The Olokun head reconsidered

doi:10.4000/aaa.266

9 | 2013
Varia

The Olokun head reconsidered

Un ré-examen de la tête d’Olokun

Paul T. Craddock, Janet Ambers, Maickel van Bellegem, Caroline R. Cartwright, Julie Hudson, Susan La Niece et Michela Spataro

p. 13-42

https://doi.org/10.4000/aaa.266

Résumés

English Français

En 2010, une exposition de têtes Ife au British Museum à Londres a fourni l’opportunité de procéder à un examen scientifique et technique détaillé d’une tête en laiton, plus connue sous le nom de tête d’Olokun, et publiée par Léo Frobenius en 1912 ; une sélection d’autres têtes Ife exposées ont fait également l’objet d’analyses. La tête d’Olokun avait été étudiée et reléguée, soixante ans auparavant, au statut de simple copie obtenue par fonte au sable. Cette nouvelle approche, particulièrement bienvenue, établit sans aucun doute que la tête d’Olokun est une fonte à cire perdue, très similaire aux autres têtes en termes de technique et de composition. Par ailleurs, tout un faisceau d’éléments suggère qu’il s’agit bien de la tête mise au jour à Ife en 1910.
Cet article livre les résultats du ré-examen technique et scientifique de la célèbre tête en laiton d’Olokun, originaire d’Ife (Nigéria), considérée en 1949 par William Fagg et Léon Underwood comme une copie réalisée par moulage de la tête originale découverte en 1910 par Leo Frobenius.
La discussion s’articule autour de trois grandes parties. La première examine les nombreux arguments utilisés par W. Fagg et L. Underwood pour reléguer la tête au statut de simple copie. La seconde décrit la façon dont on pense actuellement que la tête a été coulée. Enfin, la troisième dévoile l’examen scientifique établissant son ancienneté probable, sa production en Afrique de l’Ouest, et sa ressemblance frappante avec d’autres têtes d’Ife.
Parmi les nombreux points retenus par W. Fagg et L. Underwood, trois revêtent une certaine importance. Dans l’ensemble ils dénoncent la nature diffuse de la surface du métal et le manque de détails précis (les points a, b, f et g de Fagg et Underwood), alors attribués à la méthode de fonte au sable. En réalité, l’examen attentif a très vite démontré que l’origine de cette différence d’aspect résidait dans le fait que, contrairement aux autres têtes d’Ife présentes au British Museum de nombreux mois avant l’exposition – pour examen et nettoyage –, la tête d’Olokun n’était arrivée en Grande-Bretagne que la veille de son ouverture et n’avait, par conséquent, reçu qu’un traitement très superficiel. Sous la couche de cire appliquée à la hâte, la corrosion, la terre et même des restes du matériau de moulage d’origine ont subsisté, remplissant les interstices et brouillant les détails de la fonte (fig. 8).
Une objection plus sérieuse (g) résidait dans le fait, qu’en certains endroits de la tête, des accrétions de matière de surface paraissaient être en métal, comme si on avait là la traduction dans le métal du moulage de résidus, de poussières et de traces de corrosion, ce qui aurait pu effectivement être un argument en faveur d’une copie moulée sur l’original. Cependant, l’examen approfondi des zones concernées a montré que la plupart de ces « moulages de poussière » sont en fait constitués de vraies poussières et de résidus de fonte (fig. 9). À la base du cimier, on distingue quelques gerces de métal et le décor du cimier lui-même est recouvert de métal (fig. 10), mais cela pourrait s’expliquer par le simple fait que l’argile utilisée pour le moulage n’était pas assez fine pour pénétrer profondément dans cet interstice, ou encore que la surface du moule avait été endommagée pendant la cuisson et la coulée.
Leur point h nécessite également une explication. Ainsi W. Fagg et L. Underwood décrivent deux zones sur le front qu’ils attribuent à une tentative d’introduction de broches en fer dans un possible moule en plâtre. Cette interprétation est entièrement erronée. Pour que cela soit interprété comme la preuve que cette tête était une copie d’un original, le métal aurait dû être coulé par les canaux intérieurs dans l’original et ce que l’on voit dans le métal de la présente tête correspondrait alors au moulage des canaux de coulée de la tête originale. Toutefois ces canaux de coulée coupés contenaient encore les extrémités de broches de renforcement et par conséquent il s’agit bien, non de moulages de canaux, mais de vrais canaux qui pour une raison quelconque n’ont pas été utilisés. En outre l’examen indique que c’est la cire que l’on a coupée à cet endroit, et non le métal. L’explication la plus probable est que le moule a été assemblé, prêt à être chauffé, mais qu’il a subi un accident qui l’a endommagé avant le début de la coulée. Les canaux de cire originaux ont alors probablement dû être coupés et remplacés par deux autres à l’intérieur de la tête (fig. 2).
Dans l’ensemble il est à noter que, dans leur discussion, W. Fagg et L. Underwood ne mentionnent aucune des principales caractéristiques qui permettent de distinguer les fontes au sable ou les fontes postérieures, telle une réduction de taille de la copie qui aurait pu être constatée. De même, la présence de traits tout aussi distinctifs de la fonte à cire perdue qui peuvent être observés sur la tête ne sont pas évoqués.
L’examen scientifique a établi que le métal est sérieusement corrodé, avec une corrosion intergranulaire très importante, témoignant fortement d’un enfouissement prolongé (fig. 16 et 18). Les produits de corrosion sont les minéraux auxquels on peut s’attendre sur une véritable antiquité.
Il y a aussi une contamination très superficielle de pigments synthétiques. Ces derniers sont identiques à ceux trouvés sur les moulages en plâtre des têtes réalisés au British Museum, lors de la présence de la tête d’Olokun en 1948.
La composition élémentaire et isotopique du métal de la tête d’Olokun est similaire à celle des autres têtes d’Ife, ce qui suggère fortement qu’elle a été fabriquée à partir du même stock de métal.
L’examen du noyau de la tête indique une composition similaire à celle de l’une des autres têtes échantillonnée pour comparaison (Willett & Sayre 2006). Plus significatif encore, une partie de la matière végétale utilisée comme remplissage du noyau de la tête d’Olokun provenait de plantes qui poussent uniquement en Afrique de l’Ouest, signe évident que la tête a été coulée dans cette région.
Ainsi, dans l’ensemble, le réexamen jette un doute considérable sur le rapport de 1949. Il établit que la tête est certainement une fonte creuse à la cire perdue, probablement réalisée en fonte directe – bien qu’une fonte indirecte ne puisse pas être totalement exclue –, très similaire à celle mise en œuvre pour réaliser les autres têtes. L’ensemble de ces arguments suggère qu’il s’agit bien de la tête exhumée par L. Frobenius en 1910.

Haut de page

Entrées d’index

Plan

Introduction

The technology of the Ife heads

The Olokun head

A reconsideration of the 1949 report

A Technical and Scientific Re-assessment of the Olokun Head

Texte intégral

PDF 3,2M Partager par e-mail

Agrandir Original (jpeg, 708k)

Introduction

1   The Olokun head, in common with other heads from Ife, has been numbered and re-numbered since it (...)
2  In the conclusion of their 1949 paper Fagg and Underwood state that ‘We would make it very plain t (...)

1Since the ‘Olokun’ head¹ (figs. 1 & 2) was first shown to the German ethnographer and traveller, Leo Frobenius on his arrival in Ife in 1910 it has been the subject of intense interest and not a little contention (Frobenius 1913; Fagg & Underwood 1949; Willett 1976; Shaw 1978: 15-6). This present study is a report on the recent detailed scientific and technical examination carried out in the British Museum of the head now widely known by that name. This was undertaken to ascertain whether it is the original, as shown to L. Frobenius, or a copy as claimed by William Fagg and Leon Underwood in their 1949 publication. Since then, despite continued interest there have been few further technical studies specifically on the heads apart from F. Willett (1967: 52-56, 2004) and D. Williams (1974: 188-203), and rather more papers on their alloy composition (cf. infra and note 23). Speculation on who was responsible for the production of the supposed copy has largely focused on Frobenius².

Figure 1 – The Olokun head NCMM 38.1.2

Agrandir Original (jpeg, 564k)

Figure 2 – Sketch of the Olokun head showing the positions of particular features and from where samples were taken

Agrandir Original (jpeg, 608k)

3  In Ife Olokun is often referred to as a goddess. Olokun is more frequently viewed as a god in Beni (...)
4  Everyone assumed that the heads were of bronze, that is, copper and tin, until Moss’ analyses (see (...)
5  Note that location, discovery and subsequent history were very different from the other heads foun (...)

2In the course of his travels through West Africa in 1908 L. Frobenius heard frequent stories of a wonderful statue of the god Olokun located in Ife in the then recently created Colony and Protectorate of Southern Nigeria. Medieval Ife had been an important city state in West Africa which flourished as a commercial, political and spiritual centre with access to lucrative trade networks (Johnson 1921). L. Frobenius travelled to Ife in 1910, visited numerous sites and started excavating and uncovering an impressive collection of stone and terracotta art works. After repeated enquiries about the statue, he was taken out to the Ebolokun, an extensive grove dedicated to Olokun, the god of wealth and the sea³, and shown not a statue but a brass head. In Frobenius’ words (Frobenius 1913: 98) ‘Before us stood a head of marvellous beauty, wonderfully cast in antique bronze⁴, true to life, encrusted with a patina of glorious dark green’. Later in the same account he described it as ‘measuring 35.5 cm [cf. note 20] and cast... à cire perdue... very finely chased, indeed like the finest Roman examples’ (Frobenius 1913: 310). He was told that it had been dug up in the grove a generation or so previously and had been reburied there but was uncovered periodically and offerings made to it⁵. It was doubtless these ceremonies that led to the fame of the head spreading through West Africa. It was natural to identify the head with the dedication of the grove in which it had been found, but in fact it is more likely that the Olokun head is a stylised representation of one of the rulers of medieval Ife, most likely of the 13^th-15^th century AD .

