AMULETS CHIEFLYI N THE BRITISH MUSEUM A SUPPLEMENTARY ARTICLE (PLATES 96-100) L'archeologie, si elle est privee du secours de la philologie, devient une science conjecturale, dont les conclusions n'atteignent que le degre de vraisem- blance que peut leur preter l'ingeniosite et 1'eloquence de leurs auteurs.-FRANZ CUMONT. ;>;HEN one ventures to ask the attention of scholars to a long article on WYWYmagicaaml ulets shortly after publishing a book on the subject, a word of explanation seems to be in order. Much of that book (Studies in Magical Amulets)' was based upon observations made when I had the opportunity to examine the un- rivaled collection of magical gems in the British Museum, but only a few illustrations of them could be offered in the finished work. This was because, unfortunately as it turned out, I deferred asking for the necessary casts until I was ready to proceed with the actual writing of the Studies. By that time the war was in progress, and the treasures of the Museum were not available for the use of students until sometime after it ended. In the meantime I had been obliged to proceed with the printing of the material at hand. Now, through the courtesy of the Museum, I have obtained a number of excel- lent casts. They were made from some of the most interesting amulets in the col- lection, and would be worth publishing even without regard to their affiliations. Several of them, however, afford the means of supplementingo r correcting the account of them, and of certain kindred types, which I gave from hurried notes made in 1935 and 1937; others of considerable importance could not be mentioned in my book because of the incompletenesso f my notes. Meanwhile several interesting photographs and casts have been generously placed at my disposal by the curators of other foreign collections, and by private owners.2 'Studies in Magical Amulets, chiefly Graeco-Egyptian. Ann Arbor, University of Michigan Press: London, Geoffrey Cumberlege, Oxford University Press. 1950. In this article the abbre- viation SMA is generally used in referring to the book. I take advantage of the supplementary purpose of this paper to excuse my use of various other abbreviations and short titles. They are mostly familiar, and all can be found in the Partial Bibliography of SMA, pp. xix-xxiv. In the course of preparing this study I have incurred many obligations, which it is a 2 pleasure to acknowledge here. In the first place, for permission to publish the majority of these Hesperia XX, 4 American School of Classical Studies at Athens is collaborating with JSTOR to digitize, preserve, and extend access to Hesperia ® www.jstor.org 302 CAMPBELL BONNER In the latter part of this article all these objects are catalogued and arranged in much the same manner as that adopted for the list on pp. 253-323 of Studies in Magical Amulets, and illustrations will be found on the accompanying plates. Where it is possible to contribute anything towards the interpretation of the amulets, or to indicate their relations to previously published specimens, comments, usually brief, are added to the descriptions in the Catalogue. I trust that no apology is needed for the numerous references to passages in Studies in Magical Amulets where these or similar pieces have been discussed. They will make it unnecessary to repeat arguments used and authorities cited in the larger work. After some hesitation I have decided to include in the Catalogue descriptions of several gems that were brought to my attention by dealers who sent me impressions of them at various times during the last twenty years. Their present location is unknown to me, and I have no means of tracing the owners and requesting permission to publish, as would ordinarily be not only proper but obligatory. I hope that the interest of the objects to the expert will excuse the breach of the usual custom. Here I should mention the fact that several pieces that are published for con- venience along with the magical stones do not show any signs of magical purpose; they could be regarded simply as tokens of the wearer's devotion to the divinities represented upon them. The distinction between stones worn in the hope of securing divine protection for oneself, and others which invoke or seek to control demonic powers, is not always clear. The technique of the religious amulets, as they may be called, is much the same as that of the undoubtedly magical pieces, and they belong to the same period. A magical word or even a magical character would be enough to transfer a stone from the one category to the other (see SMA, pp. 5-7, 43, 45). A few words of explanation about the descriptions and the plates will not be out of place. Amulets were not meant to be used as seals, and, with rare exceptions, their designs were intended to be viewed directly, not by means of impressions. For this reason the illustrations have been made from casts (positive) when they could be obtained, not from impressions; and when only impressions were available, the photo- objects, I am deeply indebted to the Director and Trustees of the British Museum, to Mr. Alec B. Tonnochy, Keeper of the British and Mediaeval Antiquities, and to Mr. Bernard Ashmole, Keeper of the Greek and Roman