CAROLINGIAN EUROPE AND THE ARABS : THE MYTH OF THE MANCUS (^)

Three interrelated ideas have come to be widely accepted in récent years regarding the currency and coinage of western Europe during the Carolingian and early feudal âge. The first is that the stock of gold in the west was much larger than was formerly supposed. The second is that an appréciable proportion of this gold was in circulation in the form of coin. The third is that this coin was not, with rare exceptions, struck in Europe itself ; it consisted of Arab , which circulated under various names in western Europe in much the same way as did the florin outside Italy in the fourteenth century, or as the Maria Theresa dollar has done for the last century and a half in certain régions of the Middle and Far East {^).

(1) This paper was read at the Anglo-American conférence of Historians in Lon­ don on 14 July 1951. A summary of it was published in the BULLETIN OF THE INSTITUTE OF HISTORICAL RESEARCH, t. XXV (1952), pp. 15­16. I have left its form almost unchanged, while bringing the material and the footnotes up to date. (2) The three sLudies of cardinal importance dealing with this subject are Ugo MONNERET DE VILLARD, La monetazione nelV Italia barbarica, RIVISTA ITAHANA Di NuMisMATicA, t. XXXII (1919), pp. 22­38, 73­112, 125­38 ; t. XXXIII (1920), pp. 169­232 ; t. XXXIV (1921), pp. 191­218 ; Marc BLOCH, Le problème de l'or au moyen âge, ANNALES D'HISTOIRE éCONOMIQUE ET SOCIALE, t. V (1933), pp. 1­34 ; and Maurice LOMBARD, L'or musulman du VU' au XII' siècle, ANNALES : ÉCO­ NOMIES, SOCIéTéS, CIVILISATIONS, t. II (1947), pp. 143­69. The first of thèse, though a work of great learning and extremely useful as a collection of material, is con­ fused in its arrangement, uncritical in its handling of both the documentary and the numismatic évidence, and quite incorrect in most of its conclusions. Its gênerai ténor may be judged by the triumphant but quite unjustified statement with which its author terminâtes his third section (p. 138) : « Conto con queste ricerche d'aver seppelita una volta per sempre la teoria generalmente accettate del mono­ metallismo argenteo durante l'alto medioevo in Occidente ». Marc Bloch's article is excellent; the only reproach one can make against it Is that its author relied almost entirely on Monneret de Villard for his account of the mancus. M. Lom­ 1060 PH. GRIERSON (2)

The first of thèse ideas may well be correct, despite the fact that it conflicts with the generally accepted but somewhat dubious theory, that the passage of western Europe from a gold to a silver coinage in the course of the seventh and eighth centuries was due to the exhaustion of its stocks of gold owing to an unfavourable trade balance with the east. It is certainly correct so far as England is concerned : one has only to glance through Miss Whitelock's volume of Anglo-Saxon wills to see that in the two centuries before the Conquest every person or household of any pretensions to rank and wealth possessed many orna- ments — drinking cups, armlets, baldrics, scabbards, and so on — either entirely of gold or heavily gilded, as well as what one can only interpret as a private hoard of gold in the form of plate, ingots, and coin (}). It is also true of Italy. Only for Francia is the évidence some• what scanty, though whether this is due to defects in the évidence itself or to a real différence of conditions it is difficult to say (^). It is a question, however, which need not be investigated here.

bard's article is extremely interesting and suggestive, but in my opinion it is toc schematic and exaggerates the influence of Islamic coinage in the west. Nor does it provide the évidence by which its conclusions can be tested. See also the articles of Kennepohl and Blanchet cited below, p. 1065, n. 3, and p. 1068, n. 1. Alfons DopscH believed firmly in the persistence of large quantities of gold in the west (Die Wirlschaftsentwickelimg der Karolingerzeit, 2nd éd., t. II, Vienna, 1922, pp. 305- 308 ; Naturalwirischaft und Geldwirtschaft in der Weltgeschicbte, Vienna, 1930, pp. 119-21). Henri PIRENNE no less categorically denied it (Mahomet et Charle- magne, 5th éd., Brussels-Paris, 1937, pp. 220-22, esp. p. 221, n. 3). (1) Dorothy WHITELOCK, Anglo-Saxon Wills (Cambridge, 1930). To take one example at random, the bequests of a certain Brihtric and his wife Aelfswith, who died c. 980, include an armlet weighing eighty mancuses of gold to the king and one of thirty to the queen ; thirty mancuses of gold, a necklace of forty man• cuses, a silver cup and half a gold headband to Rochester Cathedral ; sixty man• cuses of gold, a necklace of eighty mancuses and two silver cups to Christchurch, Canterbury ; thirty mancuses of gold, two silver cups, and half a gold headband to St. Augustine's ; ail this over and above various estâtes and sums of money to other legatees (no. XI, pp. 26-9). (2) The main source of évidence from England consists of wills, and thèse are rare for France. One of the few examples, that of Eberhard of Friuli, son-in-law of Louis the Pious, does include a considérable number of gold or gilded objects (I. DE CoussEMAKER, CaHulaire de l'abbaye de Cysoing, t. 1 [Lille, 1885], pp. 1-5). The share of Raoul, Eberhard's fourth son, includes mancosos c, but it is not clear whether this refers to actual coins or to gold by weight. (3) THE MYTH OF THE MANCUS 1061