6  The Colonial Secretary, Mr L.C. Moorhouse, sent Charles Partidge, the Resident at Ibadan, to inves (...)

3The head was then reburied but L. Frobenius showed great persistence in trying to acquire it and badgered the messenger of the ruler and the priest in charge of the grove to let him purchase it. A price was agreed. During excavation a fragment was broken off the head from the corner of the crown above the proper right ear and in the end that was to be the only piece that L. Frobenius managed to take back to Germany. Local unease amongst the elders at this disturbance and removal of the head led to an audience with the ruler of Ife, the Ooni. L. Frobenius sought to allay their worries by suggesting that a copy could be made by the local tinsmith, or by a ‘galvanic process’, that is by electroforming (cf. note 2). Further complaints led the British Colonial authorities⁶ to intercept Frobenius’ party on their way to the coast and force them to retrace their steps to Ife and to surrender many of the excavated items including the Olokun head. Thereafter the head was apparently kept in the possession of the priest of the Olokun Grove, although nothing is known of the conditions of its storage, whether reburied or kept above ground, until 1934 when the Ooni took it into his palace for safekeeping. On Frobenius’ return to Europe he published his discoveries and the impressive, if somewhat depleted, collection was distributed according to the terms of his expedition grant. The naturalistic terracotta heads and photographs of the Olokun head generated some attention in Europe for a short while (Read 1911 for example) but interest rapidly waned and little progress was made in investigating material from Ife until the discovery of some 17 related copper and brass heads and one half figure in 1938 at Wunmonije Compound in Ife. On a later visit to Nigeria in 1946, H. J. Braunholtz, Keeper of Ethnography at the British Museum suggested to the Ooni that works from Ife, including the Olokun head, should come to England for study, conservation and temporary exhibition.

7  With sincere thanks to Yusuf Abdallah Usman, Director-General of the National Commission for Museu (...)

4This present study was occasioned by a major exhibition at the British Museum in early 2010 of a large selection of Ife sculpture on loan from the National Commission for Museums and Monuments, Nigeria with most of the known copper and brass heads, including the Olokun head which enabled the latter head to be examined in context with its peers⁷. The study is in three parts; an overview of the technology by which the Ife heads were cast, a re-consideration of Fagg and Underwood’s 1949 report and detailed technical reports on the Olokun head.

The technology of the Ife heads

8  The process of hollow cire perdue casting as described in the catalogue accompanying the British M (...)
9  This is described correctly by W. Fagg and L. Underwood (1949) but in his book on West African bro (...)

5The Ife heads are usually approximately life-size or somewhat less (see Willett 2004 for the most detailed overall descriptions). They are hollow castings usually with the cranium open and this is the case with both the Olokun head and also with the British Museum head which seems to represent a crowned ruler (AOA 1939. Af34.1). Presumably they were originally completed with hair or a headdress of a different, less permanent material. Inspection shows that the heads were almost certainly cast by the direct hollow cire perdue or lost wax technique (fig. 3) as described by F. Willett (1967: 52-56 or 2004: I.4) and by W. Fagg and L. Underwood (Underwood 1949: 6)⁸. First a clay core was made to the approximate dimensions of the finished head but a little smaller. Examination of the interior of the Ife heads shows that the surface is generally smooth and follows the contours of the surface of the face and head quite faithfully, although of course in negative. This suggests that the core must have been carefully modelled and possibly fired before the application of wax to the required thickness⁹. Inspection of the heads suggests that virtually all of the detail was sculpted onto the wax, including the scarification, although these lines were sharpened after casting (cf. infra).

10  On head NCMM 79.R.11 (Willett 2004) the crest has broken and the iron armature is exposed. Testin (...)

6The wax was sometimes strengthened by an internal iron armature where part of the model protruded from the main body of the casting. Thus, for example, to form the crest on the front of the headdress on the small crowned head (NCMM 79.R.11) an iron rod was first pushed through the wax into the core and the wax that was to form the crest was then wrapped or melted around this¹⁰. On firing the wax would run out leaving the iron armature in place to be surrounded by the incoming brass.

7At this stage rods of wax to form the runners, down which the metal would run into the mould and risers, to allow the air to escape, would be fixed to the wax model. After firing the wax would run out leaving the runner and the smaller riser channels (fig. 3). The remains of iron pins can be observed in the centre of many of the risers and in some of the runner joins (fig. 4).

Figure 3 – Simplified sequence of direct hollow cire perdue casting

Agrandir Original (jpeg, 304k)

A: a core in the form of the sculpture is made in clay;
B: the sculpture is covered in beeswax over the clay core. Iron rods are inserted through the wax into the core to prevent movement during firing;
C: fine details are sculpted in the wax. Tubes of wax, known as runners (a), are applied at the top. Separate wax vents (b) are inserted to allow gases to escape during casting;
D: layers of clay are applied directly to the wax surface, enclosing the vents (b) and runners (a) to form a mould;
E: the entire mould is heated, melting the wax which is drained away through the runners;
F: molten metal is then poured through the runners into the cavities left by the wax;
G: after it has cooled, the clay mould is removed and the runners and iron rods are cut off to reveal the completed sculpture;
H: the sculpture is polished to produce a smooth surface

Figure 4 – Oxidised remains of an iron armature that would have run up the centre of the of wax rod forming one of the runners, now chiselled-off (cf. fig. 2)

Agrandir Original (jpeg, 1,6M)

8Typically there were four runners, two near the forehead and two further back, and four risers of smaller diameter. They were attached to the top of the head suggesting that the heads were cast more or less upright (fig. 3). From the position of the channels it appears that the moulds were not strictly vertical but angled at about 70^o or 80^o (as illustrated in Underwood 1949: 6; fig. 3). This angling is confirmed by the position of some of the risers; for example there is evidence of a riser behind the crest on both the Olokun and British Museum heads (figs. 2 & 5). If the heads had been cast in a vertical position then the crest, pointing straight up, could have formed the lower part of a riser, but the separate riser positioned where the crest swelled shows that the head must have been inclined such that an air trap would have been created if a riser had not been added there.

Figure 5 – Remains of a riser behind the crest on the Olokun head, showing that the crest must have been inclined to necessitate a separate riser at the baluster (cf. fig. 2)

Agrandir Original (jpeg, 1,2M)

NB the crest itself must have acted as the principal riser with another one probably at the back of the head

11  The micrographs were prepared by Dr. Godrichs of the Central Laboratory of the Belgian Museums. Mo (...)

9The next stage would be the moulding, in which the wax was coated with clay. The mould is likely to have been a composite formed of successive layers, commencing with a very fine but weak clay to pick up all the detail on the wax and progressing to a stronger clay that would remain intact on firing. So far no mould fragments contemporary with the Ife heads have been found and thus it is not certain whether the first layer was applied as a thick liquid, resembling a batter mixture as assumed by F. Willett (1967: 58), or as a solid, either as pellets pushed onto the surface as was Yoruba practice (Williams 1974: 181) or as thin layers resembling pancakes as is current practice in Benin. After the moulding was completed it would be dried to leather hardness and then fired in an inverted position in order to drain and burn out all the wax and to bake the clay of the mould and core to form a ceramic. Then, the still red hot mould was swiftly inverted and the metal poured very likely from a series of crucibles in quick succession into the sprue cup and thence into the mould. Metallographic studies on four of the heads (n^os NCMM 38.1.4; KM Ife 6; NCMM 1999.2.3 & NCMM 38.1.6) by Moss (his BM sample n^os 4, 6, 7 & 10) revealed large grain sizes suggesting that the metal had been poured into the moulds whilst they were still very hot and allowed to set slowly¹¹. The taper section polished on the neck of the Olokun head as part of the present investigation showed similar large grain sizes. The slow cooling is also suggested by the analysis of the British Museum head (Af 1939.34.1) which showed that there had been sufficient time for significant segregation to take place (Table 2).

The Olokun head

A reconsideration of the 1949 report

12  ‘The copy in the present Oni’s possession was suspected by me in 1945 of being a reproduction’ (Un (...)
13  This included metallographic examination and analysis. The conservation comprised partial cleaning (...)
14  Almost immediately after the publication of Fagg and Underwood’s paper Moss (1949) cast doubt on t (...)

10In 1949 the present Olokun head was published by William Fagg and Leon Underwood as being a recent sand casting using a model made in a plaster piece mould taken from the original head. Underwood’s suspicions were apparently raised after he saw the head for the first time in 1945 in Nigeria¹². In 1947, fourteen of the heads and a half figure from the Wunmonije Compound site came to the British Museum for scientific study and conservation overseen by A.A. Moss, then in charge of metals in the British Museum Research Laboratory, prior to their exhibition in 1948¹³. The Olokun head was brought separately by the Ooni of Ife, together with the Obalufon mask (NCMM 38.1.2) just prior to the exhibition. It was examined by W. Fagg and L. Underwood in the week following the closure of the exhibition in October 1948, immediately before its return to Nigeria, by which time Moss was out of the country and could not take part in its examination¹⁴. Inevitably there must have been very little time to clean the head before it went on exhibition, and the present study has shown that soil and mould material remain in interstices in several places on the head, often covered in wax applied as part of a necessarily hurried cosmetic treatment at the British Museum in 1948. This could account for the observed ‘granular’ appearance of the head (Fagg and Underwood’s points ‘a’ and ‘b’, cf. infra).

