
.25
Rim
2N
4.2
5K
eyst
on
e4
N3.
63.1
3
id ..... o ~ > id Ij > <: trI 'Z o-J > 'Z Ij
Ij > < ..... Ij t""~ ..... Q ~
159
IU
np
rov
enan
ced
I6.
1I
123
ID.o
.E.
6824
05R
im2
N4·
5.2
5R
im2
P5
.25
Key
ston
e2
P3·
5.1
3K
eyst
one
IP
3·5
.13
Key
ston
e3.
25.1
3K
eyst
one
IP
3·75
.13
II
IS
PE
CIA
LB
OX
ED
FO
ILS
(16
SQ
UA
RE
S)
Key
ston
eG
arne
tD
isc
Bro
oche
s16
0I
Dov
er,
Gra
ve
I3.
173
BM
1963
11
-82
Key
ston
e3
N3.
63.1
3
Oth
erP
enda
nts
o16
1I
Dov
er,
Gra
ve
29B
M19
63I1~8
144
Foil
N3·
5.0
8:;>;
:I 0 rJ)
rJ)
SP
EC
IAL
BO
XE
DF
OIL
SI
(20
AN
D16
SQ
UA
RE
S)
:r::Sq
uare
-hea
ded
Bro
oche
s~ >-'
l16
2I
Do
ver
BM
1963
11-8
776
Hea
d-A
llFo
ilsP
3.25
.13
o(K
eyst
one
Dis
cB
rooc
h2.
127
Fo
ot-
On
eFo
ilP
3·33
.08
:r::on
bow
)F
oo
t-O
ther
P3.
25.1
3tIl
Foils
tjK
eyst
one
IP
3·5
.25
QO
ther
Obj
ects
II
I Mai
dst
on
eK
AS
313
lon
eFo
il0
163
IBif
rons
,G
rav
e42
P3·
5.2
5t'"
'S
ilve
rS
po
on
tj >rj 0
SP
EC
IAL
BO
XE
DF
OIL
S.....
.(2
0S
QU
AR
ES
)r- rJ
)
Squa
re-H
eade
dB
rooc
hes
164
ISar
re,
Gra
ve
4M
aid
sto
ne
408
Hea
d-O
ne
Foil
P4
.25
Hea
d-O
neF
oil
3·5
.25
Fo
ot-
On
eFo
ilP
4.2
5F
oot-
One
Foi
l3·
5.2
5I
II
IM
ISC
EL
LA
NE
OU
SF
OIL
S
Key
ston
eG
arne
tD
isc
Bro
oche
s*1
6 5IF
aver
sham
II.I
I6
IBM
1049
'70
IK
eyst
one
IP
3·5
.25
Key
ston
eI
p3·
5·'
3t,;
)
Key
ston
eI
p3.
25.1
3U
1C
entr
alP
3.25
.13
TA
BL
EI
(con
tinu
ed)
FO
ILS
UR
VE
YG
RO
UP
I
Foil
Nu
mb
ero
flin
esS
etti
ngs
posi
tion
per
mm
.E
rro
rC
atal
og
ue
Co
rpu
sM
use
um
/Co
llec
tio
nn
um
ber
Sit
eC
lass
nu
mb
erac
cess
ion
nu
mb
erP
osit
ive
Box
edO
rP
osit
ion/
Nu
mb
er(P
)S
tan
dar
dsp
ecia
l.t
lines
Sh
ape
read
able
Neg
ativ
efo
ilsb
ox
edp
erm
m.
(N)
foils
*166
Fav
ersh
am2·
448
BM
1044
'70
Key
ston
eI
P2·
5.2
5*
167
Do
ver
,G
rav
e35
3.1
74E
MIg
631
1-8
lSI
Key
sto
ne
3P
I.1
3
Pla
ted
Dis
cB
rooc
hes
*168
Dov
er,
Gra
ve
126
616
3B
MIg
631
18
58
3O
ne
Foil
N3·
5.2
5
Mis
cell
aneo
usB
uckl
es*
I6g
Fav
ersh
amA
shm
ole
anIg
og.I
34Fo
ilN
1.2
5.1
1F
oil
3·5
.25
*N
OT
ES
ON
TA
BL
EI
165
Ver
yir
reg
ula
rst
and
ard
foils
wit
hth
elin
esv
ary
ing
cons
ider
ably
inth
ickn
ess.
166
Ver
yir
reg
ula
rst
and
ard
foils
wit
hw
ide
groo
ves
inon
ed
irec
tio
nan
dn
arro
wgr
oove
sw
ith
wid
egr
oove
sn
ext
toth
emin
the
oth
er.
(FIG
.I,
e)16
7V
ery
un
usu
alfo
ilsco
nsis
ting
ofla
rge
squa
res
wit
ha
circ
ula
rp
un
ch-m
ark
insi
deea
cho
ne
(FIG
.I,
g).
168
Mo
stof
the
foils
app
ear
tobe
ofth
eb
ox
edty
pe,
alth
ou
gh
they
are
no
tcl
ear
eno
ug
hto
bere
ad,
bu
to
ne
iso
fth
est
and
ard
type
wit
hgr
oove
sse
tse
tat
40°/
135°
,th
usp
rod
uci
ng
alo
zeng
e-sh
aped
pat
tern
wit
hth
ere
adin
ggi
ven
inth
eta
ble
(PL
.II
,c
;FI
G.
I,h
).I6
gT
he
on
lyre
adab
lefo
ilon
this
tria
ng
ula
rbu
ckle
cons
ists
ofla
rge
bo
xed
squa
res,
each
of
wh
ich
isdi
vide
d,on
on
eal
ign
men
t,in
tore
ctan
gles
(FIG
.I,
f).
Ital
ics
inth
eP
osi
tio
n/S
hap
eco
lum
nre
fer
toa
sep
arat
ere
adin
gon
the
sam
efo
ilat
rig
ht
angl
esto
the
on
eab
ove.
Kl
0')
~ >-<
(1 ::r: > ~ tI > -<: t"i Z ..., > Z tI tI > -<: >-
< tI t""
t"i
>-< Q ::r:
CROSS-HATCHED GOLD FOILS
should serve to dispel any possible doubts about the contemporaneity of the twoparts of such a combination.
We must now turn to a review of the various associations of foil-types onindividual objects (TABLE II). The association of standard and boxed foils onkeystone garnet disc brooches occurs only on examples from the Favershamcemetery, with the exception of one from Dover (Catalogue no. 158) and oneunprovenanced (Catalogue no. 159), both of Class 6. I and both bearing the specialboxed foil. The Faversham series comprises six associations between boxed andstandard foils, and three between standard and standard foils. The latter allinvolve foils of3'5l./mm. spacings, but otherwise there appears to be no regularityabout the choice of standard foils in these associations.
Faversham material is also predominant amongst the other objects showingassociations of foil types but there are also examples from Dover, Howlctts, Sarreand on a gold pyramid stud from Broomfield, Essex. This last item has a combination of boxed and standard foils, the boxed foil showing different readings in eachdirection, and an average spacing of 3.08 l.jmm., indicating that it may in factbe an unusual boxed foil. Amongst these other objects four have boxed-standardand four standard-standard combinations. Again, no one particular type ofstandard foil seems to predominate in these combinations.
It is noticeable that there are no apparent combinations of boxed foils withdifferent measurements, which helps to confirm our suspicion that there may havebeen only a very few stamps producing this type of foil.