Antiquities, who has been most helpful. A word of appreciation is also due to the technical assistants who made the casts. Mr. and Mrs. Frederick Zimmermann of New York graciously allowed me to examine and publish seven pieces in their possession (Nos. 74-80). For similar favors I am grateful to M. Henri Seyrig (72-73), M. 0. Gueraud (16), Prof. G. H. Chase (58), M. J. Babelon (56, 60), Prof. Fritz Eichler (63), Mr. C. D. Bicknell, Lewis Curator (65), M. G. Fabre (66), M. J. Charbonneaux (67), Father R. Mouterde (81). Prof. Charles Seltman, on this as other occasions, has given valuable help. Prof. H. C. Youtie contributed a reading in No. 51. The casts of the stones owned by Mr. and Mrs. Zimmermann are the work of Dr. Louise Shier. The photographs were made by the studio connected with the Institute of Engineering Research, University of Michigan, and by the recently established Photographic Services of the University. AMULETS CHIEFLY IN THE BRITISH MUSEUM 303 graphic films were reversed in printing in order to show designs and inscriptions in their proper relation to the spectator. In some instances, when a modern imitator has overlooked this characteristic of ancient amulets, and has cut a stone as for a seal, I have reproduced the appearance of the impression. rhe words right and left are used from the spectator's point of view except in mentioning the physical parts of a figure in the design, as the hand or the shoulder of a divinity. " Upright oval " means that the axis of the obverse design is in the longer dimension, " transverse oval " that it is in the shorter dimension. No account is taken of the reverse, where opposite conditions are common. Measurements are given in millimeters. An unexpected result of the study bestowed upon the material from the British Museum is set forth in the pages that immediately follow this paragraph. They deal with Nos. 58-71 in the Catalogue, and would have been introduced just before that group were it not inconvenient to divide the Catalogue into two parts. Attentive observation of these pieces showed that their designs were not simply independent variations upon common subjects-like, for example, the numerous Harpocrates and Chnoubis amulets-but that there were among them either ancient replicas or else, certainly in some instances, modern copies of ancient originals. Others, again, are fabrications merely suggested by an ancient pattern, not copied from it, or are founded upon nothing better than a forger's notion of what a " Gnostic " amulet should look like. I trust that my conclusions about these objects are stated with due caution, because, to say nothing of the possibility that others may hold different opinions, it is not easy to determine the exact relations of two or more similar pieces to one another without bringing them all together for minute inspection, and that is rarely practicable. I should particularly deprecate any reflections upon the scholarship of outstanding authorities who have accepted some of these stones as genuine. Errors of judgment are quite excusable in a division of archaeology to which comparatively little expert talent has been devoted. ANCIENT REPLICASA ND MODERNI MITATIONS Many years ago Furtwangler called attention to the fact that certain glyptic designs have been carried out in two or more replicas, the work of the same artist (Autike Gemmen, III, pp. 92-93, 443-444). In some instances one piece cannot be distinguished from the other by any detail of the gem-cutter's work, in others there are insignificant variations. A good example of such doublets falls within the period in which most of our magical gems were made, and may serve to introduce this division of the present study. A. The Museum of Fine Arts in Boston possesses a sard with a representation of the story of Jonah (No. 03. 1008). This gem was described and illustrated in an article in the Harvard Theological Review, 41 (1948), pp. 32-33, with fig. 1; a brief 304 CAMPBELL BONNER description and illustration appear also in SMA, p. 312, and P1. 19, 347. There is therefore no need to discuss its details here. The article just mentioned came to the attention of Mr. Bernard Ashmole, Keeper of the Greek and Roman Antiquities in the British Museum, and led him to take note of a strange coincidence. In a letter of June 14, 1949, he informed me that a visitor had recently brought into his department an exact replica of the Boston stone. Mr. Ashmole did not record the material of this replica, but took a good sealingwax impression of it, which he very kindly turned over to me. Comparison of this impression with the Boston sard shows that the designs are indeed indentical. Mr. Ashmole had no doubt of the genuineness of the stone shown to him, and the antiquity of the Boston gem has never been called in question. The two pieces are shown together (Nos. 58-59), and a brief description of the design follows in the Catalogue which forms a part of this article. B. The situation is very different in the next group of objects (Nos. 60-62), of which the first is a large red jasper in the Cabinet des Medailles (No. 2221 in Chabouillet'sC atalogue). There is nothing suspicious about the design, a man dressed in kilted tunic of the military type, and shod with boots, standing with a serpent in each hand. Egyptian gods are frequently shown holding serpents, and in the imperial period are sometimes clothed as Roman warriors. The trophy fixed upon the head is unusual; yet Egyptian deities are often marked by symbols so attached-e. g., the scorpion of Serqet and the fish of Hatmehit-and there is no reason why the trophy, taken over from the Greeks and Romans as a symbol of victory, should not be so used (cf. SMA, pp. 244-245, P1. 