15  As the head had just been moulded in the British Museum together with the other Ife heads (Platte (...)

11The grounds for their conclusions that the present Olokun head was not the original were very fully argued by W. Fagg and L. Underwood, but no convincing evidence was advanced for it being a sand casting. This method is much used to cast relatively simple shapes, but to accurately cast a complex hollow shape such as the Olokun head would be challenging. W. Fagg and L. Underwood postulated that plaster piece mould sections were taken of the original head and from these negative sections a positive plaster piece mould was taken, again in sections. The pieces of the positive mould were then pushed into the sand to create the negative impression necessary to cast the positive after-cast. They suggested that it would have been possible to mould the outer surface in four or five sections¹⁵. The sand casting process will not be described here (see Chastain 2004 for a good practical description), but the final mould would consist of two halves with the core inside. Thus after pouring the metal there is almost inevitably a flash of metal running the length of the exterior of the casting. Returning to the Olokun head, if a sand cast had been attempted then the crest would have had to have been positioned at the junction of the two halves of the mould; that is the mould join would have run vertically down the centre of the head from the crest, down the nose to the chin. A similar vertical line on the opposite side would have run from the back of the head down to the neck. There is no evidence at all for this most characteristic feature of sand casting. The mould line would have run through both the decoration on the crest as well as through the scarified areas. These decoration and scarification lines are casts and it is difficult to envisage how a flash line passing through the decoration and scarified lines could have been removed from those areas while retaining their as-cast appearance. It is also puzzling why W. Fagg and L. Underwood chose not to mention the complete absence of the most diagnostic feature of the process they were postulating, except possibly a reference at point j (cf. infra). The presence of small amounts of clay moulding material surviving in some places on the head also mitigates against this being a sand casting.

12The specific reasons that W. Fagg and L. Underwood gave for the present Olokun head being an after-cast will be reproduced here (indented) and evaluated here in the same lettered sequence from ‘a’ to ‘o’ that they appeared in the original 1949 paper.

a) ‘The head […] exhibits no traces of chasing or polishing processes applied after casting […] in other words the exterior surface is ‘as cast’ whereas all the other heads have been very finely chased and polished’ (ibid.: 3).

16  Although L. Frobenius (1913: 310) specifically described the head as being finely chased.

13At first sight this appears true; however, detailed examination shows that everywhere mould, soil and corrosion still adhere beneath the waxing treatment (cf. supra) masking the original surface. This does mean that the detail is less crisp than on some of the other heads. However, in some areas of the scarification this extraneous material has been locally removed exposing cast lines that have been cleaned up revealing much crisper but still patinated detail, presumably from the original post-casting cleaning up (fig. 6)¹⁶.

Figure 6 – Detail of the scarification on the Olokun head where the cast lines have been accentuated by scorping probably with a steel point (cf. fig. 2)

Agrandir Original (jpeg, 1,8M)

b) ‘The whole outer surface of the head is characterized by a granularity which is absent in all the other heads and is clearly inconsistent with the wax technique adopted in making them’ (ibid.).

14This appeared to be true on first examination, although the surface accretions make close inspection of the original surface difficult. There is more adhering soil and corrosion than W. Fagg and L. Underwood appreciated, and this when waxed could produce the observed ‘granularity’.

c) ‘Careful examination of the parallel striations behind the left ear [….] shows their lower portions were cut in bronze, and not in wax as in the similar striations on the remainder of this head and on all the other heads […;] the cuts are [… considerably] less controlled in direction and in evenness of execution. […] Yet it is certain that these cuts were not made in the present head, for the unchased casting surface is continuous, even in the deepest parts of the furrows’ (ibid.).

15The features referred to are not at all clear. Because of the general diffuseness of much of the surface detail and the presence of the waxed accretions, it is often difficult to differentiate between cast and worked details, but at the point referred to here it does look as if the cast lines have been sharpened by working the present brass (fig. 6). As they are themselves corroded the work must have been done some time ago, most likely just after casting as part of the original finishing process. This feature occurs on other heads, notably NCMM 38. 1.4 (KM Ife 4 and F. Willett M11). F. Willett (2004) notes on head M11that the ‘striations behind both ears are not parallel to those on the face; they are very much shallower and very irregular; clearly they were not cut in the wax but in the metal after casting’.

d) That ‘the head is noticeably heavier for its size than the other heads […] and this is associated with an extraordinary variation of thickness which is entirely uncharacteristic of cire perdue work; in this process of which all the other heads are exceptionally fine examples, the internal surfaces bears a very close relation to the outer surface and the thickness may vary by as little as 1 mm over the whole face’ (ibid.: 3-4).

16This is not so, the Olokun head weighs 5 kg and NCMM Ife 11 for example, weighs 9 kg. In fact the weight of the head is quite close to the other crested heads, thus the rather smaller head, NCCM 79.R.11, weighs 4.2 kg, and the BM crowned head weighs 5.2 kg. The interior of the Olokun head follows the outer surface closely and the range of thickness is much the same as on the other heads, especially the 79.R.11 head and the BM head, which both seem stylistically closest to the Olokun head. Variation in thickness is in fact a feature of the direct cire perdue process.

e) ‘Inside the lower border of the neck […], is an area of heavy chiselling, quite clearly carried out in this specimen after casting [… from inside the neck]. The maker of the original head would have had no reason to apply this drastic method of correction […]. The maker of a reproduction, on the other hand, probably working from a plaster pattern, might well find himself with too thick an edge to allow of the copy passing as a reasonable facsimile, and his only recourse would be to cut down the metal near the edge’ (ibid.: 4).

17This is completely erroneous. The cut marks are casts of cut marks (figs. 2 & 7) and similar marks occur just inside the head of the small crowned head, NCCM 79.R.11). Far from being part of the post-cast working on a putative after-cast, they both represent work carried out on the original wax.

Figure 7 – A: Casts of chisel marks cut in the original wax of the neck of the Olokun head; B: Detail of marks showing their diffuse nature (cf. fig. 2)

Agrandir Original (jpeg, 760k)

f) ‘There is a complete absence on the surface of this head of sharp angles and edges such as are quite normal in cire perdue productions […], but which tend to disappear in making plaster casts’ (ibid.).

18There is no evidence for this statement at all; the head is a superb piece of cire perdue casting, the absence of sharpness is due to the apparent lack of work after casting as well as subsequent accretions of corrosion and soil. Thus it is just a restatement of point a).

g) ‘The headdress of the piece presents the appearance of being coated with an accretion of substances […] such as was in fact present in a thin layer over the whole head before it was cleaned in the Museum Laboratory; the British Museum specimen […] also presents a somewhat similar aspect, because it has not yet been cleaned and in addition to substantial traces of paint, a certain amount of lateritic soil still adheres to the headdress […]. However, no extraneous matter now remains on the ‘Olokun’ head, and the accretions, particularly noticeable at the base of the crest behind the rosette […], are entirely in bronze’ (ibid.: 4-5).

19The headdress is indeed still covered with accretions (fig. 8), as are the headdresses of the other two crowned heads. The next statement suggests that the accretions are actually casts of accretions and if so this would indeed be strong evidence that the head was an after-cast, the plaster moulds having picked up the surface dirt and other accretions. Detailed probing particularly in the region at the base of the crest (fig. 2) established that it had never been properly cleaned after casting (fig. 9). Thus far from being a copy in brass of the uncleaned surface, many of the accretions on the headdress really are of remnant mould material or core (grey), soil (red) and corrosion.

Figure 8 – Detail of the surface of the headdress on the Olokun head (cf. fig. 2)

Agrandir Original (jpeg, 2,1M)

The diffuse appearance of the casting is because it was never fully cleaned after casting, with some of the original mould material still present in the interstices. The remains of offerings and ancient iron oxide paint together with accumulated accretions of dirt and corrosion combine to obscure the detail ()

Figure 9 – Base of the crest (cf. fig. 2) where Fagg and Underwood erroneously claimed that the corrosion and dirt were in fact casts, thus demonstrating that the head must be an after-cast

Agrandir Original (jpeg, 1,3M)

17  W. Fagg and L. Underwood also explored alternatives, stating that they ‘cannot be explained by acc (...)

20However, at the base of the crest behind the headdress (fig. 2) there are small fins standing proud where metal had run into cracks in the mould and had not been removed (fig. 10). Also there are areas of brass which do have the appearance of accretions lying over the original decorated surface of the crest, which seems to have originally continued beneath the accretions. W. Fagg and L. Underwood stated this must mean that the present head is an after-cast of an earlier head¹⁷. However there are alternative explanations. It could be that if the mould was made with thin pancakes of clay around the wax (see above in the technology section) then possibly they were not pushed down to the very bottom of the crest leaving a small space which on casting filled with brass. Or the mould could have been prone to scabbing. This is a well known casting defect where small pieces of the thin surface layer of the mould become detached during the casting and the space they occupied fills with metal standing proud of the original intended surface.