The use of different foils on single objects was not random. On objects withmore than one standard foil, with the exception of one- a Class 6. I keystonegarnet disc brooch from Faversham (Catalogue no. 51) - the difference isrelated directly to the different types of setting on those objects. For instance, thefoil in the keystone settings may be different from that in the rim settings (e.g.Catalogue nos. 49, 58), or a central foil may differ from that used in all the othersettings (e.g. Catalogue nos. 50, 63). There is the same careful selection for effecton all but two of the thirteen objects sharing both standard and boxed (or specialboxed) foils.
Orientation of cellfoils
In some cases the patterns on the foils are aligned with the settings intowhich they were fitted; on a rectangular setting, for instance, the lines may beparallel with the sides of the rectangle (e.g. square-headed brooches of SurveyGroup 2, TABLE III, C, I, xi) ..More often the orientation bears a close relationshipto the position of the setting on the object: on a keystone setting, for instance, thelines may be parallel with the radius of the disc (e.g. Catalogue nos. 16, 20) orthey may be at approximately 45° to the radius (e.g. Catalogue nos. 130, 158);on composite brooches, with their all-over cellwork, the orientations may bearonly a rough relationship to the position of the setting, but the orientations aredifferent and so help distinguish one cell from the next.
RICHARD AVENT AND DAVID LEIGH
The jewellers, as one might expect from the overall design of their products,had a very strong feeling for symmetry and carried this into the minutest detailsof their work, including it would seem, the need for care in the placing of their foil.This is nowhere better seen than on some of the square-headed brooches, such asthat from Howletts (TABLE I, Catalogue no. 77) or the closely similar brooch fromDover (TABLE I, Catalogue no. 162). On the head-plate of the Howletts broochthe two top corner foil-patterns point towards the bow while the patterns on thetwo triangular settings are angled at about 45 0 to the axis of the brooch. Thesedirections may be compared with those on the Dover headplate where all the foilpieces are placed so that they are parallel (and at right angles) to the axis of thebrooch. On both brooches, the symmetry is less successful on the footplates (perhaps due to difficulties imposed by the large sizes offoil needed) but the intentionis still clear.
In contrast to these examples, there are some objects, though these are in theminority, where there is no obvious intention in the placing of the foil pieces;their orientations bear no evident relationship to setting shapes or positions. Thismay reflect either slipshod workmanship or perhaps the using up of foil offcuts.
Unusual foils
Among the objects studied in Survey Group I, there are five with miscellaneous and very distinctive foils, which cannot be compared with any of theregular foils. Two keystone garnet disc brooches from Faversham (Cataloguenos. 165, 166) have foils which are irregular variations of the standard type(FIG. r , e) while the only readable piece of foil on a triangular buckle fromFaversham (Catalogue no. r 69) consists of a series oflarge boxed squares, each ofwhich is divided, on one alignment, into a rectangle (FIG. I, f). The plated discbrooch from Dover (Catalogue no. r68) has a unique foil pattern in which thenormal squares are transformed into lozenge shapes (PL. II, c; FIG. I, h). 24 A mostremarkable foil, found on a Class 3. I keystone garnet disc brooch from Dover(Catalogue no. 167), has a pattern consisting of a series oflarge squares (at aboutI l./mm.) each of which has within it a ring-and-dot punch-mark (PL. II, B;
FIG. I, g). These circular punch-marks are not always evenly placed with respectto the chequered pattern, and it appears that two separate processes were involvedin making the stamp. First, the ring-and-dot design was rather irregularlypunched into the face of the stamp, and then the cross-hatched pattern was builtup around these punch-marks. This theory is supported by a close examination ofPL. II, B where, in places, the straight lines can be seen to override the edges of thecircular punch-marks, even though some effort has clearly been made to fit thoselines around the punch-marks. We know of three other occurrences of this foil
24 A boxed foil of similar pattern has been observed on an unprovenanced Class 2.7 keystone garnet discbrooch in the University Museum of Archaeology and Ethnology in Cambridge. Avent, op. cit. in note 9,18, plate II, no. 62.
CROSS-HATCHED GOLD FOILS
pattern in this country,25 and Arrhenius refers to various ring patterned foils frommainly Vandal and Gothic sources, with a few Frankish appearances.w We cannotdetermine without close comparison whether these are similar foils.
While we have not made any systematic study of the foils on Continentalmaterial, we have noticed, from cursory inspection of some of the few sufficientlydetailed photographs published, that there is considerable use of boxed foils,and more often than not of a special, sixteen-squared variety. Whether theseoccurrences are all of the same special boxed type as our Kentish examples wecannot of course know without more detailed work on the other side of theChannel.
Units of measurement
Over the range of foil spacings measured by us there is a marked clustering of3.25 to 4 l./mm., with a tailing off on either side (FIG. 3). The question whichmust be asked is why there is not a more even spread of line spacings over asomewhat wider range.
Clearly there were some parameters: given the techniques and skill availableit was impossible to go beyond a certain degree of fineness; on the other hand, if apattern was too coarse, insufficient of it would be seen within the boundaries ofa small setting for the cross-hatched effect to be apparent. Within these limits,however (say 1.5 to 5 l.jrnm.}, why is there still such a concentration around acentral value? One answer may be that some unit of linear measure was beingemployed and, if we accept that, then we may be further enticed into trying torelate our measurements to some known units of the period. Here there is adifficulty for, as Grierson has pointed out,27 although we have some documentaryinformation on the inter-relationships of different early measures, we have virtually no way of knowing the accurate absolute values of these units until theI5th century, from which time some linear standards survive. This need notmatter, however, since it is doubtful if fixed accurate standards existed at this time.It is likely that more variable units were employed which were related to 'natural'measures, such as palms, hands, thumbs, etc. It seems possible that one of thecurrent units employed was the thumb of about 1.I (modern) inches, or about28 mm. 28 If so, then it is a simple matter to calculate the number of lines per'Saxon' inch on the foils. For instance, our 3.5 l./mm. works out at 98 lines to the'Saxon' inch; our 3.75 at ID5 lines to the 'Saxon' inch. Hour value for this inehcould be relied upon (which it cannot) then our craftsmen might just have beenaiming for about 3.57 l.jmm., or IDO lines to the inch. This may be too fancifulfor some tastes, but it is a beguiling proposition, and if we could find some way ofverifying it, we might, in turn, have the first evidence for the true values of the'Saxon' inch during this period.
25 A. Warhurst, 'The Jutish cemetery at Lyminge', Archaeologia Cantiana, LXIX (1955), 16, 25, 34 f., pl. X,
nos. 1,2; E. T. Leeds and D. B. Harden, The Anglo-Saxon Cemetery at Abingdon (Oxford), 55, fig. 8.26 Arrhenius, op. cit. in note I, 117-118.27 P. Grierson, English Linear Measures (the Stenion Lecture, University ofReading) (Reading, 197 I), 3 f.as F. G. 'Skinner, l11eights and Measures (London, 1967),91.
TA
BL
EII
SU
RV
EY
GR
OU
PI:
OC
CU
RR
EN
CE
SO
FT
WO
FO
ILS
ON
ON
EO
BJE
CT
ST
AN
DA
RD
FO
ILS
TA
ND
AR
DF
OIL
BO
XE
DF
OIL
Cat
alo
gu
eC
lass
Sit
eN
o.S
pac
ing
Sp
acin
gS
pac
ing
Set
ting
s(L
jmm
.)E
rro
rS
etti
ngs
(l.j
mm
.)E
rro
rS
etti
ngs
(l./
mm
.)E
rro
r
Key
ston
eG
arne
tD
isc
Bro
oche
s14
72.
8F
aver
sham
Cen
tral
3·75
±0.
132
Key
sto
ne
3.25
±0.
1314
82.
8F
avcr
sham
2K
eyst
on
e3·
5±
0.13
IK
eyst
on
e-
149
3.2
Fav
ersh
amC
entr
al,
6R
im3.