The other two ideas, that there was an abundance of gold coin in circulation, and that this coin consisted mainly of Arab dinars, can scarcely be considered separately, for if the first is not true, the second does not arise. Before going on to discuss the évidence for the continued use of gold coins, it will be as well to examine briefly what thèse coins might be. This may be most conveniently done in summary form. In addition to what survived from the past — a good deal of Lom• bard and Byzantine gold in Italy, Uttle if anything elsewhere — wes• tern Europe in the three centuries between 750 and 1050 could receive gold coinage as follows :

(1) From extraneous sources : (a) Byzantine solidi, struck continuously at Constantinople through- out the period and up to 878 at Syracuse (^). (b) Beneventan solidi, becoming baser as time went on, and coming to an end c. 850 {^). (c) Arab dinars, struck by the Abbasids and later dynasties at many eastern mints, and by the Umayyads in Spain from 929 onwards

(1) For Byzantine coins, the best worlc for this period is W. WROTH, Catalogue of Impérial Byzantine coins in the British Muséum (2 vols., , 1908). Italian issues are conveniently listed in G. SAMBON, Répertoria générale délie monete co- niate in Italia. Parte I (Paris, 1912). For Sicilian issues, an essential revision of parts of Wroth and Sambon is given by D. RICOTTI PRINA, La monetazione siciliana nelV epoca bizantina, NUMISMATICA, t. XVI (1950), pp. 26-60. (2) The standard works are W. WROTH, Catalogue of the coins of the Vandals, Ostrogoths and Lombards ... in the British Muséum (London, 1911) ; G. SAMBON, Recueil des monnaies médiévales du sud d'Italie avant la domination des Normands (Paris, 1919) ; M. GAGIATI, La Zecca di Benevento (Milan, 1916-17). See also Sam- bon's Répertoria générale, referred to in the preceding note, and vol. XVIII of the Corpus Nummorum Italicorum. Thèse gold coins were struck for the last time by Radelchis ot Benevento (839-51), as also by his contemporary Siconulf of Salerno (839-49). They are very light (c. 3,8 g.) and much debased ; spécimens in my collection are of 8 carats or less. (3) For thèse coinages, in addition to the standard catalogues of S. LANE-POOLE (British Muséum), H. LAVOIX (Bibliothèque Nationale), and H. NUTZEL (Berlin), one has now G. C. MILES, Rare Islamic coins (New York, 1950) and The coinage of the Umayyads of Spain (2 vols. New York, 1950). 1062 PH. GRIERSON (4)

(d) Arab gold coins, known in Latin Christendom as taris (1/4 di• nars), struck in North African mints and in Sicily from the early tenth century onwards (^), and imitations of them struck at Salerno and Amalfi (the latter very base) in the second quarter of the eleventh century (^). (2) From domestic sources : (a) Gold tremisses struck by the last Lombard kings, by some Itahan cities in 774-81, and by Charlemagne in the period immediately foUowing the conquest (^). It should be noted that the often cited gold sohdus of Uzès, generally attributed to Charlemagne, is in fact a feudal pièce, probably of the tenth or eleventh century (*), and that the denarii auri, which ap• pear from time to time in ninth century Burgundian charters and are often cited as évidence of Carohngian gold coinage, were not necessarily coins at ail, but may have been - weights of gold. If they were coin, they were probably struck for the spécifie purpose of cérémonial payments as was the case with the gold « coins » of Uzès. (b) Sohdi struck by Louis the Pious in the years immediately following bis coronation, and the very numerous imitations of thèse made in Frisia during the next half century (^).

(1) SAMBON, Repertorio, pp. 128 ff., and after the Aghiabids were supplanted by the Fatimids G. C. MILES, Fatimid coins in the collection of the University Muséum, Philadelphia, and the American Numismatic Society (New York, 1951). The earliest dated coin known to Sambon is of A.H. 302 (= A.D. 914/5). (2) SAMBON, Repertorio, pp. 80 ff., and Corpus Nununorum Italicorum, t. XVIII, pp. 1-2 (Amalfi), pp. 307, 315-18 (Salerno). The earliest Salernitan issues are attributed to the mid-tenth century, but so early a date seems to me unlikely. The identification and attribution of the Amalfitan séries is much disputed. (3) SAMBON, Repertorio, pp. 54 ff., with références to the literature. The date of the municipal coinage is disputed. I hope to give reasons elsewhere for placing it between 774 and 781. (4) See my article. Le sou d'or Uzès, which will appear in the next issue of LE MOYEN AGE. On the question of whether Charlemagne struck gold in Frisia, see W. HàVERNICK, Die Anfânge der karolingischen Goldprâgung in Nordwesteuropa, HAMBURGER BEITRàGE ZUR NUMISMATIK, t. VI-VII (1952-53), pp. 55-60, and my comments on it in the forthcoming t. VIII (1954). (5) See my article, The gold of Louis the Pious and ils imitations, JAAH- BOEK VOOH MUNT- EN PENNINGKUNDE, t. XXXVIII (1951), pp. 1-41. (5) THE MYTH OF THE MANCUS 1063