Figure 10 – Detail of metal fins at the base of the crest (left and bottom left hand side; cf. fig. 2)

Agrandir Original (jpeg, 1,6M)

These had run into a crack in the mould during casting and had not been removed in the subsequent cleaning operations

h) W. Fagg and L. Underwood indicate in a figure ‘two depressions, one at each side of the crest, at the rear of the top border of the fillet which passes across the forehead forming the main part of the headdress. When viewed from a somewhat lower position [than they are illustrated in the photograph] they are seen as roughly semicircular fractures about 1 inch in diameter (i.e. along the top edge) and extending downwards about ¾-inch; they are concave to a depth of about ¼-inch and at the deepest point in centre of each depression are the rusty remains of an iron nail, the head of which is just visible as a [rusty] patch of oxidization in the ornament on the front surface of the fillet […]. The nature of the ‘fractures’ excludes the possibility that they occurred in the present bronze, or indeed in any bronze, and in any case they exhibit what appears to be a cast surface. They are in fact typical of fractures in plaster or terra cotta and are inconsistent with the molecular structure of metal. We believe that whoever had the replica made, being aware of the importance of including the iron nails, had holes bored to take them in the terra cotta or plaster pattern, so that they could be removed from the holes at the next stage and placed in the sand mould to be picked up in the same position in the bronze cast. In these two cases, however, the plaster broke away when the holes were half bored, and owing to carelessness was not made good before the bronze casting was made from it’ (ibid.: 5).

21The explanation given for these features (fig. 2) is unconvincing; surely if it was felt necessary to introduce iron nails this would have been done at the next stage of the process when the plaster had been removed, either into the sand if this was a sand casting or through the wax if this was a cire perdue wax casting. However, judicious examination shows that the irregularities on the surface are made up of dirt, corrosion and refractory clay (fig. 11) and thus could not have come about as described by W. Fagg and L. Underwood. It seems more probable from their position that they were once the sites for runners (cf. p. 30).

Figure 11 – Detail of the surface of the proper right depression behind the headdress (cf. fig. 2) still covered by refractory material showing that it cannot be the cut away remains of a runner or riser from the present casting

$Figure 11 – Detail of the surface of the proper right depression behind the headdress (cf. fig. 2) still covered by refractory material showing that it cannot be the cut away remains of a runner or riser from the present casting$

Agrandir Original (jpeg, 1,7M)

i) ‘The runners or feeds seem to have been in a position nearer to that adopted in the modern sand-casting method than to that employed by the bronze founders of Ife, who placed them externally or very close to the top edges in all the other extant heads […;] the manner in which the jets have been removed by chiselling is also noticeably different from that seen on the other heads’ (ibid.).

22It is true that none of the other heads have evidence of runners inside, but the position of the runners in the present casting (fig. 2) seems to have been dictated, not to follow sand casting practice, but rather by the changes necessitated by having to reposition the runners as described in point h above and in the examination section (p. 30). The second point, that the method of removing the runners by chiselling is unusual, is incorrect. This was the usual method by which the channels were removed on all the Ife heads (fig. 12). The difference is the very fresh condition of the chisel cuts, preserved inside the head (fig. 13).

Figure 12 – Diffuse chisel marks where a runner on the outside of the head has been removed (cf. fig. 2)

Agrandir Original (jpeg, 1,7M)

Figure 13 – Fresh chisel cuts where one of the two runners inside the head (cf. fig. 2) was detached after the casting (cf. the diffuse marks where a riser channels on the outside of the head was removed) (cf. fig. 12)

Agrandir Original (jpeg, 1,3M)

j) W. Fagg and L. Underwood claim ’we have detected traces of the seams typical of piece moulding (and of course inconsistent with the wax method as practiced at Ife) on the outer surface about the left ear (see Plate B). There is a faint suggestion of a joint running from near the small hole at the end of the fillet, along the anterior edge of the ear (belonging to what would almost certainly have been made as a separate piece when making a plaster cast or a sand casting, and there is some evidence of the use of a European file to remove traces of the seam of a sand cast behind the top of the ear. We cannot be certain which of these traces arose at the plaster stage and which in the final casting.’

23They went further in the caption to their Plate B and described the putative seam as continuing down below the ear, and a line can be seen on both the plate and on the present head, together with a second shorter line to the left (fig. 14). W. Fagg and L. Underwood believed the first line was indicative of the join where two of the putative plaster positives were pushed into the sand, the second shorter line was left unexplained. In fact they are both casts of incised lines, the first running for a considerable length in the narrow space between two scarification lines and thus the lines must have been on the original wax. They are probably the original setting out lines for the scarification channels. In fact whichever way the mould was constructed the mould lines and the scarification would not have respected one another, whichever came second would have gone through the first.

Figure 14 – A (cf. fig. 2): proper left side of the head showing scarification with extra line running down. B: detail of the line and also a second shorter line of scarification to the left

Agrandir Original (jpeg, 1,7M)

The first line was claimed by W. Fagg and L. Underwood to be a mould line on their postulated sand casting, the second line was not mentioned. They are clearly contemporary with the other scarification lines, mould lines would have run through them

24Point j continues:

‘The interior, particularly at the back, shows more marked ridges, which may well represent unremoved seams, but it might possibly be explained also by the contraction of the core in heating it to burn out the wax’ (ibid.).

25The first statement suggests that W. Fagg and L. Underwood believed that somehow a core was shaped from the assembled plaster pieces but these would have been negatives of the outside and thus of no use in producing a core. The ridges or fins described by them are typical of the cracks which develop in a clay core on firing and fill with metal on casting and as such are to be seen inside many of the Ife heads.

k) ‘The fracture which has removed a quadrilateral fragment from the upper border at the back of the head behind the right ear [… figs. 15 & 2] has no sharp edges and appears to present not the torn, almost fibrous appearance of bronze fractures such as one on the British Museum specimen, but rather the typical softened cast surface: if so, the reproduction must have been made since 1910, when Frobenius says that this piece was chipped off while the head was being exhumed. We must, however, record our surprise that there is no turning-in of the broken edge, such as is characteristic of the damage done, mainly in ancient times but partly during excavation, to most of the other heads; the form of the fracture seems quite consistent with a break in plaster, clay, or other friable or brittle material’ (ibid.).

Figure 15 – The 1910 break on the proper right hand side of the head (cf. fig. 2) showing the rather diffuse nature of the edge suggestive of it being the cast of a break, but is in fact due to the extensive corrosion of the metal (cf. fig. 16)

Agrandir Original (jpeg, 1,6M)

26These observations are broadly correct, although not the inferences drawn from them. If the present rather diffuse break really was the cast of the break made in 1910 then this would be the single most decisive piece of evidence that the present head is a recent after-cast. Heavy probing revealed that the metal at the break is heavily corroded. During the probing a small cube of metal (approx. 1 mm³) was removed and it was immediately apparent when examined by a low power microscope that it and the exposed surface of the break were massively corroded, with small islands of yellow brass metal remaining in the centre of the grains surrounded by red cuprite corrosion product (fig. 16 and appendix 2). This satisfactorily explains why the appearance is rather diffuse and also provides the explanation why there ‘is no turning-in’. Deeply penetrating intergranular corrosion had taken all the strength out of the metal such that it could not bend when hit, but instead broke like the friable material it had become.

Figure 16 – Surface of the small fragment which became detached from the edge of the 1910 break (cf. fig. 15) showing massive intergranular corrosion, particularly the red cuprite and the yellow of the brass metal (width of field of view c. 2 mm)

Agrandir Original (jpeg, 1,6M)

l) ‘The rough, lumpy projections on the interior surface behind the eyes and mouth again suggest a sand casting, and appear in none of the other heads. A clumsy attempt seems to have been made to suggest the modelling of the inner surface of the eyes in close relation to the outer surface, as in all the other heads, and presumably the original of this one. If this were the original, there would have been no point in modelling the core even to this extent, given that the bronze were to be so thick, as it in fact is. The irregular excrescences […] may be due to an attempt to make a cast from the incrusted inner surface of the original, or less probably, falling away of parts of the sand core before or during the pouring of the metal’ (ibid.).

27Given that the standard procedure for the other heads was to carefully model the core (see above) the occurrence of the same features on the Olokun head would seem to be a point in favour of its being original rather than against. The ‘projections’ and the ‘irregular excrescences’ (fig. 17) which should have been of brass according to W. Fagg and L. Underwood’s interpretation turned out on careful probing to be largely made of core material, part of which formed the sample for petrographic examination, reported in the Appendices and the TL sample.

Figure 17 – Interior of the head showing the remaining core material behind the eyes (cf. fig. 2)

Agrandir Original (jpeg, 1,2M)

m) ‘Samples of the blackened core material taken from behind the lips and headdress of this head appear to differ from samples from the other bronzes in that the granules are much larger; this difference, again, is consistent with the sand process rather than cire perdue’ (ibid.).

28Petrographic examination of the core material (see appendix 3) shows that the blackening is due to charcoal. The particle sizes of the quartz etc were no different than in the comparative core material taken from the British Museum head. The core material used in sand casting usually consists mainly of sand with additions of sticky materials to make it bind together more satisfactorily, whereas the core in the head is of clay tempered with some sand and charcoal. It should be noted that core material should be coarse to allow gas absorption during casting by any method, rather than specifically related to sand casting.

n) ‘The patina of the whole specimen is exceptionally thin […], and its very uniformity gives it a somewhat artificial look, typical of artificial patination’ (ibid.).

29The situation is more complex than W. Fagg and L. Underwood can have known. Detailed microscopic examination backed up by metallographic examination of a polished section and X-ray diffraction and Raman spectroscopy analysis have shown that this is a deep-seated natural corrosion (cf. infra, fig. 18 and appendices 1 & 2). Furthermore in many places the appearance of granularity probably results from the waxing treatment given to an inadequately cleaned surface. In addition some of the minerals gave unusual XRD patterns, which maybe the result of the relatively recent heating revealed by the TL determination. There are also some synthetic green pigments on the surface of the natural patina.

