Footgear — Pins, Brooches and other Dress-fasteners —
Tweezers, Nail-cleaners, Ear-picks, Mirrors, Combs, and Dressing-boxes —
Bracelets and Armlets, Finger-rings, Ear-rings, Beads and Necklaces

Examples of the footgear of Roman Britain have been found in many places where the conditions were favourable for the preservation of the leather, notably in London and at Bar Hill and Newstead. Roman writers distinguished several varieties. The solea and sandal, bound to the foot by straps, was not ordinarily used out of doors. The calceus, the close-fitting boot which completely covered the foot, was the national foot-attire for public occasions, and etiquette ordered that it should be worn with the toga in the city. It was secured by straps, which were wound round the lower part of the leg and tied in front; and their number, colour, and other details marked the rank of the wearer. The boots of the ordinary citizens were not so high, and were fastened over the instep by tongues or latchets extending from the sides. Between the sandal and the boot were various transitional forms which may be generically classed as shoes. The gallica had low sides with loops, through which a thong was laced to secure it to the foot, and the crepida appears to have been similar; and both were sometimes classed as soleae. The caliga was the strongly made sandal-like shoe with open sides, worn by soldiers, and held by straps wound round the leg. It was also worn by the inferior officers, but the higher officers wore the calceus. The soccus was a light low shoe answering to our slipper. The carbatina, apparently made of a single piece of leather, was used by rustics. The cothurnus was a hunting-boot, and custom demanded that it should be worn by tragic actors, as the soccus by comic actors. The differences between some of these have not been satisfactorily determined; still less can the Roman names, and the classification they imply, be satisfactorily applied to the footgear of Roman Britain. If by solea is understood a simple sole held to the foot by straps, it was rarely used in this country, for the large 'find' at Bar Hill yielded only one. On the other hand, a large number of shoes with low openwork sides or borders have been found, and these are usually described as sandals. They appear to correspond with the gallica and crepida — half-sandal, half-shoe. These pass, however, into the shoe which wholly encased the foot, some with openwork and others with solid uppers, and the shoe passes into the boot, of which only few examples have been found in this country. The shoes are of several types, and one of these may be the carbatina. Many of the Bar Hill shoes were certainly worn by soldiers, but none quite answers to the classical caliga.¹

The shoe was evolved from the sandal. The addition of a heel-piece and toe-cap gave the sandal a firmer hold to the foot; and by extending the heel-piece forward on either side as a tongue-like projection with an eye to receive a thong or lace passing over the instep, the strap could be dispensed with. Fig. 69, A, a child's shoe in the Guildhall, illustrates the outcome. The uppers are of two pieces of leather sewn together at the heel and the toe. They are solid for nearly an inch all round to serve as a sheath to protect the foot against stones and mud; but above that level, portions are cut out so as to leave a framework of narrow bands, apparently a survival of the straps of the sandal. From the lace-holes, the bands radiate to various points between the top of the back and the 'waist' of the sole, so that the pull of the lace is well distributed. Over the toe they run transversely, just in the direction where strength is required. An elaborate man's shoe of the form was found at Bar Hill; and in the Guildhall is an unusual variant in which the whole of the uppers is reduced to a mere skeleton of slender bands reaching down to the sole.

Fig. 69. Shoes of several types. (C, 1/2; all the rest,  1/4)

It is obvious that shoes like these, with their uppers reduced to mere filaments of leather, were only adapted for light wear. Not so the child's shoe from Bar Hill, D, which has a sturdy workaday look, and its grip to the foot is increased by a second pair of latchets. It is the type of a large class of shoes adapted for hard wear, to which many of the Bar Hill specimens belonged — presumably soldiers' shoes. These shoes were sometimes ornamented with punched work, but only sparingly so, and the leather was never reduced to bands. Those intended for heavy wear had usually a 'counter' — a stiff piece of leather to support the back of the heel.

Another type of shoe suggests a different line of development from the sandal. If the heel-piece is continued along each side of the sole to the point as a low sheath or kerb with a marginal series of holes, through which a thong can be laced from side to side over the toes and instep, we have an incipient shoe which becomes more shoe-like, from the modern point of view, by the development of the kerb. Fig. 69, B, is one of the side leathers of a shoe of the kind in which the kerb is moderately developed. Carry the development further, the lace-holes will meet and the foot will be completely enclosed.

We have now arrived at a form which resembles the modern laced shoe, except that as a rule the lacing started from much nearer the point than at present. Some of these shoes were elaborately ornamented. One in the Guildhall has the lace-holes elongated into loops and the sides are covered with a finely punched diaper with rosettes at intervals, as the first example in C. Part of another with equally elaborate patterning was found at Bar Hill. Two other examples of punched work found in these and shoes of other types are given in C. In a variant of the above type, the lace-holes of the one leather are developed into long loops which reach over the foot to those of the opposite leather. F is a restoration of a woman's shoe of the kind, in the Guildhall. In the same collection is a boy's boot, which represents an extreme variant in another direction, and remarkably anticipates the modern laced boot. The upper, which is solid, is sewn together almost as far as the bottom of the instep, and extending from this to the top of the boot are oval lace-holes, ten on either side, within a scalloped margin as in B.

Some shoes may be regarded as of mixed type. The boy's shoe from Bar Hill, G, has two heel latchets in the form of long loops, a pair of side loops, and a pair at the point. E, in the Guildhall, is a more elaborate example, and Mr. Roach Smith figures another still more advanced which combines the side-laced form of F, with heel-latchets.²

In another and primitive type of shoe, sole and uppers are made of a single piece of leather, but occasionally the sole is fortified by an additional leather. Several examples have been found at Bar Hill, one at Netherby, and another at Birdoswald on the Wall of Hadrian. In these shoes the only seam is up the back of the heel; each side is cut into two latchets with lace-holes; but the distinguishing feature is the manner in which the toe-cap was formed. This, as will be seen in the Birdoswald shoe, H, was accomplished by cutting the leather into a series of wedge-shaped strips, each with an eyelet at the end. These strips were then bent back, and the eyelets threaded together, presumably by the lace. Dr. Haverfield suggests that this kind of shoe was the carbatina, and mentions that it is still used by the Carpathian hillmen and by peasants in Italy, Roumania, and Bulgaria.³

The soles of the sandals, shoes and boots closely approximate to that of the foot. Not seldom the first or the first and second toes were indicated, and occasionally all the toes as in I, a sole in the Guildhall Museum. J, another Guildhall example, is a typical sole of the coarser shoes intended for rough wear, and it will be noticed that it still conforms to the natural shape. The sole is usually of three or four layers of leather with a thinner insole, and the heel is never raised by additional layers. In sandal-like shoes with low openwork sides, the upper is sometimes of a single piece of leather continuing across the sole; but most often, the upper is of two leathers with their lower margins tucked in between the insole and the sole. The whole fabric was fastened together by nails clenched on the insole, but this was occasionally done by stitches in the lighter shoes.