(c) Isolated gold coins struck in England — the gold of Offa (757-96), the solidus of Archbishop Wigmund of York (837-54), gold of Kings Edward the Elder (900-25), Ethelred II (979-1016) and Edward the Confessor (1042-66) (i). Thèse are each of them unique, but one remembers the in- junction of King Eadred (ob. 955) in his will that his exe- cutors should cause 2000 mancuses by weight of gold to be minted into mancuses and distributed between varions named ecclesiastical dignitaries (^}. (d) Some isolated feudal or royal gold coins, parallel to the En• glish séries, struck on the Continent ('). Most of those which survive — less than a dozen ail told — or are known from literary sources, date from the twelfth century, but some are of the period with which we are concerned : e.g. one of Bishop Bernold of Utrecht (1027-54). Thèse coins were not struck for purposes of commerce, but for occasional symbolic pay- ments, usually of an ecclesiastical character. The same is pro- bably true of the English coins just mentioned, except per- haps those of Offa and Archbishop Wigmund (*).

(1) For Offa's dinar, see below, p. 1069 ; for the gold p enny of Wigmund, see my article on The gold solidus of Louis the Pious, pp. 16-17 ; for the other three coins, see D. F. ALLEN, Edward the Confessor's gold penny, which discusses the penny of Ethelred as well as that of Edward the Confessor, and C. E. BLUNT, A gold penny of Edward the Elder, both in the BRITISH NUMISMATIC JOURNAL, t. XXV (1945- 48), pp. 259-81. (2) Select English Historical Documents of the Ninth and Tenth Centuries, ed. F. E. HARMER (Cambridge, 1914), p. 35. (3) A number of instances are collected by MONNERET DE VILLARD, art. cit. (1919), pp. 126 ff., by A. WAUTERS in Le monnayage de l'or en Belgique au XII' et au XIII siècles. BULLETIN MENSUEL DE NUMISMATIQUE, t. II (1882-3), pp. 122-7 ; t. III (1883- 1884), pp. 95-6, and by K. KENNEPOHL, Goldzahlungen in Westfalen im 11. bis 13. Jahrhundert, HAMBURGER BEITRàGE ZUR NUMISMATIK, t. III (1949), pp. 15- 20. Both Wauters and Monneret de Villard assumed that thèse coins were chance survivais from an abundant and regular coinage of gold, which is certainly in• correct. (4) Thèse weigh 4,28 g. and 4,38 g. respectively and are of spécial types, while the other English ones are gold pennies of a Uttle over 1 g. and struck with the dies normally used for silver coins. It is therefore arguable that the coin of Offa was meant to circulate on a par with the dinar and the pièce of Wigmund on a par 1064 PH. GRIERSON (6)