25±
0.13
3K
eyst
on
e3.
25±
0.13
150
3.2
Fav
ersh
am3
Rim
-2
Key
sto
ne
3.25
±0.
1315
15
Fav
ersh
am2
Rim
3.75
±0.
13C
entr
al3·
5±
0.25
4K
eyst
on
e49
6.1
Fav
ersh
am*
2K
eyst
on
e3.
75±
0.13
2R
im3.
5±
0.13
506.
1F
aver
sham
IK
eyst
on
e1.
25±
0.13
Cen
tral
3.5
±0.
1351
6.1
Fav
ersh
amI
Inte
rmed
iate
2·5
±0.
254
Key
ston
e,3.
5±
0.13
IIn
term
edia
te15
26.
1F
aver
sham
4R
im4.
0±
0.25
3K
eyst
on
e3.
25±
0.13
158
Spe
cial
Box
ed6.
1D
over
,G
rav
e29
Cen
tral
3·5
±0.
252
Rim
3.63
±0.
134
Key
sto
ne
159
6.1
Un
pro
ven
ance
d4
Rim
4.71
±0.
252
Key
sto
ne
3.5
±0.
1358
US
arre
,G
rav
eC
XV
2K
eyst
on
e3.
5±
0.25
IR
im4-
25±
0.13
Pla
ted
Dis
cB
rooc
hes
153
IS
iber
tsw
old/
Bar
fres
ton,
On
efo
il3.
0±
0.25
On
efo
il-
Gra
ve
101
632
Fav
ersh
amC
entr
al2·
5±
0.25
On
eo
ther
5.0
±0.
2569
7D
over
,G
rav
eF
Cen
tral
3·5
±0.
25O
ther
mai
nfo
ils4·
5±
0.13
Com
posi
teB
rooc
hes
154
2D
over
,?P
rior
yH
ill
All
bu
tso
me
3.25
±0.
13S
om
eri
ng
33.
0±
0.13
Gra
ve
rin
g3
foils
Squa
re-h
eade
dB
rooc
hes
77H
owle
tts*
Mai
nb
roo
ch3·
75±
0.13
IK
eyst
on
e4.
0±
0.25
sett
ings
ondi
sc
Buc
kles
92
Fav
ersh
am*
Buc
kle
pla
te2.
5±
0.13
To
ng
ue
pla
te2.
75±
0.13
oo o id ......
(1 II: ;J; id t:l ;J; < M Z .., ;J; Z t:l t:l ;J;- < ......
t:l t"'
M ......
Q II:
155
156
Gol
dC
lois
onne
and
Fil
igre
eP
enda
nts
Fav
ersh
am(P
air
to14
1)
Oth
erP
enda
nts
Fav
ersh
am
IC
ircu
lar
Oth
erfo
ils
4.0
±0.
25
3.25
1:0.
13
2C
ircu
lar
3T
rian
gu
lar
Cen
tral
3.25
±0.
13
3.25
±0.
13
*O
nth
ese
obje
cts
the
spac
ing
sar
esu
ffic
ient
lycl
ose
and
the
erro
rssu
ffic
ient
lyla
rge
tom
ake
itpo
ssib
lefo
rth
ere
tobe
on
lyo
ne
stan
dar
dfo
il,ra
ther
than
two.
Th
isis
par
ticu
larl
yso
inth
eca
seof
No
.77
,w
ith
itsla
rge
erro
rfo
rth
ere
adin
gs
onth
edi
sco
fth
esq
uar
e-h
ead
edb
roo
ch.
Oth
erO
bjec
ts15
7I
IB
roo
mfi
eld
(Go
ldp
yra
mid
stu
d)
NO
TE
ON
TA
BL
EII
To
pfo
il3.
25±
0.13
Sid
efo
ils3.
08±
0.0
8o :;0 o lf
lo: I ~ > >-3 () ~ M tj Q o r tj >rJ o ...... r- o: vo
RICHARD AVENT AND DAVID LEIGH
SURVEY GROUP 2
The square-headed brooches of Kent form a small group of objects, nearly allof which probably were produced in the first three-quarters of the 6th century.A number of them carry garnet-on-foil decoration, and twenty of these have beenincluded in the general Survey Group I presented above. We have carried out amore detailed and intensive study of thirty-one of these inlaid brooches by theuse of high-quality photographs on which the foils are more visible than theytend to be on most published photographs.
Photographs of the brooches at X 2 magnification were viewed through astereomicroscope'" while an accurately made transparent ruler, graduated inmillimetres, was placed on the surface of the photograph so that the images ofthe foils could be seen beneath the graduations. The zero of the ruler was madecoincident with a foil line and a count oflines taken to the maximum convenientnumber ofmillimetres, an estimate of ' tenths ofa line' being made by eye up to thefinal millimetre mark (the first line being counted as I). Sometimes it proved moreconvenient to take a whole number of lines and to measure the distance betweenthem to (estimated) tenths of a millimetre.
Having made a count along one line, another line a few spaces away from itwas chosen and a count made along that, and then another line, and so on takingas many readings in one direction on a piece of foil as seemed to be representativeof the overall spacing; on smaller pieces of foil fewer readings could be taken. Thephotograph was then turned through a right angle and the process repeated inthe other direction. The same sequence was then followed for each piece of foilclearly visible.
To calculate the spacings from these figures was simply a matter of dividingthe number of lines by the number of millimetres and multiplying by 2 to allowfor the magnification of the photographs. 30 Using all the readings for a singlebrooch, a mean spacing (x) could be calculated, together with its standarddeviation(s) and, for use on the graphs, a variance (S2).31 Thus all the readings oneach cell foil on a brooch could be combined to give a single figure representingthe line spacing of the sheet foil used on that brooch. (For the moment no accountis taken of the directions of readings.) For this purpose it was necessary to assume,as we did in Survey Group I, that all similar cell foils on an individual broochderived from the same sheet of foil. This seems a reasonable assumption and isborne out by the readings themselves, which show much smaller variationsbetween cell foils than those found between foils on one brooch and another. Inpreparing the graphs (FIGS. 8-10) the same assumptions were made for the cellfoils on a pair of objects, unless there was a marked difference between all thereadings on one item of the pair and the other.
29 A Wild M7A Stereomicroscope was used merely as a convenient magnifier, and its exact magnificationwas immaterial to the ensuing calculations.
30 In order to allow for slight imprecision in printing of the photographs a correction was applied: thelength of the image of a brooch was compared with its true length, as measured (usually with calipers) inthe museum. The mean spacing (x) could then be multiplied by the appropriate correction factor.
31 S2, the variance, is the mean-squared deviation from the mean; s, the standard deviation, is the squareroot of the variance.
20
VA
RIA
NC
E 19
(s"
10
')
18 17 16
,C
lJ
15 l'
]3 12 11 10
FIG
.8
FO
ILS
UR
VE
YG
RO
UP
2:
SQ
UA
RE
-HE
AD
ED
BR
OO
CH
ES
Mea
nli
ne
spac
ing
(x)
plo
tted
agai
nst
var
ian
ce(s
2X
10
2),
sb
ein
gth
est
and
ard
dev
iati
on
for
all
read
ing
son
ab
roo
ch(o
rp
air)
,ir
resp
ecti
ve
ofd
irec
tio
no
fre
adin
g.
Pai
rsof
bro
och
esar
ep
lott
edo
nce
,u
sin
gav
erag
edfi
gure
s
eo OS
(II)
stan
dar
d•
}1
5+
rea
din
gs
bo
xed
•
2'5
oFG
)
eA(iX
iI)
;;}6-
14re
ad
ing
s
3·0
oS
(i)
eA
(iii)
OQ
•L'*
' eq,
)
~}
1-5
rea
din
gs
OH
3'5
.Mt
OR
1 eg.