A notable feature of the soles of the era is the armature of hob-nails on the under surface, not merely of men's, but of women's and children's footgear. Even the lightest and most elaborate shoes usually have it, and the exceptions are comparatively few. The nails are arranged in a variety of ways. Occasionally they are loosely scattered all over the sole, or are scattered in clusters of threes; or they are confined to a marginal row all round the sole. Men's soles were usually thickly studded, the nails within the marginal row being often arranged in some pattern as indicated in J, or in close rows leaving little of the leather visible. The custom of thickly studding soles with nails was common in Italy, and Pliny in describing a peculiar fish likened its scales to the nails of a sandal. Pitt-Rivers found, with the hob-nails at the feet of two skeletons at Rotherley, several cleats from 1 1/2 to 1 3/4 ins. long, the use of which he compared with that of Blakey's boot protectors.⁴

Fig. 70. Pins, Tweezers, Nail-cleaners, and Ear-picks. (All 2/3)

We now consider some articles for fastening the attire. Of these, pins are the simplest, perhaps the most ancient, and are among the most numerous objects found on our Roman sites. They are mostly of bone and bronze, the exceptional materials being ivory, jet, silver, iron, and even glass. They are rarely less than 2 1/2 ins. or more than 6 ins. long, and while the general form is necessarily constant, they differ greatly in the form and ornamentation of their heads. The simplest are mere skewers of bone shaped by hand and with ill-formed heads; but the majority have been turned in the lathe, and in the more elaborate the heads are enriched with carving, sometimes taking the form of statuettes, busts, animals, and birds, and occasionally those of the bronze pins are enamelled. The York Museum contains a fine collection of these articles, and among its rarities are bronze pins with glass heads, bone and ivory pins with jet, agate, and silver heads, and another with a gold head. The pins shown in the two groups, A and B, Fig. 70, respectively of bronze and bone, are selected from London, Silchester, Caerwent, Rushmore, Woodyates, and Spring-head (Kent) specimens. The hooked head of the last bronze pin but one is unusual, and the last bone pin, in the Guildhall, is remarkably large, and the bust is supposed to be that of the Empress Sabina.

Pins were used for the hair as well as the dress, but there does not appear to be any special feature, either in the general form or the ornamentation, by which they can be distinguished. Probably they were used to some extent indiscriminately. We may, however, draw the broad distinction that the smaller and more attenuated were dress-pins, and that the larger and stouter were hair-pins, and from this conclude that as a rule bronze pins fall under the one head, and bone and jet pins under the other. The materials of the latter being light would render them specially appropriate for the coiffure; as also the entasis of many of them, which, by increasing their hold, anticipated the advantages of the modern 'curved' and 'falcon' hair-pins.

Pins are frequently found in the graves of the ladies of the era, and their positions often indicate whether they were used in the dress or the coiffure. There are two good examples of the latter in the York Museum. In the one, the lady's hair is still intact, and is plaited and made into a coil on the back of the head and held in position by two jet pins. In the other, there are three jet hair-pins, two small ones and a third, 7 ins. long, with a perforation near the point. Apparently this pin was threaded with a fine cord, which, being drawn over the hair and caught under the knob and tied, effectually secured it to the head. Similar large pins with eyes have been found elsewhere. In Smith's Dictionary of Greek and Roman Antiquities is shown, under 'Acus,' a female head in marble at Apt in the South of France, with the hair plaited and coiled at the back, the coil being kept in position by a single large pin. This simple style of coiffure was characteristic of the third and fourth centuries, and it contrasted with the extravagant head-dresses of the earlier Imperial period, which met with strong disapproval from the early Christian writers, as expressed in 1 Tim. ii.9 and 1 Pet. iii.3. In these elaborate productions many pins must have been used. The Apt treatment of the hair lingered to our own times in Italy and some parts of Germany.

Brooches are almost as frequently found as pins. The older antiquaries regard the brooch as a Roman introduction, but there is abundant evidence that the natives were familiar with it before the conquest, not only as an imported article, but as a product of the native metal-worker. Most of the early Continental forms have been found in Britain, and most of the forms associated with Roman remains had already been developed before the Romans appeared on the scene.

The brooches of Roman Britain may be conveniently, and on the whole satisfactorily, classified as bow-, plate -, and ring-brooches. The first were the most numerous, and, divested of their ornamental and other non-essential features, resembled the modern safety-pin. The second were an extreme variant of these, in which the bow or arch was replaced by a more or less flattened disc, rosette, or some other geometrical or animal figure, in this respect foreshadowing the generality of modern brooches. The last stand markedly apart, were apparently derived from the buckle, and have no modern representatives except in Algeria and elsewhere in northern Africa. These various brooches were mostly of bronze; sometimes of bronze-gilt, of silver, and even of gold. Enamelled enrichments were frequent. As a rule their workmanship was excellent, such as could only have been accomplished by craftsmen of skill and experience. Many certainly were imported; but there is little doubt that the majority were made in Britain, and these indicate that in this particular branch of industry the home metal-worker rivalled, if indeed he did not surpass, his Continental brother. The ornamentation sometimes consists of Late-Celtic designs of considerable purity, and these are most frequently seen on brooches found in the north and 29, where Roman influence was less felt than elsewhere. But even in a small collection of the brooches of the era, an experienced eye will hardly fail to detect survivals of these designs and a general Late-Celtic feeling.