(e) Dinars struck in Barcelona in the first half of the eleventh century, imitating those of the Umayyads of Cordova, which were only being coined very irregularly after the death of the great vizier Ahnanzor (1002) (^). This hst of the possible sources of gold coin during the period is obviously an impressive one, but it is after ail nothing more than a hst of possibilities. Only the évidence of archeology and documentary sources can tell us to what extent the coins were in fact current. The archeological évidence is meagre and disappointing in the ex• trême. For Italy we have clear évidence of the common circulation of Byzantine sohdi, both of eastern and of south Italian origin, and of Beneventan solidi. There are few descriptions of hoards available, but this is due to the unsatisfactory character of the Italian law of treasure trove and the déplorable efficiency of Italian coin dealers ; the coins themselves are for the most part common {^). There has been at least one find in north Italy which included Arab gold (^), while of course the local Arab gold of Palermo was plentiful in the south. But finds of Byzantine, Italian and Arab gold are extremely rare in Francia and Britain. If one excludes a few hoards from the extrême south of France, where coin from Spain would easily penetrate, and the imitations of with the solidus, while the gold pennies, struck to no accepted weight standard, had no obvions commercial utility. My own belief is that both séries were destined for the non-commercial function of making cérémonial payments. For a number of instances of such payments, see Le sou d'or d'Uzès, cited above, p. 1064, n. 4. (1) There is an excellent study of thèse by Joaquim BOTET Y SIS6, Les monedes Catalanes, t. I (Barcelona, 1908), pp. 22 ff., esp. 72-3 and a convenient summary by MONNERET DE VlLLARD, art. Cit. (1919), pp. 85-88. (2) Thèse hoards do not go beyond the mid-ninth century. Among the latest are those of Nocera Tirinese (Notizie di Scavi, 1916, p. 356) and Capo Schis6 near Taormina (S. L. AGNELLO in ATTI DELLO VIII CONGRESSO INTERNAZIONALE DI STUDI BIZANTINI, Palermo, 1951, t. I, p. 311). When the author was in Naples in 1949, there was in the hands of the dealers what was evidently the remains of a very large hoard of the period of Theophilus (829-42). We know from the lite- rary sources that contemporary Byzantine nomismata were still circulating in south Italy down to the mid-eleventh century. (3) A find of over 100 coins made in the river Reno near Bologna in 1857. Only 39 were recovered, 23 being Byzantine, 5 Beneventan, and 11 Abbasid, the latest bearing the date A.H. 198 (= A.D. 813-14). See L. FRATI, Délie antiche monete d'oro ritrovate in Reno nell' agosto delV anno 1857 (Bologna, 1857). (7) THE MYTH OF THE MANCUS 1065 the solidi of Louis the Pious struck in Frisia, I doubt if a total of fifty gold coins have been found in thèse countries for the whole of the three centuries with which I am concerned (^). On the other hand, the Uterary évidence is relatively abundant. It is clear from it that gold coins, described as sohdi or mancuses, were commonly used in Italy up to at least c. 850, and that in south Italy Arab gold was in current use after that time ; that gold coins (aurei, bizantï) were fairly widely known in England during at least the tenth century — there are, for example, a number of purchases of land for gold coin recorded in Ely charters (^) ; and that gold coinage of one sort or another was in continuous use in northern Spain. For France and Germany, the évidence that we possess does not suggest any wide- spread use of gold. The words used in our sources to describe the gold coins in question are aurei, solidi, bizanti, and mancus{s)i or solidi mancus(s)i. The first three of thèse require no comment. Aureus is a neutral word which could be used to describe any gold coin. Solidus is ambiguous, for it could be used to signify either a gold coin, or the value of this in sil- ver, generally assumed to be 30 silver pence, or the sum of twelve silver pence. Only the context can show which meaning is intended in any particular instance and it does not always do so. Bezant would

(1) See the list of coin finds given by MONNERET DE VILLARD, art. cit. (1919), pp. 77-8. (2) For example, the dealings of St. Aethelwold and Abbot Brihtnoth of Ely (970-81) with a certain Alfwold over the buying and selling of estâtes, involving sums of XX aurei and auro xl solidi (Liber Eliensis, II. 10 ; ed. D. J. STEWART, London, 1848, pp. 120-1). See also the charter of Dunstan to Westminster cited below, p. 1069, n. 1. Références to mancuses are even coraraoner, but in the vast majority of cases, if not in ail, a mancus auri means a weight of gold and not a coin. A large number of références are collected by MONNERET DE VILLARD, art. cit. (1919), pp. 92-97 and passim. Unfortunately he did not realize that a very high proportion of Anglo-Saxon charters are wholly or partly false and that the neces- sary critical study of them has not yet been systematically done. There are only a few Works, such as A. S. NAPIER and W. H. STEVENSONS'S The Crawford Col• lection of early charters and documents now in the Bodleian Library (Oxford, 1895), which are absolutely reliable, and the book which he used, Hubert Pierquin's Re• cueil général des chartes anglo-saxonnes (Paris, 1912) despite an impressive exterior, is about the worst possible, for it is completely uncritical. Many of his citations must in conséquence be discounted. 1066 PH. GRIERSON (8) at this time normally mean the Byzantine gold solidus. But the signi• fication of mancus is open to discussion. The Word itself is used over a period of about three hundred years, from the last décades of the eighth century to the end of the eleventh, or rather longer if one takes into account the répétition of références to it in later charters and its literary use in the Chansons de Geste (^). It is clear that it can mean a coin {^), or the weight in gold of this coin (^), or the équivalent value of this coin in silver currency (30 pence) (*). The area in which it is used is quite well defined : Italy and England throughout the period, and Spain during roughly the century and a half between c. 950 and 1100. Elsewhere, in France and Germany, there is no évidence that the coin indicated by this word was ever current, though the word was known and understood in a rather gênerai sensé (^). It never reached Scandinavia or the lands of Eastern Europe, nor is the word found in Byzantine Greek. There can be no doubt at ail that the mancus, as the word is used in Spain in the period c. 950 to 1100, is the dinar struck by the Umay- yads at Cordova and imitated by the Christian counts of Barcelona and the various Taifa kings who succeeded the Umayyads. For this there is abundant documentary évidence, which there is no need to recapitulate here (^). Several varieties of the Spanish mancus are known from our sources, which distinguish coins struck by différent rulers and in différent places. The évidence for the use of Arabie gold coinage in