[]N
t
OP
,."
oT
(II" e
CII
I)
e8 eG
eI
eKm O
E
SP
AC
ING
xlin
es]
mm
4'5
o 7:i o w W I ::r: > >-3 o ::r: ~ tj Q o r tj >T
j o H r- o: uo uo
TA
BL
EII
IS
UR
VE
YG
RO
UP
2-
SQ
UA
RE
-HE
AD
ED
BR
OO
CH
ES
gj,
co~Q
Q::1
<::0
g;.a
""~~
'"O
J:)
v.~
----:--
'V~
Mu
seu
m<::
~I'lO
0'o
c...
:::l
P<
,,~
Bro
och
aG
rou
pI
Sit
eA
cces
sion
'"0
-0"0
P<~
u"'O
---:::
-n
um
ber
nu
mb
er[fl.
l-<l-<~
<::CI~
vQ
S..0
'"0
~'"
IS
-o~
Ref
eren
ces
toIl
lust
rati
on
st'
'"S
Q~
~.g,
,,~
"£~
6i'~
~~
z
Ai
76C
hess
elD
ow
n,
BM
67
7-2
98
2.83
250
.'7
0.15
Ab
erg
,fi
g.12
6;de
Bay
e,pl
.vi
i,no
.6
;K
end
rick
,l.
O.W
.pl
.x
xx
;H
illi
er,
No
.33
and
p.29
.A
IIC
hess
elD
ow
n,
BM
67
7-2
99
2.81
150
.27
0.28
V.C
.H.,
38
8;
Med
.A
rch.
,I,
19,
fig.
3(b
).L
O.W
.A
III
Che
ssel
Do
wn
,B
M6
77
29
73.
1925
0.3
10.
16L
OW
.-----
---",----------._
---._
------~--------
-----------
Ai
and
ii2.
824
00.
210
.19
---~----
E77
Ho
wle
tts
EM
19
18
7-1
1I
4·lo
c17
0.23
0.1
4Je
ssu
p,
pl.
xii,
fig.
I;E
MG
uid
e,fi
g.5
7;
P.S
.A.,
xx
x,
pI.
2,fi
g.9;
Ab
erg
,fi
g.14
5;M
ed.
Arc
h.,
I,
pl.
i,D
;A
ven
t,N
o.29
.C
i78
Ly
min
ge,
Mai
dst
on
eK
AS
3.4
035
0.16
0.16
Arc
h.C
ant.,
LXIX
,pl
.xi
i,IA
;M
ed.
Arc
h.,
I,
Gra
ve
4419
54pl
.iv
,A
.C
II79
Ly
min
gc,
Mai
dst
on
eK
AS
4.17
32
0·33
0.22
Arc
h.C
ani.,
LXIX
,pI
.xi
i,IE
.G
rav
e44
1954
Ab
erg
,fi
g.13
5;B
row
n,
III,
pl.
xxxi
v,2.
Di
80B
ifro
ns,
Mai
dst
on
e58
72·
43II
0.23
0-43
Gra
ve
42B
42
DII
81B
ifro
ns,
Mai
dst
on
e57
42.
269
0.0
90.
05G
rav
e42
B4
2
D2.
36
200.
210.
36
Ei
82B
ifro
nsM
aid
sto
nc
KA
S4.
52
2(0
·39)
0·57
Ab
erg
,fi
g.13
4.6
20
1954
CE
ii83
Bif
rons
Mai
dst
on
eK
AS
4.3
13
(0.0
8)0
.07
62
01
95
4C
E4
·39
50.
230.
31
Fi
84G
ilto
nB
M6
27
-113
2·5
46
0.2
9-
Ab
erg
,fi
g.13
6;B
oys,
86
8-9
,pI
.,fi
g.9.
FII
Gil
ton
BM
62
7-1
12_
Ca
--
Gi
85H
ow
lett
s,B
M1
93
65
-1I
284.
1416
0.23
0.25
B.J
vI.Q
.,X
I,pl
.xv
,12
.G
rav
e7
(,);
)
0+-
~ H o ::r: :> ~ Ij :> <: ttl Z >-i :> Z Ij
Ij :> < H Ij r ttl ......
Q ::r:
....
Gii
86H
ow
lett
s,B
M1
93
65
-11
274.
074
0.02
0.0
4G
rav
e7
-~~-~~-
G4.
1320
0.21
0.20
H87
Ho
wle
tts
BM
96
11
-26
23.
52
40
.04
0.05
I88
Ho
wle
tts
BM
96
11
-26
I4
·39
220.
260
.09
J89
Mer
sham
Can
terb
ury
2636
2.86
120.
40
0.3
1A
ber
g,
fig.
137;
Sal
in,
fig.
140;
Coi
l.A
ni.,
VI
pl.
xxxv
ii,
NO
.5;
Bak
ka,
fig.
52
(a).
Kiet
90S
tow
tin
g,
Can
terb
ury
2677
4.3
827
0.23
0.21
Arc
haeo
logi
a,X
LI
(186
7),
pl.
xix,
I.
Gra
ve
9A
ber
g,
fig.
144;
Arc
haeo
logi
a,X
LI,
pI.
xix,
No
.I
Kii
d9
1S
tow
tin
gB
MO
A2
75
_0
0-
-M
ed.
Arc
h.,
I,19
,fi
g.3
(a).
L*
138
Sar
re,
Gra
ve
159
Mai
dst
on
e4
09
3·34
520
.19
0.25
Lee
ds
(194
9),
B4,
84
;A
ber
g,
fig.
120;
Arc
h.C
ant.
VI,
pl.
6:1
;B
row
n,
IV,
pl.
155.
5.M
t16
2D
ov
erE
M19
631
1-8
77
63·
6538
0.2
00.
18A
ven
t,N
o.
27.
Nt
164
Sar
re,
Gra
ve
4M
aid
sto
ne
40
83.
891
00.
30
0.4
0B
akk
a,fi
g.5
4;
(Ab
erg
,fi
g.18
4;L
eed
s(1
913)
fig.
21).
aC
hcss
elD
ow
n,
BM
67
7-2
910
3.9
615
0.7
0-
Ab
erg
,fi
g.14
3.L
O.W
.P
Che
ssel
Do
wn
,E
M6
77
-29
II
3.9
24
0.2
80
·34
Ab
erg
,fi
g.12
9;M
ed.
Arc
h.,
I,19
,fi
g.4
(a).
LO
W.
Qi
Che
ssel
Do
wn
,E
M6
77
-29
13
3.18
3(0
.14)
0.26
LO
.W.
Qii
Che
sscl
Do
wn
,B
M6
77
-29
23
3·60
2(0
.06)
-
LO
.W.
---~~
Q3·
355
0.2
5-
-_
..-
---_
.
RF
ing
lesh
am,
Pri
vat
eC
oll
ecti
on
3.7
65
0.1
00
.10
Lee
ds
(194
9),
S.2
;M
ed.
Arc
h.,
II,
19,
fig.
II
(a)
Gra
ve
E2
pl.
iv,
A.
Si
Fin
gle
sham
,P
riv
ate
Co
llec
tio
n3.
184
0·47
0·7
4L
eed
s(19~16),
pI.
xvi
(I);
Med
.A
rch.
,I,
19G
rav
eE
2fi
g.4
(c);
Med
.A
rch.
,II
,19
,fi
g.II
(b).
S11
Fin
gle
sham
,P
riv
ate
Co
llec
tio
n2.
38
20.
18-
Gra
ve
E2
T·t
Fav
ersh
amB
M10
54'7
04.
084
0·42
-A
ber
g,
fig.