The Bow-brooch, or fibula as it is customarily called — an arbitrary but convenient limitation of the word — was of ancient lineage and varied form and construction; and its history has received much attention of recent years, in this country especially from General Pitt-Rivers, Dr. Arthur J. Evans, Prof. Ridgeway, and Mr. Reginald Smith. It appears to have been derived from a simple Italian form of the Bronze Age, which anticipated the modern safety-pin in its earlier form when it was made of a single piece of wire, Fig. 71. This primitive brooch once established, it was inevitable that there should be developments in various directions. The bow was soon thickened so as to give it greater rigidity, and it became more arched so as to enclose a larger volume of the dress — thus arose the 'leech-shaped' brooch, Fig. 71. Continuing to expand, it was next made hollow for the sake of lightness, and thus became the 'boat-shaped' brooch. A lateral angularity gave the boat a lozengy shape, and eventually the angles were capped with knobs. At first the catch was a simple crook; then it was developed in a forward direction into a horizontal spiral for the point of the pin to lie upon, thus answering to the guard-loop of our safety-pins, and this eventually became a solid disc. Meanwhile a new form of catch arose, by beating out the foot of the bow and curling up its lower margin to form a hollow to receive the pin, and this was soon extended anteriorly to cover its point. The spring-coil, which at first was of a single turn, was given a double turn to increase its elasticity. All these Italian developments had long been accomplished before the conquest of Britain, but a few examples have been found in this country, probably importations of an earlier period.

Fig. 71. Modern Safety-Pin, Italian Leech-shaped Brooch, and Late-Celtic Brooch. (2/3)

While forms that appear to be later developments of the Italian brooch are found on our Roman sites, the ancestors of the generality of the Romano-British fibulae are to be sought in those of the Iron Age of the Swiss lake-dwellings and of central Europe generally. Their type, which is generally known to us as the Late-Celtic, and on the Continent as that of La Tene, shows a marked advance in construction. The spring is now bilateral, that is, it consists of two coils of two or more volutions each, the outermost of which end in a transverse loop or chord connecting the two coils, as shown in Fig. 71. The catch is equally noteworthy. The foot of the bow is produced horizontally, and its side is manipulated into a curled-up flange to receive the pin; but instead of ending with the point, the bar is turned upwards and backwards to the bow, and usually ends in a knob or disc. This form of the Late-Celtic brooch persisted for a long period, and many examples have been found in our southern counties, occasionally associated with Roman remains; but it is a pre-Roman form.

Fig. 72. Bow-shaped Brooches of several types. (All 1/1)

A variety of this brooch, and probably a later one, is more often found with Roman remains. The foot of the bow is beaten out into a plate with the lower margin curled up to form the catch; and the chord is nearly always turned inwards and presses against the root of the bow. The solid triangular catch-plate of the Romano-British bow-brooches, noticeable in all the examples of Fig. 72, may have been derived in some measure from the foregoing; but Dr. Evans has pointed out a series of transitions between the normal La Tene catch with its retroflected 'tail' and these plate-catches. First the tail was united to the bow when the body was cast, the triangular open space was retained, but the portion representing the 'tail' became an integral part of the bow. The space was next encroached upon or partially filled with ornamentation, and it then assumed a plate-like character with pierced ornamentation, its sole function being to carry the catch. Finally it became a solid plate. The effect of these changes was to make the catch an internal, instead of external, feature (compare Figs.  71 and 72).

Meanwhile, the spring was subject to modification. There was an early tendency to reduce the diameter of the coils, and, in compensation for the loss of elasticity thus incurred, to increase the number of volutions. This lengthening of the spring correspondingly lengthened the chord, thereby reducing its resistence to torsion, hence, upon closing the brooch, its 'play' resulted in some displacement of the coils. One early remedy for this weakness was the insertion of a rivet through the coils. Another and more effectual remedy was the introduction of two wing-like plates or bars, one on each side of the head of the bow and immediately over each coil. In order to tighten up the coils to these, the chord was caught over a small spur at the back of the head, and this was eventually converted into a loop or eyelet by being lengthened and hammered back to the bow, the point being often secured by a rivet. Presently the plates became semi-cylindrical so as to sheathe the upper halves of the coils. Then their ends were boxed-in and drilled to receive a rivet which passed through the coils, the pin and spring being now a separate entity held in position by this rivet. At this stage the eyelet was drilled in a small cast lug, with an ascending tail reminiscent of the upturned portion of the spur.⁵ We now leave the T-fibula to follow up a cognate line of development.

We return to the short La Tene spring with the chord turned inwards. The first development was an expansion of the root of the bow to cover the spring, and this generally took the form of an inverted trumpet-bell, as in Fig. 73, A, B, both from Caerwent. At first the pin was in one piece with the bow, but eventually it was separately made and held in position by a central lug under the 'bell' with two perforations, a forward one for the chord and the other for the axle which held the coils. Later, this gave place to two lateral lugs to hold the axle, the spring being between them. We have now arrived at the transition of the spring and hinged pins. The chord no longer attached to the head, allowed of the pin being rotated, until, in the act of closing it, the chord came into contact with the margin of the head and brought the spring into operation, as in Fig. 73, A. Perhaps this development of the trumpet-headed fibula suggested a corresponding modification of the T-fibula. By dispensing with the eyelet at the back and placing the straight chord on the opposite side, the same action was attained as in Fig. 73, D. In either case the step to the true hinged pin was a short one; but its introduction wrought a modification of the coil-sheaths of the T-fibula, which were now made solid and perforated longitudinally for the rivet. These were unnecessarily long for the purpose, but continued to be a prominent feature as they contributed to keep the bow at right angles with the surface of the dress. Still there was a trend of modifications in which they diminished in length, and this was correlated with a compensating change in the bow in which it ceased to be bar-like and assumed a light and strap-like form, as in Fig. 72, D, E.

Fig. 73. Bow-shaped Brooches of several types. (All 1/1)

It was a British custom, both before and during the Roman era, to wear brooches in pairs. Several examples have been found with their components linked together with chains, and rings for their attachment, or the attachment of cords, are common enough. The ring was either manipulated out of the rivet wire of the spring, as in Figs. 72, C, and 73, A; or was in one piece with the bow, that is, cast with it, as in Figs. 72, B, and 73, B, C. In the former, the neck of the loop was confined by a small ring, but more usually with an oblong clamp, as in Fig. 73, A. In order to keep these wire loops in a horizontal position, there was a small spur projecting from the back of the root of the bow. It is seen in Fig. 72, A, in which the loop is lost, and has been replaced by a simple rivet for the pin.