Figure 18 – Taper section viewed with polarizing filters, showing developed cuprite growth at the grain boundaries. Note also the large grain size similar to those observed on some of the other heads examined for Moss in 1948. (Width of field of view c. 4 mm)

Agrandir Original (jpeg, 1,8M)

o) A small fragment of bright uncorroded iron approximately one quarter by one eighth of an inch was observed imbedded in the corrosion slightly to the left of centre of the back of the neck about half an inch from the edge. ‘The surface of this fragment is smooth, bright and fresh-looking, and it is perfectly clear that no oxidization has taken place. The eight nails inset around the upper edge of the head, on the other hand, are heavily oxidized […], probably by the use of acid, as one would expect in a faithful reproduction’ (ibid.).

30A careful search using both the microscope and a magnet failed to find any trace of this iron. It may have been contamination in the wax coating and became detached during the past sixty years.

A Technical and Scientific Re-assessment of the Olokun Head

31The head was examined in detail with a binocular microscope augmented by careful probing of the surface deposits. X-radiography did not reveal anything that was not visible on the surface. The patina was sampled for examination by X-ray diffraction (XRD) (appendix 1) and Raman spectroscopy (appendix 2). A taper section was polished on the base of the neck for metallographic examination (appendix 1). Samples were taken of residual core material for petrographic examination and scanning electron microscopy with energy dispersive microanalysis (appendix 3) as well as for thermoluminescence dating. The various compositional analyses of the metal made previously are also reconsidered.

Examination

32Hollow cire perdue casting is the method of choice for producing large complex castings such as the Olokun head. It is also capable of picking up and reproducing quite fine detail from the mould. A distinguishing feature of lost wax castings is the general absence of mould lines, features that are present on castings made by other methods. The previous section has established that there are no features associated with sand or piece mould casting, such as the main join line where the halves of the sand mould would have met or subsidiary mould lines where separate details such as the ears were added. This, together with the general high quality of the casting, and the presence of clay mould material surviving in the interstices of the casting, would strongly indicate that the head is most unlikely to have been made by sand casting, but rather to been made by the cire perdue process.

18  See Rinne & Friel (1975) for a detailed description of the production of an after-cast of fragment (...)
19  W. Fagg and L. Underwood did not attempt to explain how the core in their postulated sand casting (...)

33It is less easy to establish whether the head is a direct hollow cire perdue casting or an after-cast, that is, made from a plaster negative mould taken of an existing casting. From this plaster negative a positive wax would have been made to be cast by the indirect hollow cire perdue process (Mills and Gillespie 1969). Although the indirect hollow lost wax procedure can produce an after-cast which is very close to the original casting there are differences, and those relevant to the Olokun head will now be considered. To produce an after-cast the original head would have been moulded with plaster cut into sections as described by W. Fagg and L. Underwood¹⁸. These would then have been reassembled and coated on the inside with wax to create the wax positive. A major difference between the direct process and the indirect cire perdue casting for producing after-casts concerns the core. In a direct hollow cire perdue operation the wax is applied to the pre-modelled core, whereas with an indirect hollow cire perdue operation the core is built up inside the hollow wax reproduction. As discussed above the negative images of the eyes, etc show that the Ife cores must have been quite carefully shaped and smoothed, to all intents a slightly reduced version of the head that was to be fashioned in the wax. This would be very difficult to reproduce for an after-cast using the moulding techniques described above¹⁹.

34When making an indirect lost wax casting, after removing the outer plaster piece moulds any damage or mould lines on the outer surface of the wax could be smoothed out and the interior filled with a suitable core material. The wax would then be encased in mould material and casting carried out in much the same way as for the direct process already described in the Technology section above.

20  The 1913 English translation of Frobenius’ original German publication stated that the head was 14 (...)

35After-casts are invariably smaller than the originals from which they were moulded, with shrinkage of between 2 % and 4 % for copper alloys being typical. This is due to the shrinkage of the metal as it sets in the mould, and is a well known and unavoidable feature of after-casts (Craddock 2009: 72, 181). L. Frobenius recorded the height as being 35½ cm in his original German publication and this is virtually identical to the recent measurements made in the British Museum of 35 cm to within the precision of the readings²⁰.

36The plaster moulding of the original picks up all the surface detail, including any damage, and in the case of a once buried antiquity such as this head, any residual dirt and corrosion that had not been cleaned off. These residues will inevitably be translated into metal on the after-cast. W. Fagg and L. Underwood claimed in point g that such accretions were observable in the metal. As stated in the comments on g, re-examination suggest that most of the observed features on the headdress are real accretions not casts. There are some that could be casts and could be taken as evidence of after-casting although as stated above there are alternative explanations.

37Another feature on the head suggestive of after-casting are the two depressions behind the headdress (fig. 2). An alternative explanation to that advanced by W. Fagg and L. Underwood [their point h) see above] but which still regarded the head as an after cast would be to interpret them as the sites of the two runners of the putative original casting which had been chiselled away after the original cast. This would neatly explain the cut surfaces overlain with refractory clay material (fig. 11), the presence of which shows that they never could have been operative channels during the present casting. These contrast with the very obviously freshly cut surfaces of the runners inside the head immediately adjacent (fig. 2).

38However, there is a problem here, namely the remains of the iron armatures in the centre of each of the depressions. If these depressions are no more than the topography of the original runners picked up by the after-cast how can the iron armatures still be present? Additionally, removing the brass channels was usually quite a difficult task. The channels on all the other heads and the other channels on the Olokun head had been laboriously chiselled off flush with the surface of the head. In contrast two cuts were made deep into the body of the casting across a far greater area requiring a much greater effort than was necessary. Furthermore, the cuts are absolutely smooth with no evidence of the chisel marks that are prominent on all the other channel cut offs. This suggests that it was the wax that was cut rather than the brass. A possible explanation was that two original runners were indeed located in the two depressions but that just before firing there was some mishap, the mould was dropped or some such calamity and the wax rods and the surrounding mould broke. Following this the top of the mould and core was scraped away exposing the remains of broken wax rods which were easily cut away, leaving the ends of the iron armature. New wax rods were then put in place, without iron armatures, just behind and inside the head.

39It could be argued that the present runners were located inside the head to ensure that their remains were as inconspicuous as possible and did not interfere with any of the moulded features from the original head. However, if the head was made with the intention of deceit then it is surprising that no effort was made to disguise the chiselled remains of the interior channels. This could so easily have been achieved by grinding.

40Thus overall, the detailed examination of the head shows that it is a hollow cire perdue casting, and strongly suggests it is an original direct casting, rather than an after-cast made by the indirect process.

Metallography

41The taper section (appendix 1; figs. 2 & 18) showed that the metal was a casting with a large grain size similar to the unpublished taper sections prepared by A. Moss in 1948 on four of the other heads (cf. supra and note 11).

Corrosion and patina

21  Chrome green was used to patinate the plaster casts of the heads made at the British Museum in 194 (...)

42Detailed microscopic examination backed up by XRD and Raman spectroscopy has helped to resolve some of the problems regarding the patina. Although presently quite thin compared to that on some of the other heads it is visually very similar to that on the other two crested heads. The patina is firmly attached, being especially noticeable at the extremities of the casting where a synthetic patina could be expected to wear away quickly exposing the metal beneath. The patina did not fluoresce under ultra violet radiation and did not respond to a range of organic solvents used to test for the presence of an organic binder. XRD showed cuprite was present but the green corrosion could not be easily identified (appendix 1). Raman spectroscopy detected both cuprite and malachite natural corrosion products and in addition some very bright specks on the surface of the patina contained Prussian blue and chromium salts, a combination known commercially as chrome green (appendix 2)²¹.

43Both the taper section on the neck (fig. 18) and the break on the head (fig. 15) show extensive intergranular corrosion, very similar to that on the other heads sampled by Moss (see note 11), that could only have come about from prolonged burial (appendix 1). Many of the iron nails are now totally corroded in situ within the surrounding brass. It is not easy to understand how this could have been achieved artificially.

44This evidence of long burial is reinforced by the core material surviving in contact with the inside of the Olokun and BM heads which is heavily impregnated with calcium, copper and zinc salts (see appendix 3).

45The head has a deep stress crack running under the chin and similar stress cracks were observed on some of the other Ife heads (such as that on the lower jaw and neck of NCMM 38.1.8) which appear to be corrosion cracking rather than casting flaws.

46The location of the Olokun head between 1910 and 1934 is not known. If somehow, as has been suggested, L. Frobenius did succeed in having a copy made and buried in place of the real head back in 1910 it is most unlikely that such a deep and comprehensive corrosion could have developed in 24 years. It is also significant that the nature of the break on its re-excavation back in 1910 [point k)] strongly suggests that the metal was already heavily corroded. However the evidence of the composition (see below) would suggest that if it is a copy then the metal used only became available after 1938/9.

The Core

47Samples were taken of core material from the Olokun head and from the British Museum head for scanning electron microscopy and petrographic analysis (appendix 3). This showed both differences and similarities. In particular the Olokun head core contains appreciable quantities of charred material, and this explains the dark appearance noted by W. Fagg and L. Underwood. Also the sample from the Olokun head had clearly been exposed to considerable heat, bringing about its partial vitrification. This is likely to have been very localised as core material in other parts of the Olokun head did not seem to be vitrified.

48There are some broad similarities between the clays used for the cores of the Olokun and BM head (figs. 19A & B). Both contain quartz, feldspar, pyroxene, zircon and ilmenite, and they are iron-rich and non-calcareous, but they are not from identical sources. This is indicated by the presence of microfossils in the Olokun sample but not in the British Museum sample. The Olokun sample had higher contents of magnesium oxide and quartz inclusions in the matrix. In addition there was a higher incidence of mica in the British Museum head sample, which is very fine and might be regarded as a natural component of the matrix. Thus, there were no unusual minerals present that could have convincingly demonstrated a common source or suggested a particular geological region, but equally the suite of minerals present in both heads were sufficiently similar not to exclude a common source.