97
;M
ed.
Arc
h.,
1,1
9,
fig.
3(c
).1+
Tii
Fav
ersh
amE
M10
54'7
0_
00
--
o 7J o en en I ::r: ;J;. .., o ::r: t"J ti o o r ti >
'!j o ..., r-< en
NO
TE
SO
NT
AB
LE
III
V:l
Ul
Irre
gu
larl
yb
ox
edfo
il.*e fa
Pai
rsar
ed
eno
ted
by
asi
ngle
lett
eran
dd
iffe
ren
tiat
edb
ysm
all
Ro
man
nu
mer
als.
bW
her
ea
dra
win
gis
cite
dfo
ra
pai
rof
bro
och
esit
iso
ften
no
tcl
ear
fro
mth
ed
raw
ing
orits
cap
tio
nw
hic
ho
ne
ofth
ep
air
isre
pre
sen
ted
.S
om
etim
esth
ed
raw
ing
con
tain
sel
emen
tsof
bo
thb
roo
ches
.c
Th
eab
sen
ceof
are
adin
gin
dic
ates
that
nofo
ilco
uld
bere
ad.
dT
her
eis
no
evid
ence
that
thes
etw
ob
roo
ches
wer
efo
un
dto
get
her
,b
ut
they
qu
ite
evid
entl
yfo
rma
pai
r.M
easu
rem
ents
wer
en
ot
mad
eon
the
foils
onth
edi
scs
ofth
ese
bro
och
es.
Th
est
and
ard
dev
iati
on
(s)
use
din
this
tab
leis
that
asso
ciat
edw
ith
read
ing
sin
all
dir
ecti
on
so
na
bro
och
,ta
kin
gn
oac
cou
nt
ofdi
visi
onin
toh
igh
and
low
read
ing
s(c
it.no
tes
31,
32).
Bo
xed
foils
.t
Sp
ecia
lb
ox
edfo
ils.
AbergArch. Cant.AventBakka
Brown
BM Guide
B.M.Q.Boysde BayeHillierMed. Arch.JessupKendrickLeeds (1913)Leeds (1936)Leeds (1949)P.S.A.SalinV.C.H.
RICHARD AVENT AND DAVID LEIGH
ABBREVIATED REFERENCES USED IN TABLE III
N. Aberg, The Anglo-Saxons in England (Uppsala, 1926).Archaeologia Cantiana.R. Avent, op. cit. in note 9.E. Bakka, 'On the beginnings of Salin's Style I in England', Universitetet i Bergen ArbokI958, Historisk-antikuarisk rekke Nr. 3 (1958).G. B. Brown, The Arts in Early England, III and IV, Saxon Art and Industry in the Pagan Period(London, 1915).(R. A. Smith) A Guide to the Anglo-Saxon Antiquities in the Department of British and MedievalAntiquities (British Museum, 1923).British Museum Quarterly.W. Boys, Collectionsfor a History of Sandwich, Pt. II (Canterbury, 1792).J. de Baye, The Industrial Arts of the Anglo-Saxons (London, 1893).G. Hillier, History and Antiquities of the Isle of Wight (London, 1855).Medieval Archaeology.R. F. Jessup, The Archaeologyof Kent (London, 1930).T. D. Kendrick, Anglo-Saxon Art (London, 1938).E. T. Leeds, The Archaeologyof the Anglo-Saxon Settlements (Oxford, 1913, reprint 1970).E. T. Leeds, Early Anglo-Saxon Art and Archaeology (Oxford, 1936, reprint 1968).E. T. Leeds, A Corpus of Early Anglo-Saxon Great Square-Headed Brooches (Oxford, 1949).Proceedings of the Society of Antiquaries if London.B. Salin, Die altgermanische Thierornamentik (1904, 2nd impression, Stockholm, 1935).Victoria County History, Vol. I of the relevant county (Kent or Hampshire).
Because of the technique used and because of the far more numerous measurements taken on each piece of foil, and on each brooch, the major source of variation in measurements is likely to be in the variations on the foils themselves, andought to be considerably greater than that due to experimental error. Thus, incontrast to the possibilities provided by Group I, we are able to use the statisticalconcept of variance to help us judge the degree of irregularity of spacings on anyone foil. The measurements are presented in TABLE III and plotted in FIG. 8. On thelatter the overall mean spacing for all the foils on a brooch (or pair) is plottedagainst the variance, or irregularity, ofthat foil pattern. The greater the variance, themore irregular is thepattern, and vice versa. By plotting the information in this way wehave a means of distinguishing and perhaps grouping foils both by their spacings,in lines per millimetre, and by their degree of regularity. Thus two foils with aclosely similar spacing may be distinguished by very different variances (i.e. oneis very regular, the other very irregular). Clearly this can be very helpful. If,however, both the spacings and the variances are close, although there is then astrong case for calling them the same foil, we may now invoke the evidenceprovided by visual inspection of a photograph, or better still, the brooch itself.The eye can often see features which characterize foils but which are not readilyamenable to measurement. We have been able to take not only good photographsat X 2 magnification, but also a series of macrophotographs, taken at highermagnification of details of the foils.
However, other information can be obtained from the figures to refine theinvestigation further. It seems likely that when a stamp was cut the spacing inone direction did not always exactly match that in the direction at right angles toit; the pattern formed was of rectangles rather than true squares. Fresh calculations were therefore made, based on the readings for each cell foil, in order toderive a mean spacing in each direction on that piece of foil. It was then possible
CROSS-HATCHED GOLD FOILS 37to label the 'high-reading' direction and the 'low reading' direction. Where themeans were identical they were arbitrarily labelled. All the original figures werethen combined for a single brooch (or pair) to give mean high (Xh) and mean low(Xl) readings, again with standard deviations and variances (S2).32 To some extentthis is an arbitrary operation and, where the spacings are very similar in bothdirections (i.e. we have nearly true squares), there may be little or no significancein the measured differences. Where the differences between the readings are morepronounced, however, they may be quite significant and so may help tocharacterize different sheet foils. We may thus do further graphical plots of meanoverall spacings against mean high and mean low readings (Xh and Xl) and thiswe have done, though not published here since they add little to FIG. 8. However,they have been consulted in arriving at a final grouping. It is more useful to plotthe spacings against the difference between the high and low readings (Xh-Xl)(FIG. 9), since this gives us another method of expressing the irregularity of a foilWhere this factor is low, there is little difference between the spacings in eachdirection and we have an even foil composed of more or less true squares. Wherethis factor is high, then the spacing in one direction is much higher than that inanother, and the foil will look uneven, the spaces between the lines being nearerrectangles rather than true squares. A further graph can be drawn, which doesnot materially add to the information already plotted, but which gives us a usefulvisual indication of the overall regularity of the foil (FIG. 10). This graph is a plotofvariance (S2) against (Xh-Xl). The nearer a point lies to the origin of this graph,the better the foil it represents.
In plotting the graphs, different symbols have been used to indicate the degreeof reliability to be placed on the relevant measurement. A spacing based on morethan fifteen readings on a single object is clearly much more reliable than onebased on two or three readings, such as we were forced to depend on when some ofthe foils were missing or obscured, and this has to be taken into account whendrawing conclusions from the graphs.
In deciding whether apparently individual foils with fairly similar spacingsfrom two different items may in fact derive from a single large sheet foil, we mayuse a statistical test which says that we may so consider them provided theirvariances do not differ by a factor of more than about 3.33 Another more refinedstatistical test, of the difference between two means.s! proved impossible to use
3' S', This variance differs from that already used (s"), being the combined standard deviation for highand low readings, weighted to allow for the relevant number of readings in each direction, ni, and nj:
~ .Sh'+ ~-,Sl'ni, + ni nh + ni
About 450 pairs of readings were made (lines and millimetres) and the considerable body of calculations,including photographic correction factors, standard deviations and variances, demanded the help of aprogrammable calculator (Hewlett-Packard 125).