We have now carried the evolution of the bow brooch through two concurrent types — the 'T' and the 'trumpet,' each beginning as a spring brooch and ending as a hinged one. To these in their later developments belong most of the Romano-British fibulae. True it is, that there are many forms which do not at the first sight seem to conform with these types, but they generally prove to be of intermediate character, or their fundamental identity is obscured by abnormal developments of the bow, the head, or the foot. In every large collection of these objects there are forms so fundamentally different as to suggest some other origin altogether, and they may prove to have been evolved on the Continent.

Fig. 73, E, is one of these aberrant forms, and belongs to the 'cruciform type.' It differs from the foregoing types chiefly in its catch being external to the bow, and somewhat box-like with a slit in the side for the pin to enter. The cross-bar is generally long and terminates in knobs, and there is usually a knob behind the head of the bow. There are several variants of the type. The catch-bar is especially subject to modification, being often wider and longer than in our example, and its upper surface decorated. Sometimes it is a conspicuous feature and assumes a fan-like form. The bow may be short and wide, and the cross-bar plate-like. D is an unusual example from Charterhouse, lacking the knobs and having a 'spring-hinge' pin, instead of the almost invariable hinged pin. Gold brooches of the type have been found at Odiham, Hampshire, in Scotland, and elsewhere. There is little doubt that these cruciform brooches are late Roman, and are the precursors of the remarkable barbaric fibulae which followed the Roman era, transitional forms of which are illustrated in Hans Hildebrand's Industrial Arts of Scandinavia.

We now pass to the Plate-brooch. This form of brooch is unquestionably very ancient, but there is little doubt that it was derived from the bow-shaped brooch. The 'plate,' as we have already noticed, represents and plays the part of the bow, but it apparently began as an ornamental feature of it. Whatever its origin, however, the plate-brooch appeared on the scene of Roman Britain fully developed. The 'plate' afforded ample scope for the display of artistic ingenuity. In its simpler, and perhaps earlier, form it was a metal disc, flat or centrally raised like a button or the head of a large stud, with turned mouldings and usually a central boss or knob. In a favourite design there was a broad cavetto between the central ornamentation and a beaded margin, and this was sometimes relieved by spoke-like ridges or plates, or the whole central portion was treated as a rosette. The margin often had a series of small rounded projections. Occasionally the 'plate' had the form of a wheel with four spokes, the spaces between these being pierced. Other simple geometrical forms, as squares and lozenges, were less frequent, and these also were often bossed or domed and their margins relieved with roundlets or other ornaments. Combinations gave rise to more elaborate forms, as four discs arranged quatrefoil fashion, and elongated patterns consisting of two discs or lozenges united side by side, or of one central disc or square with two triangular or peltate wings. Brooches in the form of animals were not uncommon, horses, hares, birds, and fishes often displaying a barbaric quaintness, being the favourite subjects. Many were enamelled, and on no other class of objects is the art of the Romano-British enameller better seen or studied. With few exceptions the enamel was champlevι, that is, it was deposited in cavities in the metal basis. Sometimes a considerable expanse of metal was visible, and served as the ground of the enamelled ornamentation; but usually the visible metal was reduced to narrow walls or ridges which separated the different colours. The designs were mostly geometrical, as 'checks' of two colours arranged chessboard-fashion, concentric zones of different colours, roundlets of one colour on a ground of another, and so forth. Delineations of animals seem never to have been attempted, and those of foliage only rarely. Occasionally the brooches were decorated with 'mosaic' enamel. In these, metal walls were dispensed with, or were confined to the primary divisions of the design, and the chief feature was the fine patterning of minute rosettes, squares, crosses, spirals, dots, and 'checks,' built up in the same manner as millefiori glass (p. 180). Fine examples of these brooches have been found at Caerleon, Lydney, and Rushmore.

Fig. 74. 'Plate' Brooches, mostly enamelled. (All 1/1)

The examples of plate brooches in Fig. 74 will give the reader an idea of the diversity of their forms and decoration. In A, from Caerwent, the projecting roundels contain green enamel and the central knob one of darker colour. B, from London, has blue and green enamels. C, from Caerwent, is of unusual form, representing the Greek omega, and without enamel. D is a Lydney example bearing traces of enamel, and with an open centre. E, from Richborough, has an inset of white enamel in its raised centre. F is an openwork brooch from Caerleon, with remains of rich blue enamel. G, a peculiar peltate form, is enamelled in red, blue, green, and yellow, and its design has a marked Late-Celtic feeling. It was found at Wolvershill near Banwell, and similar brooches have been found at Castor, Irchester, and Leicester. H is another Caerwent example with red and green enamels and a pierced centre. I, from Wappenham, Northamptonshire, is of tinned bronze, with seven studs of bone held by bronze rivets, the intervening portions of the plate being engraved. A quatrefoil brooch of the same unusual decoration was found near Ipswich. Fig. 75, A, B, are two examples of enamelled zoomorphic brooches, the one found in Gloucestershire and the other at Rotherley.

Our next is a typical example, Fig. 75, F, of the S-shaped or dragonesque brooches, a small but highly interesting class which may be regarded as a connecting link between the plate-brooches and the ring-brooches next to be described. It was found at Faversham, Kent, and resembles the letter S with its serifs developed into grotesque and somewhat horsy heads with large ears and attenuated necks. The curved pin is loosely coiled, as in the ring-brooches, round one of the necks. In using the brooch, the pin was thrust through a sufficient volume of the dress, and its point was passed between the lower neck and the body, the pressure of the dress keeping it in that position. All these brooches appear to have been enamelled, the usual colours being red, blue, green, and yellow, and in both shape and decoration they have a strong Late-Celtic feeling. There are about eighteen known examples found in this country, and a few have also been found on the Continent.