Figure 19 – A: Petrological thin section from the core of the Olokun head showing a vitrified fabric, with abundant poorly-sorted quartz and micritic calcareous fragments infilling the voids of the paste. (XPL, 5.4 mm field of view). B: Petrological thin section from the core of the British Museum head showing a non-calcareous and slightly micaceous fabric with poorly-sorted quartz and abundant opaques and iron oxides. (XPL, 0.87 mm field of view)

Agrandir Original (jpeg, 976k)

49Metal salts from the corrosion of adjacent copper alloy casting had penetrated both the core samples (appendix 3, Tables 3 and 4). In addition the Olokun head core was heavily impregnated with calcium salts presumably from the local burial environment, suggesting prolonged burial (fig. 20).

Figure 20 – SEM-EDX elemental X-ray map for calcium showing the penetration of calcium salts (red) into the core material in the Olokun head suggestive of prolonged burial

Agrandir Original (jpeg, 928k)

50The Olokun head core sample was rich in the charred remains of woody plants, most likely the fines from the charcoal used in the casting operation. Scanning electron microscopy examination of the numerous charcoal fragments indicate that at least three different woody taxa are represented, one of which is Tropical African in origin and two of which may or may not be (figs. 21 & 22 and appendix 4).

Figure 21 – VP-SEM image of two charcoal fragments in the Olokun Head core sample. The fragment on the right is a TLS (Tangential Longitudinal Section) with exclusively uniseriate rays

Agrandir Original (jpeg, 544k)

Figure 22 – Micrograph of Olokun Head showing charcoal fragment in TLS with rays 4 (or more) cells wide (0.87 mm field of view)

Agrandir Original (jpeg, 1,2M)

51Thus overall the examination of the core strongly suggests the casting was made in West Africa. The addition of organic material either as plant fibre or as dung to the clays making up the core for West African bronze castings was discussed by D. Williams (1974: 189).

22  Specifically the malachite should have been decomposed at temperatures above about 200⁰C (Mellor 1 (...)

52A portion of the core sample from the Olokun head was submitted to Oxford Authentication for thermoluminescence testing (Sample n° N1 10j97). Their conclusion was that it had ‘last been fired less than 200 years ago. … and thus the date could lie between the 19^th and 20^th centuries but not beyond these dates’. This is a disturbing result. It could be argued that the antique head had been heated either accidentally or as part of a ceremony since its discovery at some time in the mid 19^th century. There is evidence for glass bead making in the 19^th century in the Olokun grove, and it is possible the head could have become accidentally locally heated. If the antique patinated head had been heated at temperatures in the region of 500^°C high enough to remove the TL then this should have made visible changes to the patina²². It is possible that if this putative heating had occurred in the 19^th century then some decades of reburial could have elapsed before the uncovering in 1910 and during which the malachite could have partially reformed.

Composition

23  Major analytical programmes have been undertaken on West African copper alloy castings, usually in (...)

53As already intimated in the introduction the Olokun head has been analysed on several occasions since 1949 (Tables 1 & 2), usually together with other West African copper alloy castings²³. The composition provides some of the strongest evidence that the Olokun head is likely to be the original. Chemical analysis of the head in the 1940s showed that it is of leaded brass, very similar to that used in several of the other Ife heads, as noted by A. Moss (1949), and qualitative emission spectrographic analysis suggested that the trace elements were also likely to be similar (Table 1; Barker 1965). Further analyses (Werner & Willett 1975; Craddock & Picton 1986) confirmed the percentages of the alloying elements and quantified the trace elements showing that the composition of the leaded brass used for the majority of the heads and for the Olokun head were quite similar. More conclusive evidence that the Olokun head was made from the same stock of metal as the other Ife heads was provided by the lead isotope analyses carried out in the 1990s and reported by F. Willett and E. Sayre (2006) which showed that ‘Interestingly head 16, the so-called Olokun head also matches this group (that is ten of the heads from the Wunmonije Compound)’.

54If the head is indeed a copy this raises interesting questions as to the source of the metal. Up until the time that the analyses were made by A. Moss (1949), it was assumed that the so-called bronzes were just that, an alloy of copper and tin. Thus anyone making a copy would have chosen that alloy. Following the first publication of the composition of the Ife heads including the Olokun head W. Fagg and L. Underwood in their response offered two explanations.

24  These can have very strange compositions with several percent of arsenic or antimony in addition t (...)
25  The bronze was a massive neck ring that carried the marks of recent sampling. This raised the inte (...)

55The first was that random old Nigerian scrap copper alloys could have been used including the ubiquitous manillas²⁴. However, subsequent analyses of Nigerian copper alloy art metalwork generally, and more specifically, the spectrographic analyses made of an eclectic group of seven copper alloy items from Ife itself (Willett 1964) make that hypothesis unlikely. Five of the latter were of copper alloy with a few percent of lead, zinc, tin and arsenic, one was of bronze and just one, from a previously unrecognised fragment of a head was of the same leaded brass alloy as the other heads²⁵. The Olokun head is so close in composition to the leaded brass of the other Ife heads that it would have been an extraordinary coincidence if such a random mixture of scrap had hit on the right composition across a range of major and minor elements.

26  In practice most metal forgeries even now use contemporary modern alloys that are often anachronis (...)

56W. Fagg and L. Underwood’s other suggestion was that the head or the detached fragment that went with L. Frobenius to Europe could have been analysed to determine the composition of the metal to be used in the copy. This scenario does require a very comprehensive analysis, probably at this date necessitating a combination of wet chemical analysis for the major alloying metals and emission spectrographic analysis for the trace elements (although spectrographic analysis was not really available as an analytical tool until the mid 1930s). It would have been necessary to analyse for a very wide range of potential alloying and trace elements, none of which could have been assumed a priori not to be present, such that subsequent analysis by more sophisticated techniques have not been able to detect any inconsistencies in the alloy as made up. This scenario infers an attention to detail that borders on the obsessive and one that is unparalleled in the copying of metal works of art, either legitimately or as forgeries (at least amongst those that have been uncovered, that is)²⁶. The coincidence of the lead isotope data, the significance of which no one had any cognisance of before the 1960s, makes it virtually certain that the Olokun head is of the same stock of metal as the Wunmonije heads.

57Thus if the present Olokun head is a copy made in the 20^th century then the only feasible source of metal is from amongst the Ife heads. That is, brass of the same composition from the same group, possibly damaged brass heads were melted down to form the alloy for the present casting.

58However, in order to make a cire perdue casting complete with runners, risers and the sprue at least twice the weight of metal of the final casting is required. Thus a minimum of two heads worth of metal would be required and as far as can be ascertained such a quantity of metal was not available before the Wunmonije heads were discovered in 1938-1939. This means that if the head is a copy then it must have been made at some time between 1939 and 1945 when L. Underwood first cast doubts upon it.

Table 1 – Chemical and emission spectrographic analyses (given as weight %) of the Ife heads including the Olokun head (n°. 16) from Barker (1965); *Including gold 2.4%; n.d.: not detected)

BM Sample n°	Sn	Pb	Cu	Fe	Zn	Total
1	0.9	15.9	68.8	0.9	11.6	98.1
2	0.6	13.8	71.2	0.7	13.3	99.6
3	0.2	1.2	96.8	0.2	n.d.	98.4
4	0.2	n.d.	99.7	0.1	n.d.	100.0
5	0.1	0.1	99.2	0.1	n.d.	99.4
6	0.05	0.3	99.2	0.2	n.d.	99.7
7	2.1	3.6	74.0	0.2	15.3	99.6*
8	0.7	11.1	71.8	0.6	15.0	99.2
9	1.3	13.6	74.2	0.5	9.6	99.2
10	0.05	n.d.	99.1	0.1	n.d	99.2
11	1.3	11.3	74.1	0.8	11.9	99.4
12	1.i	9.9	77.5	0.6	9.3	99.4
13	n.d.	14.5	73.4	0.9	9.7	98.5
14	1.7	6.4	76.2	0.4	13.9	98.6
15	0.9	4.4	79.8	0.6	11.8	97.5
16	1.0	11.0	74.2	0.6	12.3	99.1
17	0.05	0.0	99.4	0.1	n.d.	99.5
18	0.8	11.4	73.4	0.6	3.1	99.3

Table 2 – Composition (given as weight %) of the Olokun head and the British Museum head (Craddock & Picton 1986). Atomic Absorption analysis (see Hughes et al. 1976 for details of the methodology). Manganese and cadmium were also sought but were below the detection limit of 0.005% for all the elements analysed.

Sample	Cu	Zn	Pb	Sn	Ag	Fe	Sb	Ni	Au	Co	As	Bi
Olokun head	78.3	12.82	7.78	0.47		0.38	0.19	0.065	0.002	0.009	0.39	0.053
BM head (neck)	71.5	15.50	11.80	0.80	0.065	0.74	0.25	0.080			0.550	0.003
BM head (Headdress)	70.0	17.40	10.80	0.80	0.095	0.34	0.20	0.035			0.50	0.002

Conclusion

59This investigation has taken a number of approaches, both observational and analytical to produce a wide range of evidence relating to the likely age of the Olokun head. It is perhaps salutary to reflect that if the head had been metallographically examined and analysed by A. Moss along with the other heads in 1948 its authenticity would have been conclusively re-established. Conversely, if the head had been amongst those selected for TL dating as part of the programme carried out by Stuart Fleming on core material from a number of West African copper alloy castings (Willett & Fleming 1976) its suspected recent origin would have been confirmed. The various tests carried out for this re-examination produced apparently conflicting results. In the real world certainty is illusory, data, especially visual, can often be interpreted in a number of ways; but also in the real world there can only have been one sequence of events. The task here is to try and evaluate which is the more probable.