33 The Variance Ratio test (F-test), e.g. M. J. Moroney, Facts from Figures (London, '951), 234 ff;J. E. Doran and F. R. Hodson, Mathematics and Computers in Archaeology (Edinburgh, 1975),66. The varianceratio of 3 is an acceptable approximation for our usual sample sizes, though it might rise to 5 where thesample becomes very small.
34 The t-test for the difference between two means, e.g. Moroney, ibid., 231fT.; Doran and Hodson,ibid., 54.
c..>:l CO
Xh-X
lD
IFF
ER
EN
CE
inS
PA
CIN
G.7
0
oS
(i)
·60
()J
OO
(i)
eA
liii)
eC
li)
eD
id .... 0 :r: > id t:I > -< MO
PZ
OE
~ > Ze
C(i
i)eK
t:IeG
t:I >e6
-< ....O
Re1
t:I r- tIl .....-
";)
-Q
'·0
SP
AC
ING
x:r:
lines
Imm
oH
3·5
o1
-5
rea
din
gs
3·0
()
6-1
4re
ad
ing
s
2·5
eA
(rXii)
•15
+re
ad
ing
s
oI 2·0
·'0 •10·'0 ·20
.30
FIG
.9
FO
ILS
UR
VE
YG
RO
UP
2:
SQ
UA
RE
-HE
AD
ED
BR
OO
CH
ES
Mea
nlin
esp
acin
g(x
)plo
tted
agai
nst
diff
eren
ces
betw
een
mea
nhi
ghan
dm
ean
low
spac
ings
(Xh
-X
l).P
airs
ofbr
ooch
esar
epl
otte
don
ce,
usin
gav
erag
edfi
gure
s
CROSS-HATCHED GOLD FOILS
VARIANCE 17
(5'xlO') 16
15
~J14
·13
12
11
10
eAD;i)
eC(ii)
OP
e6 eK(i)el OEeG
eA(i)(Ii)
eC(i) tiD
OR 05(i)
39
e 15+ readings () 6-14 readings
FIG. 10
o 1-5 readings
FOIL SURVEY GROUP 2: SQUARE-HEADED BROOCHESDifference between mean high and mean low spacings (Xh - Xl) plotted against variance(S2 X 102), S being the combined standard deviation for high and low readings. Pairs of
brooches are plotted once, using averaged figures
due to the small size of most of the samples, i.e. the limited number of readingswhich could be made on each brooch.
Using these graphs, the photographs and photomacrographs, and the abovecriteria, we are able to arrive at a total of at most fourteen different foils used ontwenty-eight brooches which have visible foils. The groupings are presented inTABLE IV. This is not to say that we may not have even fewer stamps represented,but our evidence and methods of detection cannot securely group the foils anymore tightly. Certain imponderables also limit our conclusions: without knowingfor certain the size of stamp used, we cannot know the amount of variation to beexpected over anyone foil; furthermore, it may be that on occasions two foils
RICHARD AVENT AND DAVID LEIGH
TABLE IV
SURVEY GROUP 2-RE-GROUPING OF DATA
DifferenceBrooch Group I
Mean Group Standard betweenor number Site spacing (x) mean deviation high and
Pair (l.jmm.) (I.jmm.) (s) low means(Xh-XI)
Di,ii 80, 81 Bifrons 42 2.362·37
0.20 0.36S ii Finglesham E2 2.38 0.18 -
Ai,ii 76, - Chessel Down 2.82 0.21 0.19
J 89 Mersham 2.86 0-40 0.3 1
Fi 84 Gilton 2·54 0.29 -
Qi,ii Chessel Down 3·352·95 0.25 -
Si Finglesham E2 3. 18 0·47 0·74
Aiii Chessel Down 3. 19 0.3 1 0.16
Ci 78 Lyminge 44 3.40 0.16 0.16H Howletts 3.5 2 3.56 0.04 0.05R Finglesham E2 3.76 0.10 0.10
P Chessel Down 3.9 2 0.28 0·34
0 Chessel Down 3.96 0.70 -
T't Faversham 4.08 0.4 2 -It
Cii 79 Lyminge 44 4. 17 0·33 0.22Ki 90 Stowting 9 4.38 4.3 1 0.23 0.21Ei, ii 82, 83 Bifrons 4·39 0.23 0.3 1
I Howletts 4·39 0.26 0.09B 77 Howletts 7 4. 10 4.2 1 0.23 0.14Gi,ii 85, 86 Howletts 7 4. 13 0.21 0.20
L* 138 Sarre 159 3·34 0.19 0.25
Mt 162 Dover 3.653·77
0.20 0.18Nt 164 Sarre 4 3.89 0.30 0.40
NOTES on Table IV:
* Boxed foilst Special Boxed foilst Irregularly Boxed foils
with very similar spacings were used on one brooch, and, using our method foraveraging calculations for a whole brooch, it would then be difficult for us todetect this.
In addition to these groupings there are a number of observations on theresults which are worthy of note. The trio of brooches from Chessel Down (A iA iii, PL. III, c) share two foils. One of these, used on two of the brooches (A i andA ii), has a spacing of 2.82 l./mm., with a standard deviation of 0.21; the other,used on the third (A iii), has a spacing ofg.1gl./mm., with a standard deviation
CROSS-HATCHED GOLD FOILS
of 0.31. This confirms other findings about these brooches which suggest theseparate manufacture of a pair and a single brooch. Diagnostic of this are:
i) the differing use of punched decoration round the outer edge of theheadplates ;
ii) the use of two garnets on the foot-plate terminal of A iii instead of a largerone;
iii) the differing use of niello decoration. Compare for example the presence ofniello on the bar of the foot terminal of A iii and its absence on the 'pair';the double zigzag on the side-terminals of A iii compared with single zigzag on the other two; the poorer execution of niello decoration round thecorners of the headplates of A iii compared with the other two;
iv) numerous minor differences in the chip-carved animal decoration; comparefor example the work on the shoulders of the foot-plates.
These differences all point to the manufacture of a pair of brooches, as was normalin the Kentish and Isle of Wight repertoire, followed by the manufacture of athird which was intended to match the pair and was probably made by the samecraftsman or craftsmen, though perhaps on a separate occasion. This trio, then,provides us with valuable corroboration of the foil evidence and of the methodsused to identify foils. The archaeological implications of this discovery are notrelevant to the present study, but it can be said, in passing, that the two otherocurrences of trios in the Kentish square-headed series - one set from ChesselDown and one set from Milton-next-Sittingbourne'" - do not appear to showthese differences of manufacture.s"
Items F, G and I (two pairs and one single) are all very similar broocheshaving, for instance, the same arrangement of garnets on the head-plate (onerectangular and two lentoid) and, so similar in most other respects that one wouldreadily attribute them to a single workshop. G and I have, almost certainly, thesame foil, one with close spacing and low irregularity, while item F (judging bythe one of the pair that has readable foils, F i) has a markedly different foil, with acoarse spacing and a more uneven quality.
Items G and I are both from Howletts, as is item B, a square-headed broochbearing a Class 2. I keystone disc brooch on its bow. All three items share the samefoil which may possibly suggest that the high quality B and the poorer quality Gand I were made in the same workshop, although we cannot be sure without moreevidence: it is possible that the same foil was available in more than one workshop.