Fig. 75. 'Plate' and 'Ring' Brooches, Studs, and other Dress Fasteners. (All 1/1)

The Ring- or Penannular-Brooch is a common object in almost every collection of Romano-British antiquities. It is a simple yet ingenious contrivance. After inserting the pin in the dress, the ring was revolved until its gap was above the point; then the pin was pressed down, and the pin having passed through the gap, the brooch was 'locked' by giving it a quarter-turn, the pressure of the dress within the brooch maintaining the ring in this position. These brooches were rarely larger than 1 1/2 ins. in diameter, and their decoration was almost exclusively confined to the terminals of their rings. The simplest terminals were effected by hammering back the ends of the wire of the ring, as in Fig. 75, C, the returns being slightly ornamented by groovings. More pleasing terminals were knobs, which were usually grooved longitudinally. In a variant of this, as in D, a Guildhall example, the knobs are bent back, and the ends of the wire were sometimes flattened and made to imitate serpents' heads. In E, from West Coker, the terminals are rectangular plates.

It is remarkable that the ring-brooch should not have got beyond an elementary stage in Roman times, considering its wonderful developments a few centuries later, especially in Ireland and Scotland; but these developments, it must be admitted, seriously reduced the usefulness of this form of brooch.

Besides pins and brooches, other forms of dress-fasteners have been found on our Roman sites, but they are far from common. About half a dozen bronze studs have been turned up at Silchester, with flat (as Fig. 75, G), convex, and conical circular heads, H being unusual in having two shanks. There are several in the Guildhall, one with an enamelled head. In the same museum are about eighteen double hooks of bronze, which are described as dress-fasteners and might well be called hook-links. The simpler are of a single piece of bronze wire, flattened and twisted in the middle with the ends pointed and bent into hooks, as I. Others are more or less elaborate productions in wirework, as J, the framework of which is wrapped with thin coiled wire and ornamented with three beads. Small dumb-bell-shaped objects of bronze and bone have been found at Newstead and elsewhere, which appear to have been used as the 'frog' buttons or 'olivets' attached to the 'loops' of modern military tunics, that is, a cord from one side of the garment was secured round the middle, and the dumb-bell was buttoned through a cord-loop attached to the other side. The curious bone objects, shaped like a corkscrew handle, K and L, from London and the Victoria Cave at Settle, were probably for the same purpose. A variant of these fasteners consisted of a disc like that of a stud, but with the shank developed into a horizontal loop by which it was attached to the dress by a braid or cord. Two with enamelled discs have been found at Slack, and it is probable that some of the small enamelled discs, which have been described as the fronts of brooches, were really the heads of studs or these dress-fasteners.

Of toilet requisites, tweezers, nail-cleaners and ear-picks are seen in most collections of British Roman antiquities. The first (volsellae) were used for removing superfluous hairs, and are ordinarily a narrow band of bronze bent into the form shown in Fig. 70, E, the looped head serving the double purpose of increasing the elasticity of the arms and of providing an eye for a ring or cord. More elaborate examples have solid handles, turned and finished off with a knob. F, from Rotherley, is of wire doubled upon itself and twisted to form a handle, the free ends being flattened to form the arms. Nail-cleaners are usually narrow plates of bronze notched at one end to form two sharp points and with an eye at the other for suspension, as in H, L, and J. I, from Cirencester, is unusually large and is ornamented with engraved lines and concentric circles. K, from Lydney, has a handle turned with many mouldings; and others are of wire hammered flat below, and twisted above to form a handle with a loop at the end. Occasionally they have only a single point. Ear-picks resemble diminutive spoons with minute bowls, and the simpler sort are made of bone or of flat strips of bronze as in N. M, from Rushmore, has a bar-like handle, turned above and ending as usual with an eye.

These instruments are often in sets of two or all three, threaded on a ring, like the tweezers and nail-cleaner, D, found in London. The rings are as a rule quite plain, but sometimes they are ornamented, or one side is developed into an ornamental plate; or a bronze band bent into an arch and united at the base by a bar takes the place of a ring. In another London set, C, the nail-cleaner of which is of unusual shape, the instruments are riveted together. Very rarely two functions may be combined in a single instrument, as the combined tweezers and ear-pick, G, also found in London.⁶

Mirrors (specula) are rarely found, but the Colchester cemeteries have yielded a considerable number. Although looking-glass — glass backed with a metallic film — was known to the ancients, its use was exceptional, and no example of it has been found in this country. The Roman mirrors were ordinarily of white bronze or yellow bronze plated with tin or silver, and were highly polished. They were, as a rule, circular discs with handles, which, although of excellent workmanship, were rarely if ever ornamented to the same degree as the Etruscan and Late-Celtic mirrors, and compared with the latter the examples found in this country are much smaller.

There are twenty-three hand-mirrors in the Colchester Museum, of which more than half retain their handles. According to Mr. A G Wright, they range from 2 1/2 to 5 ins. in diameter, and are mostly of white bronze, the rest being of pale yellow bronze plated with tin, and several apparently with silver. The reflecting surface is slightly convex in order that the image of the face or the head, being reduced, may be seen as a whole within the field. The front is in some cases bordered with an engraved band, a row of ring-and-dot ornaments, or a row of small perforations; while the back is generally relieved with concentric groovings. Fig. 63, D, presents the back of one of these mirrors, which is further ornamented with a marginal row of conical depressions. Its looped handle is thoroughly typical, and is surmounted with a trilobed plate which is soldered to the back of the disc.

Another form of Roman mirror — the box or pocket mirror — is of rarer occurrence. A fine example was found at Coddenham, Suffolk, in 1823.⁷ It was nearly 2 1/2 ins. in diameter and 1/4 in. in thickness, and the two halves — the lid and the box — were made of a bronze medallion of Nero, each half containing a small convex tinned reflector. In the Colchester Museum there are four rectangular mirrors ranging from 3 3/4 by 3 Ό ins. to 6 by 5 ins., which are quite plain, and with little doubt were fitted in the lids of toilet-boxes.