(1) The most récent article on the latter is that ot Adrien BLANCHET, Les mon• naies dans la chanson de Roland, COMPTES RENDUS DE L'ACADéMIE DES INSCRIP• TIONS ET DE BELLES-LETTRES, 1942, pp. 36-48, who argues that the références to manguns (i.e. mancuses) and besants fins indicate a date of composition c. 1075- 1090. His reasoning seems to me extremely hazardous. (2) Solidus mancus can scarcely mean anything other than a coin ; it is mancus standing alone that is ambiguous. (3) This is the normal meaning tobe attached to it in English documents, where bracelets, cups, etc. are regularly described as being of so many mancuses. (4) This value is given by the Anglo-Saxon writer Aelfric, and confirmed by many other sources, as for example the text cited below, p. 1073, n. 1. (5) The évidence for ail this is collected by MONNERET DE VILLARD, art. cit. (1919), pp. 73 ff., and although some of the citations will not stand up to a critical examination the main lines are certainly correct. (6) See the références to BOTET Y SISO and MONNERET, above, p. 1066, n. 1. (9) THE MYTH OF THE MARCUS 1067

Christian Spain is overwhelming, and can occasion no surprise, since Arab influence in the Christian kingdoms of the north is abundantly demonstrated in other fields. Similar conditions, however, did not obtain beyond the Pyrénées. The view that the mancus of Italy and England is also an Arab dinar, at least when it is not a weight or a sum in silver, rests upon four considérations : (1) the fact that although some old German glosses treat mancus and bezant as if they were interchangeable, the wording of a West• minster charter of Dunstan appears to make a distinction between mancusi auri and bizanctei nummi (}). If, so the argument runs, the mancus is not a bezant, it can only be a dinar. (2) the existence of some imitation dinars, made probably in Francia or England, and in particular Offa's dinar bearing the date A.H. 157 (= A.D. 773/4) (2). This was found in Italy, and scholars have very reasonably related it to be the 365 mancuses which Offa promised as an annual gift to the Pope (^). (3) the fact that there is an Arab p ast participle, manq 'sh, meaning something « engraved », and that naqash, the verb from which it comes, is regularly used for « to strike » (coin). (4) finally, the assumption that words do not normally change their meaning. If mancus meant a dinar in tenth century Spain, the pre- sumption is that it also meant a dinar in eighth century England.

(1) J. M. KEMBLE, Codex diplomaticus Aeoi Saxonici, t. VI (London, 1848), no. 1223, pp. 13-19. It is cited by MONNERET DE VILLARD, art. cit. (1919), p. 102, as proving conclusively that the mancus and the bezant were distinct coins. (2) Offa's dinar is an imitation of an Abbasid one dated A.H. 157, but with the words OFFA REX included in the design ; the others are of the same date but with no words on them, and the tact that they are not genuine originals is only apparent from the blundered way in which the Arab legend is rendered. See especially John ALLAN, Ojfa's Imitation of an Arab dinar, NUMISMATIC CHRONICLE, 4th séries, t. XIV (1914), pp. 77-89, and MONNERET DE VILLARD, art. cit., pp. 95-6. (3) Allan categorically rejected the view held by a number of earlier scholars that Offa's dinar was struck in connection with his promised gift to the Pope. Monneret de Villard, who rightly criticizes (p. 79, n. 5) the extrême inadequacy of Allan's knowledge of documentary allusions to the mancus, regards this con• nection as probable, a view with which I completely concur. 1068 PH. GRIERSON

None of thèse considérations is décisive. (1) The alternative « bezant or ma nous » is a false one. Though the charter of 959 cited by Monneret de Villard is, like ail early Westminster charters, a palpable forgery of the late eleventh or early twelfth cen• tury, the passage in which the références to bezants and mancuses occurs may well have been taken over from a genuine document (^). But the antithesis is between gold by weight and actual bezants, the first being normal and the latter a little exceptional (^). There is nothing in the document to suggest that the mancuses are specifically coins, while the bizanctei nummi cannot be anything else. (2) the équation « Offa's dinar = a mancus », though rejected by Allan, may well be correct ; the existence and weight of the coin fits in too well with the literary évidence to be passed over. But it does not prove that the meaning of the word was limited to dinars, any more than bezant in later usage necessarily meant a Byzantine no- misma or a florin a coin of Florence. (3) the dérivation of mancus from the Arabie manqûsh was first