21, 374-376). I have not seen the reverse of the stone, which contains in four lines arrangements of the seven vowels. There is little doubt that it is this stone, then in the possession of Thomas Le Cointe, which is illustrated in Chiflet's AbraxcasP roteus, pl. 23, No. 94.3 In a few minute details Chiflet's engraving is untrue to the original. The snakes held by the man are straighter, and the trophy on his head, which, in the original, is very slightly inclined from the vertical, is here tilted farther and looks as if it might topple from his head. There are also inaccuracies in the inscription, some of which, such as the confusion of alpha, delta, and lambda, prove nothing, but others are important. On the side toward which the man is looking, the word nearest the margin begins with a letter (damaged by a slight grinding down of the margin) which is almost certainly p; the engraver has omitted it entirely. Similarly, on the other side of the stone, he has omitted the v which begins the middle line and the 8 which begins the outside one. Another minute detail is of some importance in connection with other gem inscriptions as well as in this instance. The original engraver used clearly marked but not exaggerated serifs at the tops and bottoms of several letters. Chiflet's 3This work follows the essay of J. Macarius (L'Heureux), Abraxas seu Apistopistus, in a volume published at Antwerp in 1657. AMULETS CHIEFLY IN THE BRITISH MUSEUM 305 engraver has so overdone these serifs that the letters a, 8, X, are topped by a con- spicuously long horizontal stroke for which there seems to be no precedent in ancient epigraphy. Observation of these details enables one to state with some confidence that two other stones bearing the same design as the Paris stone are modern copies based upon the engraving in Chiflet or the reproduction of it in Montfaucon (L'Antiquite expliquee, II, 2, pl. 160, 4). The first of these is B. M. 56026 (No. 61). Here the stone shows the same inaccuracies as those noted in Chiflet's engraving; but the maker of 56026 has added errors of his own. He has omitted the first four letters of the word aXco8oKcapv(vmo iddle line at the left), perhaps because they resemble the last four of the preceding line; and he has written a reversed N in this line and one in the following. He has also followed Chiflet's engraver in placing the long horizontal stroke over the triangular letters, and in one or two places it is as long as the base of the letter. Other minutiae, not worth discussing here, but significant to a close observer, strengthen the conviction that the work is not ancient. This same design is used for the reverse of another stone in the British Museum, 56360 (No. 62), and the marks of Chiflet's engraver are again present, some of them exaggerated. The snakes are even straighter, the angle of the trophy to the vertical is still greater, the horizontal stroke across the tops of the triangular letters is still longer in some instances. Several glaring errors occur in the inscription on the right hand side of the stone. The design is again a modern fabrication based on the old engraving. The obverse design also is somehow related to an engraving in Chiflet (pl. 19, No. 78) or to its original; but the British Museum stone cannot be identified with that original because Chiflet's gem has a different reverse. T'here is also a difference in the obverse design, to which attention will be called later, but with that one excep- tion, the slight differences that are perceptible may be due to the engraver. The central figure of this obverse design is bearded, has his arms folded on his breast, and wears a crown with triangular points, which is of a mediaeval rather than an ancient type. The lower part of the body is so closely swathed as to give the appear- ance of a herm. From his elbows two curved lines descend to the heads of the two outermost of four nude figures which form a group directly under the feet of the royal person. They face a central axis, have their arms crossed upon their breasts, and seem to be dancing on a segment of a celestial sphere, indicated by two concentric arcs connected by slanting lines, and with three stars in the spaces between. The other details, including a meaningless inscription, contribute nothing of importance. In Chiflet's illustration, wings are attached to the backs of the two outermost dancers; in the B. M. stone they are absent and in the illustration they were probably added by the draftsman who supplied the copy for the engraver. So far as I know, the design as a whole is quite without parallel in ancient glyptic art, and it seems to me 306 CAMPBELL BONNER to be a modern fabrication. It probably represents some Renaissance scholar's con- ception