The comb (pecten) was in common use among the Greeks and Romans. Those of the latter were mostly of bone and box-wood, and the employment of this wood for the purpose was so prevalent that buxus was an alternative name for this toilet appliance. Wooden combs were used in Roman Britain, but, as might be expected, only a few specimens have survived, those usually found being of bone. The ordinary form was double, that is, it had two rows of teeth, one on either side of the body, the teeth of the one being coarse and of the other fine — a form that continued throughout the Middle Ages and still survives in our 'tooth-combs.' Fig. 63, F, is a wooden example in the Guildhall Museum. The bone combs were often made in several pieces held together between two strips or cleats by means of rivets; and if made of a single piece, the cleats were used as stiffeners. E is a typical example from Woodyates,⁸ both in form and construction. It appears to have been originally held together by bronze rivets, but was afterwards repaired by iron ones. The cleats are the only portions which offer a field for ornamentation, and in this case it consists of parallel grooves. One found at Wroxeter has a row of concentric circles between two beads; but the ornamentation is never elaborate. Combs of similar forms and like construction are frequently found with Anglo-Saxon remains, and it is sometimes difficult to distinguish them from the Roman. Metal combs are rare. One of bronze exactly resembling a modern tooth-comb, only larger, and a similar iron one, were found at Chesterford.⁹

Small combs with a single row of teeth and flat triangular backs are occasionally found on Roman sites, and a plain example was turned up at the last place. Similar Continental examples, more or less ornamented and with cases to sheathe the teeth, were evidently pocket-combs. They are regarded as late Roman, and were apparently the prototypes of the larger Anglo-Saxon combs of the form. A small comb with an ornately shaped back and converging teeth found at Wroxeter¹⁰ may have been worn in the hair, as part of the coiffure. A comb and a large hairpin were found adhering to the hair of a lady in a coffin at York.

The remains of small ornamented boxes have been frequently found in the graves of women, and their scattered contents, which usually include brooches, bracelets, and other articles of the toilet, show that they were dressing- or trinket-boxes. One is sculptured on the tombstone of the Palmyrene woman at South Shields, Fragments of many derived from the local cemeteries are to be seen in the Colchester Museum, and the woodwork of one of these has so far survived as to show that it was neatly dovetailed at the angles. The mountings of these caskets, mostly of bronze, consist of ornate corner-pieces and plates, held in position like their lock-plates (of which two are shown in Fig. 68) with ornamented nails, ring and other hinged handles, bosses, and various ornaments. The mountings of a casket found at Icklingham are replaced on a modern box in the British Museum; and those of another, including its contents, found at Rushmore, are figured by General Pitt-Rivers.¹¹ The keyhole cover of the latter was in the form of a hinged boss ornamented with a bust in a Phrygian cap. Several of the Colchester caskets appear to have had mirrors fitted within their lids, as mentioned in a paragraph above.

Of articles of pure adornment, those which are comprised under the general term armillae are conspicuous in our Romano-British collections. The term is convenient, for it is often impossible to decide whether a given specimen is a bracelet, armlet, or anklet, as they are not distinguishable by peculiarities of form or pattern. Relative size helps us, but not much. If one is too large for a bracelet, it may be reasonably concluded that it encircled the arm; but a child's armlet may be as small as the mother's bracelet. The women of the era certainly wore them as bracelets and less frequently as armlets, for they have been found in graves occupying these positions on the skeleton; but whether they were used as anklets is not so clear. If, then, these articles are specified below as bracelets and armlets, the reader must keep in mind these limitations and uncertainties.

As a class, the Romano-British armillae are not conspicuous for variety of form, construction or ornamentation. They are resolvable into few types, and the decoration, when present is of a simple sort, never including enamel, and this is remarkable because they could easily have been designed to present an admirable field for its display. They are rarely in other material than bronze and jet, and if a precious metal is used in their construction, it is used sparingly. Those of metal are light and slender, and many would now be designated bangles. The massive gold armillae with Late-Celtic ornamentation, which are occasionally figured as Roman, are almost certainly the productions of contemporary Scottish or Irish metal-workers, or are pre-Roman.

Fig. 76. Bracelets and Armlets, Finger-Rings, Ear-rings, and Beads. (A to G, 2/3; the rest, 1/1)

The bracelets most frequently found are of two bronze wires twisted into a cable, one wire being manipulated into a small hook at one end, and into an eye at the other, the free wire being coiled to form a collar, or instead of this, the ends may be confined by tubular collars, as in Fig. 76, A, a Lydney example. The component wires were often attenuated towards their ends so as to produce a pleasing swell in the cable. Sometimes these bracelets were not made to open, as a small example in the Guildhall Museum composed of a bronze and an iron wire twisted and looped together at the ends to form a central ornament.

Our next example, B, from Rushmore, is of less frequent occurrence. It is made of a single bronze wire, expanded about the middle, and sliding on itself, each end being coiled round the wire at some distance from the opposite end. Its large size is suggestive that it was an armlet, for which it would be well adapted, its elasticity exerting a pressure on the arm which would keep it in place. The unusual bronze armlet (it is too large for a bracelet), C, was found with a skeleton at Deepdale Cave near Buxton. The hoop is square in section, and each attenuated portion is bent into a row of loops, the two rows being parallel and held in position by the surplus wire being wound round the contiguous parts of the hoop. Similar armillae have been found elsewhere, and a finger-ring of similar manipulation at Silchester. The slender bracelet, E, from Caerwent, is transversely ribbed and has three bead-like swellings and a hook-and-eye clasp. Bracelets of this type have been sparingly found, and are apparently derived from a prototype in which several beads were threaded on a wire, the intervals being wound with finer wire.

Our next example, F, from Richborough, stands for a large class. It is a hoop made of a narrow band of bronze. The exterior is ornamented by the angles being filed out at intervals in such a manner as to leave a simple key-pattern in relief which is ornamented with engraved lines. A variety of patterns was produced by these means and sometimes by punching in addition, of which we give five, G. The Richborough example has an overlapping joint so that it could be sprung open in passing over the hand. A similar bracelet found at Aldborough had its ends bent back to form two loops apparently to allow of its being tied by a cord. Others again and perhaps the larger number have their ends riveted together. D is a variant from Lydney in which the hoop is plain and ends in an ornamented hook-and-eye clasp. Another type of bangle was apparently cut out of a thin plate of bronze and ornamented with a scalloped edge.