(1) The charter is one of a celebrated group of forgeries which have been often discussed ; see J. ARMITAGE ROBINSONS'S introduction to his édition of John FLETE, The History of Westminster Abbey (Cambridge, 1909), pp. 12 ff., and David KNOW- LES, The Growth of Exemption : Westminster, DOWNSIDE REVIEW, t. L (1932), pp. 413-20. Armitage Robinson points out, with regard to this particular charter, that the pseudo-original is in a late eleventh century hand and has a completely spurious seal, that the date and the names of the witnesses are chronologically thirty years apart,and that the dispute with the bishop of London over privilèges, which forms the main subject matter of the charter, reflects the conditions of the Conqueror's reign. It is in fact bound up, both in subject matter and phraseology, with the other forgeries of the séries. Mr C. N. L. Brooke, who has been kind enough to advise me on it, points out further that its use of canon law quotations, partly Pseudo-Isidorian in origin, argues a post-Conquest date, and that it introduces an archdeacon, which a genuine charter of Dunstan could not have done. (2) Bezants are only mentioned once, mancuses repeatedly. A typical passage showing the use of the various terms employed — pounds of silver, mancuses of gold, and bezants — is the foUowing : « Quidam autem minister régis ... petiit a me (i.e. Dunstano) quoddam amminiculum, qui commodans ei XXX libras ar• gent! suscepi ab eo VIII mansos ... Item X cassatos emi ab Aelfego duce ducentis mancusis auri ... Et illam possessionem in Scepetune erai ab Aelfleda vidua LX bizancteis nummis ». (11) THE MYTH OF THE MANCUS 1069 proposed by Adrien de Longpérier (i) and was accepted by the great orientalist Reinhart Dozy (^). That manqûsh could be used as an epithet for a dinar is certain, and one can think of analogies in other languages, like signatus or figuratus in Latin and x'^QÔ^yi^a'<:oç in Greek. But this use is rare and poetical ('), and it does not appear that man• qûsh is actually used as a synonym for a dinar (*), or if it is such use must be a very rare one and would scarcely have been the word for the coin borrowed by its Latin users. Arab scholars in gênerai regard the dérivation of mancus from manqûsh as untenable (^). (4) finally, in the history of money, the names of coins are anything but stable. The penny of Offa of is not that of Queen Eliza- beth II of England, though the same name covers both and there is no break in the évolution of the one into the other. The word florenus started as that of the silver coin of Florence in the late twelfth century. When Florence created a gold coinage in 1252, the new coin was known as a florenus ami, and since for several décades this was virtually the only gold coin in use, ail new gold coins as they were created came to be known as florins, however much they might differ in weight and value from the original one. The word ducatus was that originally given to a base silver coin of the Norman kingdom of Sicily struck in 1140 for the duchy of Apulia. If then came to mean any silver coin larger

(1) In a note by J. Y. AKERMAN, The gold « mancus », NUMISMATIC CHRONICLE, t. V (1843), pp. 122-4, and again by LONOPéRIER in the REVUE NUMISMATIQUE, 1844, pp. 291-3. (2) Supplément aux dictionnaires arabes (Brill, 1881), t. II, p. 712. (3) Dozy cites a passage from tlie Kitàbu 'I-Agliânï (« Booli of Songs ») com- piled by Abu '1-Faraj of Isfahan (ob. 967) in which a handsome youth's face is compared to al-dïnâr al-manqûsh, « a well-engraved dinar », in the sensé that his features were well-defined and clear, as regular as those of the Cufic inscription on a coin. This is scarcely sufficient to provide an etymology for mancus. (4) Dozy cites the Muhit-al-Muhit, a dictionary printed at Beyrout in 1870, as his authority for the use of manqûsh elliptically for dinar. But he admits that his authority is one which « il faut se servir avec prudence », and Professor A. J. Arberry informs me that it may in fact be discounted. (5) This was emphatically the view of John Allan, and Professor Arberry, who has been kind enough to advise me on the whole question, is of the same opinion. When the name of a coin is borrowed from one language to another, it is the com- mon and not the rare one that will be used, and in this case it is not even esta- blished that manqûsh was ever the name of a dinar at ail. 1070 PH. GRIERSON (12) than the ordinary Italian denaro, and so was naturally applied to the silver grosso of Venice, created in 1202. When Venice started striking gold coins in 1284, thèse were known as ducati ami, and the word was quickly limited to the gold. Finally, the old dérivation of the word ducatus from the ducatus Apuliae was forgotten, and a new one, based on the final word of the reverse legend, was substituted in its place (}). Ail thèse, however, are négative considérations. They merely prove that the identification of mancus with dinar is less probable than had been supposed ; they do not disprove it. The crucial point of the argument lies in the précise circumstances in which we first hear of mancuses. We first hear of mancuses in the last décades of the eighth and the first of the ninth century. The areas are the former Byzantine terri- tories of Ravenna and Istria, the former Byzantine territories around and Naples — the great abbey of Farfa, near Orvieto, is an important source of évidence — and neighboring Frankish (formerly Lombard) régions (^). For example, in 786 a certain Paul the Falconer bound himself not to go back on the terms of a contract he had made with the abbot of Farfa on penalty of paying vobis vel vestris succes- soribus auri solidos mancosos centum, and in 799 a native of Milan named Sarengo rented some property from the same abbey in return for an annual payment in auro aut in argento vel pannos valentes man- coses decem (^). Thèse solidi mancusi, either as priées or fines, are cited time and again in charters as if they were the normal coin of the country, though there is sometimes a suggestion that they may be becoming dif- ficult to obtain in what was later Frankish territory. For example, a lost charter of Charlemagne to San Zeno of Verona — it is known from its confirmation by Louis the Pious of 19 November 815 — specified