60The examination unequivocally concluded that the head is a cire perdue production. It is most likely to be an original direct cire perdue casting but the possibility that it is an indirect cire perdue casting cannot be entirely excluded.

61The method used to cast the Olokun head and the composition of the metal is very similar to that of the Ife heads discovered subsequently. If the head was made in the 20^th century then the necessary supply of metal from the other Ife heads would only have become available late in 1938 and thus the head would have had to have been made at some time between then and 1945. This scenario is rendered most unlikely by the highly corroded state of the metal. The deep intergranular corrosion observed on both the neck and the top of the head, apparently already present when the head was re-excavated in 1910, the presence and penetration of metal salts into the core material, and the totally corroded iron nails set in the brass combine to support prolonged burial. The charred plant remains in the core include woody taxa native to tropical Africa which strongly suggest that the head was cast in West Africa. However the sophisticated indirect cire perdue technology necessary to produce an after-cast was not being practised in Nigeria at that time.

62Against these arguments are the results of the TL test which suggest that the head was strongly heated at some time during the last two centuries. If that single determination on a small piece of vitrified clay is correct, then the most likely scenario to explain the date and the apparently unaffected state of the corrosion is that the head was locally heated shortly after its discovery in the mid 19^th century followed by reburial during which the patina could reform.

63Thus the combined evidence suggest that the present head is that shown to L. Frobenius in 1910 and that it is an original direct hollow lost wax casting of some antiquity, most likely coeval with the brass heads discovered at the Wunmonije Compound in 1938.

Table 3 – Olokun head (HO2) core sample: SEM-EDX results of four bulk analyses at 100x, with average and standard deviation (results are reported as normalised % oxides)

	Olokun bulk analysis 1^st 100x	2^nd bulk analysis 100x	3^rd bulk analysis 100x	4^th bulk analysis 100x	Average	Standard deviation
Na₂O	0.9	0.8	1.1	1.0	0.9	0.1
MgO	1.4	1.1	1.6	1.4	1.4	0.2
Al₂O₃	10.7	9.7	12.6	13.2	11.5	1.6
SiO₂	54.3	60.4	36.6	51.2	50.6	10.1
P₂O₅	1.1	0.8	1.9	1.0	1.2	0.5
SO₃	0.0	0.2	0.2	0.2	0.2	0.1
K2O	0.3	0.2	0.4	0.6	0.4	0.2
CaO	16.5	13.3	22.1	17.3	17.3	3.6
TiO₂	1.7	1.1	4.9	1.2	2.2	1.8
MnO	0.0	0.0	0.2	0.0	0.0	0.1
FeO	4.6	3.8	7.9	4.3	5.1	1.9
CuO	0.9	1.0	1.2	0.8	1.0	0.2
ZnO	5.8	6.1	6.6	6.0	6.1	0.3
PbO	2.0	1.3	2.9	1.9	2.0	0.7

Table 4 – British Museum core sample (BM1): compositional data of SEM-EDX of two bulk analyses at 170x on a third fragment with average and standard deviation, and results of a bulk analysis of a fourth fragment at 80x. Results are reported as normalised % oxides

	BM 1 x170 fabric 1^starea on small fragment	BM 1 x170 fabric 2^nd area on same small fragment	Average	Standard deviation	BM 1 bulk x80 on clay fragment
Na₂O	0.2	0.0	0.1	0.2	0.0
MgO	0.4	0.2	0.3	0.1	0.2
Al₂O₃	31.1	16.0	23.6	10.7	25.3
SiO₂	45.9	26.9	36.4	13.5	31.9
P₂O₅	0.6	1.2	0.9	0.4	1.5
SO₃	1.4	0.6	1.0	0.5	0.3
K₂O	0.5	0.3	0.4	0.1	0.1
CaO	0.6	0.5	0.5	0.1	0.3
TiO₂	1.0	2.1	1.6	0.8	0.9
MnO	0.0	0.0	0.0	0.0	0.3
FeO	14.2	42.9	28.5	20.3	29.1
CuO	1.2	3.1	2.2	1.4	7.6
ZnO	0.4	1.0	0.7	0.5	0.4
PbO	2.6	5.3	3.9	1.9	2.2

Haut de page

Bibliographie

Barker H. (1965) – Examination of the Ife Bronze Heads. Man, 65, p. 23-4.

Cartwright C.R. (2005) – The Bronze Age wooden tomb furniture from Jericho; the microscopical reconstruction of a distinctive carpentry tradition, Palestine Exploration Quarterly, 137, p. 99-138.

Chastain S.D. (2004) – A Sand Casting Manual for the Small Foundry. 2 vols. Chastain, Jacksonville Fld.

Courty M. A., Goldberg P. & Macphail R. (1989) – Soils and Micromorphology in Archaeology. Manuals in Archaeology, Cambridge University Press, Cambridge.

Craddock P. (2009) – Scientific Investigation of Copies, Fakes and Forgeries. Butterworth Heinemann, Oxford.

Craddock P.T. & Picton J. (1986) – Medieval copper alloy production and West African bronze analyses - part II. Archaeometry, 28, 1, p. 3-32.

Craddock P.T. & Hook D.R. (1995) – Copper to Africa: Evidence for the international trade in metal with Africa. In: D.R. Hook & D.R.M. Gaimster (eds), The Scientific Study of Artefacts from Post-medieval Europe, British Museum Occasional Paper 109, London, p. 181-93.

Craddock P.T., Ambers J., Hook D.R., Farquhar R.M., Chikwendu V.E., Umeji A.C. & Shaw T. (1997) – Metal Sources and the bronzes from Igbo-Ukwu, Nigeria. Journal of Field Archaeology, 24, p. 405-29.

Drewal H.J. & Schildkrout E. (2010) – Kingdom of Ife. British Museum Press, London.

Fagg W.B. & Underwood L. (1949) – An examination of the so-called Olokun head from Ife, Nigeria. Man, 49, p. 1-9.

Frobenius L. (1913) – The Voice of Africa I. Hutchinson, London.

Goucher C.L, Teilhet J.H., Wilson K.R. & Chow T.J.
(1978) – Lead Isotope Analyses and Possible Metal Sources for Nigerian “Bronzes”. In: G.F. Carter, Archaeological Chemistry II, American Chemical Society, Westerville, Ohio, p. 278-92.

Hughes M.J., Cowell M.R. & Craddock P.T. (1976) – Atomic Absorption Techniques in Archaeology. Archaeometry, 18, p. 18-36.

Joel E.C., Sayre E.V., Vocke R.D. & Willett F. (1995) – Stable lead isotope characterisation of various copper alloys used in West Africa: an interim report. Journal of the Historical Metallurgy Society, 29 (1), p. 25-33.

Johnson S. (1921) – A History of the Yorubas from the Earliest Times to the Beginning of the British Protectorate. Routledge, London.

Mellor J.W. (1967) – A Comprehensive Treatise on Inorganic and Theoretical Chemistry III. Longmans, London.

Mills J.W. & Gillespie M. (1969) – Studio Bronze Casting- lost wax. Maclaren and Sons, London.

Moss A.A. (1949) – Further light on the ‘Olokun’ head of Ife. Man, 49, p. 120.

Platte E. (2010) – Bronze Head from Ife. British Museum Press, London.

Read C.H. (1911) – ‘Atlantis’ rediscovered. Burlington Magazine, 18, p. 330-35.

Rinne D. & Friel J. (1975) – An account of the recreation of an ancient statue. The J. Paul Getty Museum, Malibu Ca.

Shaw T. (1970) – Igbo Ukwu. Faber & Faber, London.

Shaw T. (1978) – Nigeria. Thames and Hudson, London.

Spataro M. (2002) – The First Farming Communities of the Adriatic: Pottery Production and Circulation in the Early and Middle Neolithic. Società per la Preistoria e Protostoria della Regione Friuli-Venezia Giulia, Quaderno 9, Trieste.

Underwood L. (1949) – Bronzes of West Africa. Alec Tiranti, London.

Werner O. (1970) – Metallurgische Untersuchungen der Benin-Bronzen des Museums für Volkunde Berlin. Baessler Archiv., N.F., 18, p. 71-153.

Werner O. & Willett F. (1975) – The composition of brasses from Ife and Benin. Archaeometry, 17 (2), p. 141-56.

Wheeler E.A., Pearson R.G., LaPasha C.A., Zack T. & Hatley W. (1986) – Computer-aided wood identification. North Carolina Agriculture Research Service Bulletin 474.

Wheeler E.A., Baas P. & Gasson P.E. (1989) – IAWA list of microscopic features for hardwood identification. IAWA Bulletin 10, p. 219-332.

Willett F. (1959) – Bronze figures from Ita Yemoo, Ife, Nigeria. Man, 59, p. 308-11.

Willett F. (1964) – Spectrographic analysis of Nigerian bronzes. Archaeometry, 7, p. 81-3.

Willett F. (1967) – Ife in the History of West African Sculpture. Thames and Hudson, London.

Willett F. (1976) – True or False? The false dichotomy. African Arts, 9 (3), p. 8-14.

Willett F. (2004) – The Art of Ife: A Descriptive Catalogue and Database (CD-ROM). Hunterian Museum and Art Gallery, University of Glasgow, Glasgow.