Jet bangles are not uncommon, and there are many in the York Museum; where also are bracelets made of several pieces of carved jet and of several pieces of bone united by lead and copper bindings; also two glass bangles, a small green one with blue and white lines, and a larger, dark red, with white and purple stripes. Penannular armillae, although frequently found with pre-Roman Britain remains, are rare. There are several in the Guildhall Museum, one of silver ending in grooved knobs, and another of tinned bronze with the ends expanded into ornaments resembling serpents' heads. Bronze 'arm-purses'¹² have been found at Thorngrapton, Birdoswald, and elsewhere in the north. In these, a portion of the hoop is expanded into a boat-shaped cavity, with a hinged lid on the inner side closing with a spring snap. The first example contained coins ranging from Claudius to Hadrian.

There are several bracelets of beads, mostly of jet and glass, from burials at Colchester in the Joslin Collection. One bracelet is of sixteen blue ribbed beads with two coins of Nero as pendants. There is a small chain bracelet from Colchester in the York Museum, and it is not unlikely that some of the pieces of fine bronze chain seen in most collections are portions of similar bracelets.

It is noteworthy that the armillae found on our Roman sites show little, if any, Roman influence; on the contrary, they seem as a class to have been derived from indigenous prototypes of pre-Roman times. The cabled bracelets so closely resemble the British neck-torcs that one can hardly hesitate to trace them to that source; as also the wire and ribbon examples to similar pre-Roman forms. The ancient Britons also had jet armillae, and it has already been noticed that their penannular form survived into Roman times, while jet, amber, and glass beads are of common occurrence in their graves. Another noteworthy point is that while not few of the British armillae were of gold and highly decorated, the precious metals are singularly wanting in the Romano-British. This is remarkable when we consider that the 3rd and 4th centuries were characterized by a love of display and personal adornment, and it seems to indicate that the wearing of armillae was not fashionable with the wealthy, but was mainly confined to the poorer classes, during these centuries.

Finger-rings are of great antiquity, and were at first objects of utility rather than of pure adornment, being seals adapted to be carried on the finger or thumb. Among the Romans the earliest rings appear to have been of iron or stone; but gold rings were early conferred as a military distinction, and the privilege of wearing them was afterwards extended to ambassadors, to senators and chief magistrates, and then to knights. Tiberius next extended the jus annuli aurei to all who had a certain property qualification, and his successors to all whom they willed. Severus conceded the right to all Roman soldiers, after which the gold ring gradually ceased to carry with it any distinction. The devices engraved upon the signet-rings were varied, and included mythological subjects, portraits, and allusions to the family history of the wearers, thus in a sense answering to our crests. Originally the men wore only a signet-ring and the wedded women a marriage ring; but under the later emperors, rings, often of a costly sort, were worn in great profusion.

Finger-rings are frequently found on our Roman sites, and they appear to have been worn by all classes. They are not confined to the sites of towns and country mansions: even the small and remote Romano-British village at Rushmore yielded twenty to the spade of General Pitt-Rivers.¹³ Bronze is their usual material, then follow in descending order, jet, silver, iron, gold, amber, and glass. Such is their diversity of form and ornamentation that it is scarcely possible to classify them. Many are hoops which, if ornamented, have their ornamentation diffused all round; many have their ornamentation concentrated to one spot, the rest of the circuit being a hoop, and to these belong the signet-rings and the forms derived therefrom; while the residue consists of rings of intermediate character or of aberrant forms.

The simplest Roman 'hoops' are of bronze wire bent into a circular shape with the ends meeting, but more often overlapping, and more often still the wire is made into a double coil — as Fig. 76, H, or even a coil of three turns. Such rings were probably home-made; but in skilful hands the ends of the penannular ring were ornamented, or, if they overlapped, each was bent back and assumed the shape of a serpent's head, while the double coil sometimes took the form of a serpent twined round the wearer's finger. Of endless hoops, two found at Rushmore are simple examples, one being of bronze wire with the ends brazed together, and the other of white metal square in section. Another Rushmore example, I, which is not uncommon, is of base silver, circular within and octagonal without, and it provides us with a starting-point for continuous ornamentation. The periphery of a Silchester ring, J, is cut into a series of concavities, that of another, K, is punched with a fine herring-bone pattern, while that of a third is diagonally grooved.

We now turn to the more interesting class of signet-rings and rings of kindred form suggested by or imitating them. In the bronze ring, L, found in London, the hoop swells into the bezel, which contains a paste intaglio of a bird. There is a similar ring in silver with a jasper intaglio of Mars in the Caerleon Museum, and two of bronze in the Guildhall Museum; in fact, these rings are not uncommon, and probably represent an old form which died out in the 2nd century. Iron rings are occasionally found, and they all appear to be of this form. There are two in the Guildhall Museum, the one with a jasper intaglio of a man holding a patera and cornucopiae, and the other engraved with a galley in some other stone. A Wroxeter example has its stone engraved with a fawn springing out of a nautilus,¹⁴ and a Melandra Castle one has a ram.¹⁵ Iron was not used for these rings on account of its cheapness. From Roman writers we know that many had a preference for iron signet-rings long after those of more costly metals and alloys had become general. In the imitation or bastard signet-rings of the form, the bezel lacks an intaglio, or instead of an intaglio there is an inset of unengraved glass or stone, or of enamel.

In the more elaborate rings of this type, the shoulders of the bezel are ornamented and the setting of the stone takes the form of a rim or border often also ornamented, hoop, shoulders and setting now ceasing to flow into one another and appearing as separate ornamental entities. Usually the setting is highly raised, in order that the impression from the intaglio should not be disfigured by the impress of the shoulders. The highly ornamented rings are, as a rule, of silver and gold, but their technique varies considerably, many of them being of decadent execution and reflecting a taste for display. Instead of intaglios, their settings sometimes contain cameos, which again are often of inferior workmanship. Two examples of these ornate rings are given,— one of silver, N, from Great Chesters, containing a stone with a bevelled edge,¹⁶ and the other of gold, O, from Sully near Cardiff, with an onyx cameo of Medusa's head.¹⁷ The last was associated with three other gold rings of similar character and a large number of coins which proved that the hoard was buried in the first quarter of the 4th century, and this confirms the attribution of this class of rings to the 3rd and 4th centuries.