of a Gnostic " universal father" with a group of cosmic spirits dancing on the celestial sphere. C. Another cut of Chiflet's (pl. 17, No. 19) plays a part in the treatment of the next two objects. The first of these is a chalcedony in the Kunsthistorisches Museum at Vienna (No. 63). In Chiflet's time it was in the cabinet of the Archduke Leopold Wilhelm. The stone is an oval lentoid, with sharp edge, 22 x 19 x 6.4. On the obverse, at the top, is an ouroboros enclosing seven characters, of which five, including two forms of epsilon ( E and C), are Greek letters. At each side are two characters of the " ring sign " class; but the eight points of the crossing diameters end in short strokes at right angles to the lines, not, as usual, in small circles. Below is a long inscription, of which the first nine lines contain a number of well-known voces rncagicaes, everal of them in corrupt forms, then the words oVXceErEM ataLv. The Latin name Maianus is not common, but is adequately attested. As for the use of the dative, it is uncertain whether we should-r egard it as dativus commodi, as in Thuc. 7, 53, or as a mere blunder for the accusative. Among the magical words we recognize more or less corrupt versions of Iao Sabaoth Adonai semeseilam, abrasax, ackracm- machamari, sesengenbarpharanges, arouanta, Michael, arorachthi, and in lines 6-7 what is probably a careless copy of &rt[hmncchiniamb&an w, ord that I have seen else- where only in connection with scorpion amulets (SMA, pp. 77, 200, 273). The reverse is covered with an entirely meaningless inscription, the elements of which do not even make a recognizable magical word, with the single exception of lao; and the greater part of it consists of mere combinatio'nso f the vowels in threes. Thus the potency of the inscription consisted solely in the fact that it was made up of letters in groups of three and was unintelligible (see SMA, pp. 193-194). After the first line the writing is divided into three columns, which, from line 3 on, consist of three letters each, with vowels greatly preponderating. Chiflet's draftsman made his drawing almost three times the size of the original, and it is inaccurate in several details. The ring signs at the top of the obverse are reduced to mere asterisks, and the head of the ouroboros, indistinct in the original, is brought out clearly and given a kind of forward-cocking crest, which I cannot detect on the photograph of the original. There are also several errors of reading, especially on the reverse, where the engraver has further erred in barely indicating the division of the inscription into three columns, which is so conspicuous on the original. There is reason to think that at least three stones were cut with Chiflet's engrav- ing as pattern; King, at any rate (Gnostics, pp. 289-290), had seen three with the same inscription except for the apparently inevitable errors. One of them is probably our No. 64, a sard in the British Museum (56276) of about the same size as the Vienna chalcedony. If it is the same as a stone mentioned by King (p. 290), it came from the Towneley collection, but the Museum now has no record of its provenance. AMULETS CHIEFLY IN THE BRITISH MUSEUM 307 On the obverse the maker has followed Chiflet's draftsman in reducing the four ring signs to asterisks and in making the head of the ouroboros clear. He has also followed Chiflet's errors in the inscription (e. g. CCC6a t the end of line 4, Xe line 9), and introduced several more of his own. He has departed from the proper division of the lines, and at the end he has omitted the prayer bvXa'eTE MatavW4 for lack of room. On the reverse matters are even worse. Chiflet's errors are copied, (e. g. MACO for HAG:i)n line 4, 6M H for 6G) H line 12), others of his own are numerous, and the whole epigraphic style breaks down in the latter half of the inscription, where we find impossible forms like a Latin L, a thin broken-backede psilon, and other such mon- strosities. If anything further were needed to establish the spurious character of the stone it would be at hand in the circumstance that this forger took the trouble to cut the long inscriptions retrograde, ignorant of the fact that both designs and inscrip- tions on genuine magical gems are almost always cut to be read by direct view, not from an impression. Retrograde inscriptions of a single word sometimes occur on ancient amulets; but I know of no long inscriptions so treated. The tests that prove the British Museum stone to be derived from Chiflet's engraving may be applied with a like result to an amulet published in the Catalogue of the Wyndham Cook Collection (p. 58, No. 264). The maker used a stone of a different shape, and perhaps for that reason omitted the ouroboros and asterisks at the top. The inscription is better executed than that of No. 64, and in that respect follows the Chiflet cut more faithfully, though not without errors. Since the above-mentioned publication is adequate, there is no need to describe the piece or to comment upon its details. It is scarcely necessary to emphasize the importance of the part that Chiflet's engravings played when they came into the hands of forgers of the late seventeenth and eighteenth centuries. The making of spurious " Gnostic " gems had begun before, as Chiflet's own plates show, for it is impossible to believe that some of the objects that he illustrates are ancient. But the convenience of his cuts for the imitator's purpose is obvious. We have seen it in this group and the preceding one (B), and shall find it at work again in Group E. To these descendants of Chiflet's engravings we should add two more which I have discussed elsewhere (SMA, p. 147, p. 290, No. 221; p. 281, No. 173). D. The next group (Nos. 65-68) is rather puzzling. If all four pieces could be set side by side on a table, it might be possible to determine their relations to one another definitely; but when one uses only casts, impressions and photographs, some points must remain in doubt. Yet it is certain that all the members of the group are directly or indirectly the product of one maker. The design common to all of them is best represented by a stone in the Lewis Collection at Corpus Christi College, Cam- bridge (No. 65. It is No. C 17 in J. H. Middleton's Catalogue). Middleton says that 308 CAMPBELL BONNER Mr. Lewis bought the stone from M. Feuardent, who obtained it in Bombay. It is a red jasper scaraboid, very convex on one side, which we treat as the obverse, flat on the other. The obverse design is the cock-headed anguipede, noteworthy in this instance for the slender, elongated proportions of the human trunk. In the field and round the margin are many letters, mostly vowels, making no sense. Other details will be noted in the Catalogue. The center of the reverse is occupied by the lion-headed Chnoubis serpent, radiate as usual. His body makes a double coil at the middle of the field and then descends in an almost straight line. Round him are the triads of animals which are commonly placed round Harpocrates, as if adoring the young sun god, scarabaei above, goats and crocodiles at right, birds and snakes at left. At either side of Chnoubis are two meaningless letters. There are unusual features about the designs of both sides. Taken singly they carry little weight, but taken together and in connection with the existence of approxi- mate replicas, they arouse some suspicion. In the first place, the obverse is cut as for a seal, and so the anguipede carries his shield on his right arm. This is rare on magical amulets. Yet the reverse design is cut for direct view. This is proved by the fact that the goats are on the right side, as in all genuine amulets that show the animal triads (See SMA, P1. 10, 203-208; 210 is exceptional because there for artistic reasons the maker has broken up the triads. See also Southesk Catalogue, N 24, remembering that all Lord Southesk's cuts are made from impressions). The substitution of Chnoubis for Harpocrates is not inappropriate, since both are solar figures (SMA, p. 142), and yet I believe it to be without parallel except in this group. There are also a few details of the design that are faulty. The legs of the crocodiles, instead of bending naturally, are set like short straight pegs under the middle of the body. The birds, which should be hawks, seem to be pigeons, though they might even be taken for ducks; and the snakes have quite unnatural mouths 4 and odd-looking crests. This last point applies also to the snakes -that serve as the legs of the cock-headedg od on the obverse. There is a close replica of the Lewis stone, another red jasper, in the Cabinet des Medailles of the Bibliotheque Nationale (2198 bis, our No. 66; Babelon, Guide, pp. 70-71; not in Chabouillet's catalogue of 1858). It gives the anguipede the same dis- proportionately long body that was noted on the Lewis stone. The casts of this latter seem to be slightly smaller than the impressions (I have no positive casts) of the Paris specimen-;b ut this may be an unavoidable consequence of the process used in making the reproductions. The only difference that can be readily perceived is in the M behind the anguipede's whip arm; on the Paris stone it is set a little higher with 4They have also been taken for peacocks (De Ridder, Cat. sotmmaire des bijoux antiques, No. 1616). AMULETS CHIEFLY IN THE BRITISH MUSEUM 309 reference to the elbow, and the letter itself is a little larger. The Paris jasper has an A clearly cut on the shield, where the Lewis stone has what looks like V; but it is really part of an A, the left hand stroke of which was too lightly incised or else has been worn away. On the Paris stone the lines of letters just below the arms of the god have been erased, or else so abraded by wear that only faint traces are visible. Before leaving these two objects attention may be called to the fact that red jasper is rarely used for Chnoubis amulets, and is not commonly used for the cock-headed demon, though examples occur. For the Chnoubis design chalcedony,p lasma, and black jasper are often employed; for the anguipede dark green jasper and haematite predominate. The two remaining members of this group are made of bronze, one in the Louvre (Bijoux 1616, our No. 67), the other in the British Museum (56550, our No. 68). I have suggested elsewhere (AJA, 53 [1949], p. 271) that a highly convex form, which has no advantages for engraving on bronze, may indicate that such objects as these are casts made