3.; Chcsscl Down trio, British Museum Accession Numbers 67, 7-29, 14, 15, 16; Milton-nextSittingbourne, British Museum Accession Numbcrs 1905, 4-18, 19, 20, 2 I; E. T. Leeds, 'Notes on Jutishart in Kcnt bctween 450 and 575', Medieval Archaeol., I (1957), pl. iv, B; E. T. Leeds, A Corpus ifEarly AngloSaxon Great Square-Headed Brooches (Oxford, 1949), S'3; A Guide to the Anglo-Saxon Antiquities in the Departmentof British and Medieval Antiquities (British Museum, 1923), fig. 56; N. Aberg, The Anglo-Saxons in England(Uppsala, 1926), fig. 127.
36 Therc is one recently noted parallel for the enlargement of a set, in this case from one brooch to a pair,and that is from Donzdorf, W. Germany; E. M. Neuffer, 'Der Reihcngrabcrfricdhof von Donzdorf (KreisGoppingen)', Forschungen und Berichte zur Vor- und Fridigeschichte in Baden-Wiirttemberg, II (1972), 15 ff.,Abb. 5a; for illustration also see G. Haseloff, 'Salin's Style 1', Medieval Archaeol., XVIII (1974), pI. vi, c.Bakka cites another possible example, the pair from Basel-Klcinenhungcn, Grave 74: E. Bakka, 'Goldbrakteaten in norwegischen Grabfiinden: Datierungsfragcn', Friihmittelalterliche Studien, VII (1973), 78.
RICHARD AVENT AND DAVID LEIGH
Item C, consisting of a pair of brooches, C i and C ii, from Grave 44, Lyminge, quite clearly has two different foils, one of moderately fine spacing and ofvery high quality, the other of much finer spacing, but less even in its execution.In this case, there is no reason to believe that two brooches were made on differentoccasions, for they are very similar in all other respects. They are, however, someof the largest in the Kentish cabinet, and employ an exceptionally large area offoil. One can perhaps assume that the craftsman who made the pair did not haveenough foil of one sort to cover both brooches. Ifso, then we may infer that he hadsupplies of stamped foil, but not the means to stamp his own. Alternatively, hemay simply have wanted to achieve a contrast between the two brooches.
Each of the remaining pairs of brooches with readable foils (items G, D, S, Qand E) shares the same foil with the exception of the pair from Finglesham (S),which has widely differing spacings and regularities. The reasons for this are notobvious, although the Finglesham pair is a somewhat unusual example of itstype.P?
Other unusual foils came from Chessel Down, Mersham and Faversham(0, J and T). The foils on brooch 0 (PL. II, A) are quite evidently bungled: oneof the authors was able to make better examples than these at his first attempt.Other decoration on the same brooch, from Chessel Down, also appears to berather sub-standard. Item T, a most unusual pair from Faversham, also has visually poor foils, and displays a clear attempt at boxing (5 X 5). They were probablyimported from Scandinavia. Item J from Mersham, one of the earliest squareheaded brooches, has a coarse and moderately irregular foil spacing, and mayperhaps be of Scandinavian manufacture.
If we discount these last unusual and highly irregular foils, we are left with amaximum of nine standard foils in regular use on the Kentish square-headedbrooches at this time, plus a boxed and a special boxed foil.
RECONCILIATION OF TWO SETS OF DATA FOR SQ.UARE-HEADED BROOCHES
FIGURE I I is a simple graphical representation of the similarities or otherwisebetween the two sets of measurements for square-headed brooches. If both sets ofmeasurements agreed exactly they would be joined by vertical lines. As it is,there is a more-or-Iess even slope, which accords with one set of measurementsbeing about 0.2 l.jmm. higher than the other. There are clearly a few points ofdiscrepancy, e.g. item F i, but in such instances the poor quality and lack ofreadable foils make accurate measurements almost impossible.t"
37 S. E. Chadwick (Hawkes), 'The Anglo-Saxon cemetery at Fingelsham, Kent: a Reconsideration',Medieoal Archaeol., II (1958), 57-8.
38 In the course of carrying out our survey we encountered numerous difficulties in observing the foils onobjects or their photographs sufficiently clearly to provide reliable figures. These may be listed for thebenefit of other workers: i) the foils were lacking because either there was no foil placed under the garnetoriginally, or the foil and garnet had both been lost, or the garnet had been reset, without its foil; ii) thegarnet (or glass) was so blemished or opaque that the foil could not be clearly seen beneath it; iii) thepattern on the foil, though visible, was squashed or otherwise damaged; iv) in the case of a photograph, thefoil may not have been sufficiently well lit, or else light may have been reflecting off the surface of thegarnet. To these difficulties should be added the problem of attempting to compare foils, by eye andmemory, on brooches from different collections.
CROSS-HATCHED GOLD FOILS 43
Group 2
84 80 76 89 138 78 87 162 16488 82 Group 1
77 908185 86 83
79
I I I2'0 30 4·0
SPACING linesfmm
FIG. II
FOIL SURVEY GROUPS I and 2
Comparison between the two sets of data for the same (square-headed) brooches
Minor discrepancies aside, and ignoring for the moment the systematic error,the two sets of readings provide good confirmation of each other, the generalorder and groupings of the brooches being followed more-or-less exactly. Thiswelcome observation serves to bolster our trust in the larger set of measurementsfrom which we may with confidence draw conclusions about relative order anddistributions, even if we cannot employ them as accurate, absolute measurements.
CONCLUSIONS
At the beginning of this paper, we posed two questions. How were the crosshatched gold foils, which were used as backings to inlays on Anglo-Saxonjewellery,produced and is it possible to identify specific foils or groups of foils by themethods we proposed to employ in this analysis? As to the first, our experimentalwork leads us to believe that the cross-hatched patterns were stamped on tosheets of gold foil with stamps which were probably made from fine-grained woodor bone and that these stamps could have been used over a considerable period oftime. To enable us to tackle the second question, we devised two new and complementary methods of analysis and have thus been able to produce a considerable body of general information about the types and uses of these gold foils.In addition to distinguishing the main groups of standard and boxed foils, we havediscovered certain unusual foils and in particular one distinctive group which wehave called special boxed foils. We have shown that it is possible to trace anapparent chronological progression from finer to coarser line spacings amongststandard foils and an unusually high incidence of boxed foils amongst the Faversham material. A detailed, statistical, study, using photographs and photomacrographs, of square-headed brooches has shown that, at the very most,fourteen different foils were used on twenty-eight of those brooches which havevisible foils. Using these methods, it has also been possible to undertake a reassessment of a trio of square-headed brooches from Chessel Down. Further, the results
44 RICHARD AVENT AND DAVID LEIGH
of these analyses have suggested that a fixed unit of measurement may have beenin use at this time in this context. Throughout this study it has become increasinglyclear to us that the Anglo-Saxon craftsmen used these gold foils with considerableskill and sensitivity, deliberately selecting different foils for different parts of thesame object to emphasize certain features in its overall design and often takingconsiderable care to ensure that individual cell foils were aligned in such a way asto maximize their effect.
In this paper we have only been able to concentrate on one, very specific,area of technical expertise employed in the production of jewellery during thisperiod. We are certain that with more work along these lines, perhaps improvingon our techniques, it will be possible to identify individual foils with certaintyand hence relate objects more closely both to each other and to the workshopswhich produced them. When this evidence is combined with other complementarytechnological, artistic and archaeological evidence, we shall be closer to anunderstanding of the Anglo-Saxon jeweller's craft and eventually, perhaps, hisplace in Anglo-Saxon society.