We have already described some examples of engraved gems, and as it is unlikely that this delicate art was practised in Britain it is unnecessary to give it more than the briefest notice. This art was at first confined to the production of seals, but under the Greeks it attained such perfection and was so appreciated that precious stones were not only carved in intaglio but also in relief (cameos), as pure works of art for the connoisseur and collector. The Romans equally esteemed them, and they were produced in large numbers by Greek artists settled in Italy, but from the first century the glyptic art gradually declined. The examples found in this country, whether in rings or loose, are in both precious stones and paste, and are mostly intaglios. Few belong to the best Roman period, the larger number being mediocre and some even barbaric, the work probably of provincials. Some of the loose gems may have fallen from rings, but many, and especially the cameos, are too large to have ever adorned these articles. It is probable that these were appreciated for their own sake, as also for the various virtues they were supposed to possess — these virtues depending in part upon their subjects. The number found in this country, however, is not great. There are thirty-three intaglios in the Pump Room at Bath, which were obtained from the main outfall drain of the baths in 1895, where they were apparently all dropped together in the 2nd century, perhaps accidentally by a jeweller. Eight different stones are represented in the series, nearly half being sardonyx, and the rest onyx, sard, agate, chalcedony, amethyst, heliotrope, and plasma. Nearly one-third of the subjects are taken from the animal world, and include a griffon and a crane. Next in point of number are gods and goddesses and other mythological personages, the residue being charioteers, athletes, a horseman, a shepherd, a youth making an oblation, two heads, and a trophy. There are some good examples of engraved gems in the York and Shrewsbury Museums.

Few ear-rings of the era remain, and as these are mostly of gold, it may be that being small and delicate objects, those of inferior metals and alloys have perished beyond recognition. The prevailing form is a small disk or a precious stone in a setting, with a wire hook attached to the back. Fig. 76, R, is a Silchester example, with a circular gold plate of delicate pierced work, and S, in the Chesters Museum, is a rectangular blue stone in a ribbed setting. There are several set with stones in the York Museum, and another from Silchester has the form of a serpent holding an emerald in its mouth. One in the Pump Room at Bath has a pear-shaped carbuncle, and linked to its setting are two gold wires, which probably terminated in small ornaments. Much more elaborate was a gold ear-pendant found at Housesteads. The base of the hook was expanded in the form of a small leaf, and from it depended successively two acanthi and two S-spirals, all linked together and having a total length of 2 3/4 ins.¹⁸ Two found at Gellygaer are of a different type, each being a fusiform piece of metal ending in fine points and bent into the form of a penannular ring. The larger is of bronze and the smaller, T, of base silver. The points being pressed together into the perforation of the ear-lobe, the ear-ring was necessarily worn permanently. There are several ancient gold ear-rings of this type in the collection of the Royal Irish Academy, and similar are still worn in northern Africa. Two in the Colchester Museum, found with the remains of a child in a lead coffin in the vicinity, are of gold wire bent into the form of the bracelet shown in Fig. 76, B. In the Guildhall collection is another of pewter in the form of a simple ring with the ends twisted together. It is probable that some of the small penannular rings, which have been described from time to time as children's finger-rings, were worn as ear-rings.

Glass beads, of two prevailing shapes, cylindrical and globular, are of common occurrence on Roman sites. The ordinary cylindrical beads appear to have been made from round or polygonal tubular canes of blue or green glass of about the thickness of a thin tobacco-pipe stem, broken into the requisite lengths, and rounded at the ends by partial fusion. In a larger and elaborate variety, the cane was clothed with several layers of different colours, and the shoulders of the bead were bevelled off with a series of facets, thus exposing the edges of the layers as a succession of zigzag bands, as indicated in Fig. 76, . The globular beads are usually somewhat flattened, varying from 1/4 to 3/4 in. in diameter. The larger sizes are generally decorated with superficial zigzags, meanders, stripes, or 'eyes' of white or yellow, the body usually being dark blue. U is a Gellygaer example. Other shapes are also met with, a frequent one being an oblong or oval plate of coloured glass perforated longitudinally. Many of the glass beads are hardly distinguishable from those found with Anglo-Saxon remains; but a characteristic Roman variety, V, is somewhat melon-shaped and ribbed, and made of a pale blue vitreous frit. Of beads of other materials, those of jet are not uncommon. They are of various shapes, and are sometimes carved with incised ornamentation. Amber, coral, ivory, and bone beads are sparingly found, and those of stone are rare — there are an alabaster bead with projecting spines and another of chalcedony in the Guildhall Museum.

Now and again sets of beads of necklaces and bracelets have been recovered, mostly from graves, and several examples of these may be seen in the York and Colchester Museums. In the former museum are two necklaces still intact, the one of yellow and green glass beads and the other of blue glass and coral beads, strung on fine silver wires. A necklace in the Guildhall consists of twenty-four bone and ivory beads with a perforated piece of tusk for a pendant. Many small objects have been found, mostly of jet and bronze and perforated for suspension, which may have been pendants of necklaces, as for instance a jet bear and Medusa's head at York, and a bronze drop ornamented with a violet stone at Colchester. Most of these were probably regarded as amulets. Coins were sometimes used as pendants, and probably also the larger and more enriched beads.

The Author's Notes:

1. Roman Forts on Bar Hill, p101.

2. Illustrations Rom. Lond. p132.

3. Cumb. and West. Archaeo. Soc. xv, p183.

4. Excavations, ii, p190; also iii, p102.

5. The pyramidal ornament with its terminal boss in Fig. 72, A, a fibula found in Deepdale Cave, Buxton, and the projection at the back of C with a disc of red enamel held by a small pin above it, are legacies of the ascending tail and its rivet, but are purely ornamental, as the pins are hinged.

6. Illustrations of Roman London, plate xxxiii, 8, 11, 10.

7. Archaeologia, xxvii, p359.

8. Pitt-Rivers, Excavations, iii, p132.

9. Brit. Arch. Assoc. iii, p208.

10. Uriconium, p278.

11. Excavations, i, p61.

12. Arch. Jour. viii, p88; xvi, p84.

13. Excavations, i, p51.

14. Uriconium, p318.

15. Melandra Castle, p113.

16. Archaeo. Aeliana, xxiv, p42.

17. Numismatic Chron. xx, p64.

18. Bruce, Roman Wall, p200.