(1) The traditional dérivation, still given in ail works of référence, is that the word cornes from the hexameter which forms the reverse legend of the gold ducat : SU tibi Christi datas, quem tu régis, iste ducatus. The incorrectness of this is proved by the fact that the word is applied by a chronicler to the silver ducat, on which there is no such legend, before the gold ducat came into existence. See N. PAPA- DOPOLi, Le monete di Venezia, t. I (Venice, 1893), p. 81. (2) Much of the évidence is collected by MONNERET DE VILLARD, art. cit., pp. 80-4. (3) Regesto di Farfa, ed. I. GIOROI and U. BALZANI, t. II (Rome, 1879), pp. 119, 136. (13) THE MYTH OF THE MANCUS 1071 that on the festival of the saint the abbey should pay the bishop of Verona and his clergy aut manculos viginti aut quinquaginta solidos argenti (i). There is one document which is more revealing than even private and royal charters. Sometime late in the reign of Charlemagne, three of his missi held an enquiry in Istria into the jurisdiction of the Pa- triarch of Grado and complaints against Duke John of Istria (-). The enquiry took in part the form of asking what was the custom up to and on the day on which the land passed from the Greeks to the Franks, and local représentatives answered on oath the questions put to them. To a question regarding the tribute paid by the cities of Istria, it was answered that the city of Pola paid 66 solidos mancosos, Rovigno 40, Parenzo 66, Trieste 60, Albona 30, and so on, up to a total of 344 man- cuses. AU this was the custom tempore Graecorum. Now it is inconceivable that thèse mancuses can be anything but the local gold coinage of Byzantine Italy. Arab gold was at this time only being struck in the eastern provinces of the CaUphate, not in Spain or north Africa ; it was no doubt known in western Europe, but comparatively small quantities can as yet have come in. On the other hand, gold was being struck at Rome up to and including the reign of Constantine V (741-75), as it was to be in Sicily for another century to come, and reckonings of payments — in particular, ail assessments of taxes — would naturally be in terms of the impérial coinage. The mancus must be the local solidus of Byzantine Italy. Why, if this was the case, was it not simply called a solidus? The answer to this question is given by the documents, which regu- larly distinguish between solidi « of pure gold » (solidi obrizi, buoni, adpretiati) and solidi mancussi, and by the coins themselves.

(1) F. UQHELLI, Italia sacra (2nd ed. Venice 1730), t. V, cols. 705-6. There is no modem édition, the volume of Louis' charters in the German Monumenta édition having not been published, but its genuineness is not open to question. Cf. J. F. BôHMER and E. MûHLBACHER, Regesten des Kaiserreichs unter den Karolingern, 751-918 (2nd ed. Innsbruclc, 1908), no. 597. (2) UGHELLI, op. cit., t. V, cols. 1097-1100. The document is discussed by L. M. HARTMANN, Geschichle Italiens im Mittelalter, t. III (i), Gotha, 1908, pp. 28-32, and in his Untersuchungen zur Geschichte der Bgzantinischen Verwaltang in Ita• lien (Leipzig, 1889), pp. 61-2, 91, 156-7, but he makes no comment on the nature of the monetary unit employed. 1072 PH. GRIERSON (14)

A study of the latter shows that from the beginning of the eighth century the sohdi struck in Italy diverged considerably from the im• périal standard in weight and fineness. A dechne in weight alone seems to have been characteristic of the mint of Syracuse ; a dechne in both weight and fineness is apparent in the solidi of Rome, which by the middle of the century often weigh no more than 3,5 g. and are of less than 8 carats fine (i). The issue of base solidi at Rome ceased when after the Frankish conquest Pope Adrian abandoned the coinage of gold in favour of that of silver. At Syracuse, by the end of the eighth century, the fluctuations in the weight of the sohdus were over, but it was being regularly struck to a weight one-seventh less than that of the solidus of Constantinople, i.e. a value of 21 instead of 24 sili- quae, or 3.87 instead of c. 4,40 grammes (^). It was the « déficient », i.e. light-weight solidus —solidus mancus (^) — of the north and centre that was the original mancus of the texts (*).