from semi-precious stones, the beauty of which is enhanced by cutting en ccabochon.I t is probable that these two bronzes are casts from the same original, since in all essentials they are alike. It is true that the London bronze does not show the letters round the margin clearly, but that seems to be because its edge was ground off slightly, with resulting damage to the inscription. But whether they were cast from it or simply cut in imitation of it, that original differed in several respects from the original of the jaspers of the Lewis Collection and the Cabinet des Medailles. The chief differences are as follows. The proportions of the anguipede on the bronzes are less elongated than on the other two amulets; there is a difference of 3 mm. in the length from shoulder to the bottom of the kilt. Further, in the bronzes the cock's beak is tilted upward, making a broader angle with his neck. Consequently the comb is inclined from the vertical, while in the two jaspers it is upright; and on the bronzes its notches are more clearly indicated. These are minute points, yet decisive. To sum up: the two jaspers of this group are closely similar and evidently derived from a common source, and the two bronzes are probably casts from an original which was not the same as that of the jaspers. Yet in all four the differences are so slight and the agreement in peculiar characteristics so striking that all must be ultimately derived from one designer; and he has departed so far from the usual type, especially in the reverse, as to cast some doubt upon the genuineness of the whole group. If the two originals were ancient, which I should not venture to deny absolutely, their designer has in any event misunderstood and wrongly rendered several details of a well-known traditional type. E. This last group (Nos. 69-71) comprises three unrelated pieces all presenting some characteristics that betray the hand of an imitator, or at least lead one to suspect it. First, a sard in the British Museum (56069, our 69), with an attractive design of Harpocrates standing in the cup of a flower. He holds a flail over his right shoulder, 310 CAMPBELL BONNER and raises his left hand towards his face; his head is encircled by twelve rays. Some other details will be mentioned in connection with the appraisal-of the work. On the reverse is an ouroboros enclosing three magical words, covuapra ac3Xava0avaX)3a the last two of which are very common. Outside the ouroboros runs aKpa,uqaXajapE, another inscription with no recognizable elements except cJaKacoO(e rror for ca/3aw6) and a8o, which is probably part of a8wvat. The parts of this outer inscription are separated by vacant spaces in a manner which will require comment later. The maker of the obverse design chose a familiar type, but varied it in a manner which lays the work open to suspicion for several reasons. 1. When a lotus is part of the design, Harpocrates usually sits on the flower (or capsule) with his knees drawn up, or with his legs hanging down as if the flower were a chair, or else he kneels, often with one leg extended over the edge of the flower (see SMA, Pls. 9-10, Nos. 189-210). I remember no example in which he stands on a lotus, although standing types are common, especially when the god is shown as a youth rather than as a child. The validity of this observation is not affected by a peculiar case like that of a little terracotta flask in the Fouquet collection (Perdrizet, Terres cuites doel a Collection Fouquet, Text, p. 94, No. 238; pl. 36, middle row, right and left). It is made in the shape of an Eros standing in a lotus flower with his hands tied to a column behind him. The lotus shows that here, as often, Eros and Harpocrates are assimilated (see below); but the figure stands merely because it is one of the numerous objects derived from the type of Eros standing bound to a column or a tree (references in SMAI, p. 121, note 68). There may be dynastic sculptures or paintings unknown to me which represent Harpocrates standing on a lotus; but I have seen no such design on a genuine amulet of the Roman period. 2. In this example the flower is certainly not a lotus, as the history of the type requires, but a bell-shaped flower something like a tulip. The maker has opened a sector of the margin to show the young god's leg in the flower-cup, here manifesting some aesthetic judgment, since the effect would have been awkward if the rim of the flower had hidden the youth's leg from the knee down. In connection with this criticism and the preceding one, it may be remarked that the present design was probably made under the influence of some fine gem of earlier style representing Eros rising from the cup of a flower. The most striking example is the Demidoff banded agate, Eros rising from a pomegranate flower and holding branches of fruit in his hands (Furtwaingler, Antike Gemmen, I, pl. 24, No. 50; cf. No. 49 on the same plate, and a different conception of the subject, pl. 27, No. 1). The rapprochement is not inappropriate, since some of the genuine representations of Harpocrates, especially in the minor arts, have been assimilated to the Greek god of love (SMA, p. 144); but so marked a departure from the Egyptian type is still disturbing.