APPENDIX
EXPERIMENTAL WORK
In order to test a theory for the production of foils we carried out some experimentsusing fine-grained wood to produce stamps with which to impress a cross-hatchedpattern of lines on metal foil. Tests were made on an ordinary softwood, pine, and ontwo hard soft-woods, yew and box. Two sorts of knife-blades were tested: a no. 15scalpel blade and the slightly blunted blade of a Stanley Slim-knife. Gold foil provedto be too costly for the tests; gold leaf is not of the required thickness and aluminium foilis too hard. Eventually, high-purity tin foil was chosen, since it could be obtained in thecorrect thickness (0.025 mm.), and it has similar working properties to gold. Further,in order to conserve the tin, some of the preliminary testing was done by makingimpressions in plasticine. Observations during the experiments may be summarized asfollows:Choice of wood. Pine, a particularly soft wood, had too coarse a texture and althoughthe first set of cuts was satisfactory, the second set, at right-angles, resulted in the lossoflittle chips of wood between the lines. It was also difficult to maintain an even depthof cut, and the knife tended to wander about and was difficult to control. Box provedto be much more satisfactory. It was possible to cut lines to an even depth and the hardwood gripped the knife so that one could maintain even spacing and draw straight linesby eye. There was no loss of wood when cutting the second set of lines. Although hard,it is sufficiently elastic, on the macroscopic scale, to pull when the second set of lines iscut, so that there was a tendency for the second set of cuts to draw or comb the first set,rather like a combed effect on cake icing. With care in the choice of knife and theangle at which it was held this effect could be reduced, but not eliminated. It is not aneffect we have ever observed on foils. Yew produced the best results. It was possible todraw thin, parallel lines of even width and to do this without loss of wood or anycombing effect.Choice of knife. Attempts to use the scalpel blade, originally chosen on the erroneousassumption that it would enable the necessary fineness to be achieved, were quicklyabandoned since the very fine blade tended to be too tightly held by the wood, and thechannel cut by the thin blade was so narrow that some displacement of wood took plaee
CROSS-HATCHED GOLD FOILS 45on either side of the groove. This resulted in the appearance of two grooves on eitherside of the ridge in the plasticine impression. The slightly blunted Stanley-knife bladeproved much more satisfactory and, provided it was held at the correct angle, thedisplacement effect would be avoided.
Angles of cut. An important lesson was that the angle of cut must be diagonal to thegrain direction of the wood. If either direction of cut was allowed to follow the grainthen the grain took hold of the knife and determined its direction, rather as tram-linesused to trap bicycle wheels. In the few examples of foil where we have been able toobserve grain, the angles of the grooves are not parallel to the grain, though not necessarily at 45 0
, as they ideally ought to be. In the vertical plane, it was found preferable tohold the blade at a low angle to the wood, so that a length of blade was in contact withthe wood, rather than using it nearly vertically with only the point in contact. Usingthis low angle of cut on yew the combing effect could be entirely eliminated.
Line-spacing. 10 mm. squares were first marked out in I mm. units and an attemptwas made to cut three to four lines to the millimetre, trying to keep the lines paralleland of even depth. To the experimenter's surprise he was able to carry out this task withreasonable success after only one or two attempts, finding it quite possible to cut linesas close as this, if not as evenly as might be wished, nor always parallel. The experimenter is somewhat myopic and normally wears spectacles. However, he found itslightly easier to do the work without spectacles, perhaps implying that such work wasbest done originally by short-sighted individuals. The work soon caused eye-strain,however, and later parts of the experiment were conducted using a low-power stereomicroscope (x 6). As skill developed it was found that the spacing of the lines could bemade more even if they were cut entirely freehand without regard to I mm. marks.No straight-edge was needed as a guide since it was very easy to cut straight lines in thewood for distances up to about 20 mm., beyond which movement of the hand wouldhave caused a jerking effect. This may be an important observation since if the lineson our foils were indeed created by freehand methods, we have some indication of themaximum size of anyone stamp.
Line depth. By measuring with a microscope it was possible to monitor the line depthsproduced and, by varying the pressure on the knife blade, achieve a depth of about0.03 mm., corresponding with measured depths on some original foils.P"
Boxing. Wc wished to know whether boxed (strong) lines were created first and thefiner lines drawn between them, or vice versa; or whether the strong lines were drawn ontop of existing fine lines on a standard pattern. The experiments showed that by far theeasiest method is to draw the fine, standard pattern first, and then to draw the knife overselected fine lines afterwards in order to strengthen them, This bears out our observations on certain of the boxed foils, where divergences in direction of the strong linesenable the original fine lines to be seen in place (PL. I, D). Using this argument, we cansee that it should have been possible, in theory at least, to convert a stamp of standardpattern to a stamp of boxed pattern simply by re-cutting some of the lines.
Production ofstampedfoil. Although the test patterns were cut on a larger piece of wood,it was found easier to apply the necessary hammer blow if the surface area of the woodenstamp was reduced to a minimum, so that the stamp formed, in effect, the tip of a handtool. In this way all the pressure could be concentrated on the patterned area. This isanother observation to support the idea that not very large areas of foil were stamped atone time. The tin foil was placed on various supports: lead sheet, leather and beeswax.
39 This figure is based on nine readings made on two loose foils from Taplow in the British Museum.The measurement of depths of grooves or heights of ridges on loose pieces offoil was accomplished with thehelp of a Wild M20 research microscope. The focusing knob on this is calibrated in microns (jrrn) so thatby focusing first on the top of a ridge and then at its base one is able to calculate the depth by simplesubtraetion.
RICHARD AVENT AND DAVID LEIGH
Tolerable impressions could be obtained on all three, although lead seemed to bepreferable. The best results were obtained by hitting the stamp on to the foil with asharp blow of the hammer.
RESULTS
The stamped tin foil that we produced (PL. III, B) presents a tolerable imitation of thegold foils which we have observed. There is no mistaking the difference, however, andwe may ascribe this to a number of factors: gold foil would probably behave slightlydifferently from tin foil; a better support medium could well have been used (pitch?);(NB.: It was noticeable that better results were produced by impressing into plasticine(PL. III, A) rather than into foil on lead, though foil on plasticine did not work at all);there may well have been a better blade to use for cutting the stamp; the skill of theexperimenter left much to be desired and would probably have improved with morepractice and a greater aquaintance with fine metalworking techniques; no doubt thereare other 'tricks of the trade'.
ACKNOWLEDGEMENTS
The general survey of foils was prepared with the help of an award from the Society for MedievalArchaeology's Colt Fund and we are grateful to the Society for this assistance. We are also grateful to thefollowing for allowing us ready access to objects in their collections and for their help at various timesduring our research: Mrs Leslie Webster, Department of Medieval and Later Antiquities, British Museum;Miss Louise Millard, Royal Museum and Public Library, Canterbury; John Musty, Ancient MonumentsBranch of the Department of the Environment; Mrs Margaret Warhurst, Merseyside County Museums,Liverpool; David Kelly, Museum and Art Gallery, Maidstone; David Brown, Ashmolean Museum,Oxford, and Lord Northbourne.
We also wish to thank Drs ], Haworth, A. B. Nix, and B. A. Rogers of the Department of Mathematics,University College, Cardiff, for helpful discussions on the statistical problems and for suggesting themethod of plotting variances; the Forest Products Research Laboratory for supplying samples of yew andbox woods; Andrew Oddy of the British Museum Research Laboratory for providing microscope facilities;the Departments of Archaeology at Southampton University and University College, Cardiff, for microscope and photographic facilities; Mrs Jean Barber and Miss Sabina Thompson for assistance with thetyping, and Howard Mason and Keith Bellamy for preparing the drawings and photographs respectively.Our special thanks go to Dr Tania Dickinson for reading the manuscript and making many helpfulsuggestions.
NOTEThe Society is much indebted to the Council for British Archaeology for a grant towardsthe cost of publishing this paper.