(1) Spécimens in the author's collection are so base, with densities of between 10 and 12,5, that it is impossible to ascertain their exact composition by gravi• métrie methods. The alloys used evidently contain copper, and probably tin and zinc as well. (2) See the figures and discussion by RICOTTI PRINA, art. cit., pp. 29-31, 40-55. He accepts, however, the Arabie dérivation of the word mancus (p. 58, note 16 bis). The theoretical weight of the solidus was 4.54 g., but in practice it was normally about 4.40 g. (3) This ought, in good grammatical usage, to have been given the plural solidi manci, when in fact we aiways find mancussi or mancosi. But the eighth and ninth centuries were not a period of good grammar, and I suspect that the word was treated as if it were grammatically équivalent to another, (per)cussus, which is also frequently used to characterize solidi. A parallel can be found in the con• fusion shown in Frankish charters of this period between coactus, the compulsion to pay a fine, and cactus, referring to the quality of the gold in which a fine must be paid. (4) That mancus raeant « defective », and had nothing to do with manqûsh was believed by Louis BLANCARD (in REVUE NUMISMATIQUE, 3rd séries, t. III, 1885, p. 230) and is the view of Adrien BLANCHET, art. cit., p. 43, but M. Blancard thought that it referred to the debased Barcelona mancus, in contrast to the good Umayyad dinar, and M. Blanchet that it referred to the fact that the mancus was worth only 30 deniers, while the ancient Frankish équivalent of the solidus had been 40 deniers. M. Blancard's opinion is certainly incorrect, since the word man• cus antedates the appearance of the Barcelona coin by two centuries, and the view I put forward in the text seems to me préférable to that of M. Blanchet, since (15) THE MYTH OF THE MANCUS 1073

This identification makes it possible to understand something of the évolution of the word. It was first locally used in Italy for a de- based solidus. In this form the many English travellers —• merchants, political envoys, missionaries, and pilgrims — who passed to and from Italy so frequently in the course of the eighth century became familiar with it. It would corne their way, after ail, more often than either Byzantine solidi or Arab dinars, for it would be the coin in use in Rome itself. When Offa promised to make an annual payment in gold, he would naturally do so in terms the pope could understand, in terms of the gold coin used in Rome, and this he did. This is turn would mean that he must know the weight of the coin (^), and in this way the reckoning in mancuses by weight would pass to England. We need not assume that many actual coins would make the journey from Italy to Britain ; if the weight were adopted by the court it would, through royal estâtes and goldsmiths, pass in time into common use, as did the pondus Caroli of Charlemagne at about the same time. The fact that the relations between England and Italy were indeed by way of Francia, but direct and not through the intermediacy of Franks, explains why we do not find the mancus playing the same part in French documents of the late Carolingian and early feudal periods as it does in Italian and Enghsh ones. The later vicissitudes of the word are fairly évident. Although the supply of Itahan mancuses must have run dry in the course of the ninth century, the word had by then established itself and was used either casually of any gold coin, or more precisely of gold coins of a certain weight, or as a money of account in silver. With the création of a gold coinage in Umayyad Spain in the early tenth century, the familiar word was naturally applied to the new coins. Though thèse differed from the original Italian mancus in appearance and fineness, they may well have approximated to it in weight, and it was as a standard of weight or as the generic term for any gold coin that the word mancus was by then mainly used. the mancus is heard of in Italy before it appears in France, where it was al ways rare, and the solidus of 40 deniers was a purely Frankish phenomenon. (1) The weights of actual spécimens are so variable that one cannot say pre• cisely what this was, though it was clearly less than that of the true solidus. I imagine that it was a little over 4 g. 1074 PH. GRIERSON (16)

The implications of this reinterpretation of the meaning of the word mancus on our reading of the économie history of western Europe in the Dark Ages are clearly far-reaching. Monneret de Villard's picture of western Europe between the eighth and twelfth centuries as being familiar with large quantities of Arab gold coin was in its essentials endorsed by M. Lombard in the brilliant article referred to in the first page of this paper. The validity of one of its lines of thought is seriously affected by the argument put for- ward in this study. For M. Lombard, there was from the ninth century onwards importation of gold from the Arab world into Western Christen- dom, mainly in exchange for slaves. Some of this gold found its way back into Arab lands, purchasing spices and articles of luxury from the east ; some of it passed by way of exchange to Byzantium, and played its part in maintaining the économie prosperity of the Eastern Empire ; but a substantial proportion remained in Europe, and in due course proved an important factor in the économie revival of the West. The évidence for this importation of Arab gold is in part hterary, in part archeological, and in part inferential. The inference is that it was used to pay for slaves, but exchange against articles of luxury, or rather, reinvestment in articles of luxury, is much more hkely. The archeological évidence is virtually non-existent (^). The literary évidence almost entirely dépends on the assumption that the mancus was always an Arab coin, and this I have shown was not the case. The importation of Arab gold in substantial quantities into western Europe in the Dark Ages must therefore be regarded as non-proven, and is indeed in the highest degree unlikely.

Philip GRIERSON.

(1) This statement of course refers to gold, and to western Europe. The évi• dence of coin hoards testifies to the import of Arab dirhems into eastern and nor- thern Europe on an enormous scale. Professor Sture BOLIN, in a most important article recently made available in an English translation (Mohammed, Charle- magne and Ruric, SCANDINAVIAN ECONOMIC HISTORY REVIEW, t. I [1953], pp. 5- 39), argues, despite the lack of archeological évidence, for huge imports of Arabie silver into western Europe as well, and makes out a good though not entirely convincing case for this.