Willett F. & Fleming S.F. (1976) – A catalogue of important Nigerian copper-alloy castings dated by thermoluminesence. Archaeometry, 18 (2), p. 135-46.

Willett F. & Sayre E.Y. (2006) – Lead isotopes in West African copper alloys. Journal of African Archaeology, 4 (1), p. 55-90.

Williams D. (1974) – Icon and Image. Allen Lane, London.

Haut de page

Annexe

Appendix 1: X-ray diffraction (XRD) analysis and metallographic examination (S. La Niece)

A small sample of red material was taken from the top edge of the head at the back. XRD analysis was carried out using a Debye Scherer powder diffraction camera and an X-ray tube with copper target. The diffraction pattern was recorded on film 17574. The pattern is a good match with that of cuprite (Cu₂O) as for the standard IICD pattern number 5-0667.

Taking a piece of metal for full metallographic examination was not possible without causing unacceptable damage to the head. Instead, a small area on the bottom edge of the neck was abraded flat on SiC papers of decreasing grit size down to #1200 grit. This process is not able to produce the quality of polished section that would be expected of a mounted specimen. The taper section was examined using a Zeiss Axiovert 100A inverted stage metallographic microscope.

The microstructure of the metal of the head, as revealed in the as-polished state showed large equiaxed grains with corrosion at the grain boundaries. Polarizing filters identified the grain boundary corrosion as cuprite (Cu₂O) (fig. 18). Cuprite was also identified by XRD on the top edge of the head. Brass alloys are relatively resistant to corrosion attack and the extent of the corrosion at the grain boundaries seen here is unlikely to be the result of artificial patination of a fresh casting, and has the appearance of metal which has undergone corrosion over a number of decades.

Appendix 2: Raman Spectroscopy (J. Ambers)

Samples of green material taken from two areas on the head (one of the raised spheres on the ‘crown’ and the other behind the left ear) were analysed by Raman spectroscopy using a Dilor Infinity spectroscope with green (532 nm) and near infrared (785 nm) lasers. Samples were examined without any surface preparation, using a spot size of around 5 microns and powers of 0.5-2.5 mW at the sample. The material from the ‘crown’ was found to contain chrome yellow and Prussian blue, a mixture sometimes termed chrome green (see note 21). This is a synthetic pigment which was not manufactured until the 19^th century. The material from behind the ear is coloured by a mixture of malachite and cuprite. This is entirely consistent with corrosion products formed by natural processes. The ancient iron oxide paint visible in figure 8 was not detected.

Appendix 3: Petrographic and chemical analysis of the cores (M. Spataro)

The core deposit inside the hollow casting of the British Museum head (BM1) was sampled with a spatula. The sample was friable. The Olokun Head (HO2) sample was previously collected by Paul Craddock. Both fragments were consolidated in epoxy resin before being made into polished thin sections. These were examined using a polarising microscope (Leica DMRX) and by scanning electron microscopy-energy dispersive X-ray spectrometry (SEM-EDX) (Tables 3 and 4).

Multiple bulk SEM-EDX analyses were carried out on each sample at various magnifications (80x–170x, giving sample areas of between c.1.5 x 1.1 and 0.7 x 0.5 mm), using ‘Glass VP’ calibration. Fourteen elements (Fe, Al, Mn, Mg, Ti, Na, Si, P, S, K, Ca, Zn, Cu and Pb) were quantified and the results were converted into oxide percentages, which were normalised (oxygen by stoichiometry) to take into account the fact that oxygen and carbon are not measured, and semi-quantitative, because of clay porosity (for more detail, see Spataro 2002: chapter 2).

Petrographic analysis

The Olokun Head sample (HO2) shows a dark red and vitrified fabric, with some poorly-sorted sub-angular to sub-rounded quartz based on Courty et al. (1989: 68) – size range between 0.3 x 0.15 and 0.04 x 0.03 mm – feldspar, pyroxene, amphibole, occasional clay fragments, coarse ilmenite (?), occasional clay pellets, zircon, very rare and fine muscovite (fig. 19A left). Some residual organic structures (charcoal less than 1 mm long) are present. Abundant post-depositional material infilling the voids left by the burning out of organic matter is visible, including calcareous fragments.

The British Museum head sample (BM1) has a brown, non-calcareous and slightly micaceous fabric, with some sub-angular to sub-rounded quartz (size range between 0.2 x 0.15 and 0.03 x 0.03 mm), some muscovite mica with very fine lamellae, occasional zircon, feldspar, pyroxene, clay fragments, clay pellets containing sub-angular quartz, some coarse ilmenite (?), opaques and iron oxides (fig. 19B right). There is some post-depositional material (see SEM-EDX analyses below).

In thin section, the main differences between the two samples are:

HO2 is more highly fired than BM1,
organics and post-depositional calcareous fragments are present in HO2 only,
some muscovite is present in BM1, while it is very rare in HO2, and
coarser pyroxene is present in HO2.

On the other hand, a few features are common to both samples:

iron-rich fabric with poorly-sorted quartz,
coarse ilmenite,
presence of pyroxene, and
occasional clay pellets and pure clay fragments.

SEM-EDX analysis

The SEM examination of sample HO2 shows some microfossils (Globigerinidae [possibly], foraminifera). Four bulk analyses by SEM-EDX at 100x (an area of c.1.4 x 1.00 mm) were carried out on sample HO2. The results (Table 3) indicate a high silica content and variable calcium oxide, which is probably mainly post-depositional (as suggested by the optical petrography). The sample contains low concentrations of alumina, sodium and potassium oxide, and a very variable iron oxide content (3.8-7.9%; Table 3), low percentages of copper and lead, and zinc in high percentages (ca. 6%; Table 3). The presence of copper, zinc and lead suggests that contamination from the copper alloy took place during the casting process, or subsequently, as the metal corroded. The voids in the charcoal are filled with calcium salts (>50%), some aluminium-containing material (mainly clay), and copper, zinc and lead corrosion products.

Spot analyses were also carried out on minerals sparsely present in the fabric, identifying zircon and iron-titanium oxides, possibly ilmenite.

The BM-1 sample has high alumina and iron oxide contents. Iron is particularly high and might be partly post-depositional (although in thin section this is not apparent). Two fragments were analysed at 100x. A third fragment with a matrix which varies in colour was analysed in two different areas at 170x, which were very variable in composition, particularly iron content (14.2-42.9%; Table 4). A fourth fragment, analysed at 80x, has a very high iron oxide content (Table 4).

Elemental map composition

A larger area of each sample (37x; c.3 x 2.2mm) was analysed by SEM-EDX to create a map of its elemental composition, in order to be able to relate the compositional data to the clay fabric, mineral inclusions and post-depositional alterations.

Magnesium is found in low percentages throughout the HO2 sample. Aluminium is present throughout the sample, as expected because of the clay matrix, with some higher concentrations due to aluminosilicate clay minerals (e.g. rich in aluminium and iron). Silicon is abundant in mineral inclusions, mainly quartz and feldspar. Phosphorus is almost absent; potassium seems to be concentrated in one mineral, possibly one of the feldspathoid group (associated with aluminium and silicon). Calcium is present throughout the fabric, but is more abundant where, as a post-depositional product, it fills voids in the clay and the gaps around and within the charcoal (fig. 20). Titanium is concentrated in specific minerals and is mostly associated with iron, suggesting the presence of iron titanium oxides, such as ilmenite. Iron is present throughout the clay in moderate content.

Copper, zinc and lead are present throughout the clay fabric and are not associated with any of the inclusions; in particular, zinc, being a volatile element, has been absorbed by the clay to a high level (ca. 6% ZnO; Table 3), most likely during the casting process and/or through corrosion during burial. Copper is also dispersed throughout the matrix and it infills the voids left by the organics, again suggesting contamination from the alloy head, either during casting or burial.

The compositional elemental maps of BM1 show low contents of magnesium, mainly concentrated in a grain of pyroxene (also rich in calcium and silicon). Aluminium is scattered throughout the fabric of the sample, reflecting the aluminosilicate of the clay minerals. Silicon is widely present, with concentrations in minerals, mainly quartz and feldspar. Potassium is concentrated in very few minerals (feldspathoid group); calcium is concentrated in a mineral which is also rich in magnesium and silicon. Titanium and iron are concentrated in opaques (in thin section) such as ilmenite. Iron is more widely spread in the clay. Copper, lead and zinc are present in very low concentrations throughout the fabric.

Discussion and conclusions

Comparison of the two clay fabrics is complicated by the fragmentary nature of the samples and the likelihood that both clays have been contaminated by elements from the metal alloy of the heads and calcium salts during and/or after casting. Differences in their chemistry may thus reflect differences in the alloys of the heads and/or in the burial environments. There are some similarities between these clays; both contain quartz, feldspar, pyroxene, zircon and ilmenite, and they are iron-rich and non-calcareous, but the two clays used are not from identical sources. This is indicated by:

the presence of microfossils in HO2 and not in BM 1,
higher content of magnesium oxide in the matrix of HO2,
the presence of charcoal in HO2 and not in BM 1 (the higher incidence of quartz inclusions in HO2 (which may, however, have been added as temper) and the higher incidence of mica in BM 1, which is very fine and might be regarded as a natural component of the matrix.

The presence of metal corrosion products within the matrix of both core samples would appear to indicate that a significant period of time has elapsed since casting. The absence of post-depositional calcareous material in BM1 does not mean that it has not been buried, but suggests that it may not have been buried in the same location as HO2 or that the calcium content in the cores differs and so different solubilisation and redeposition processes have occurred. The composition of both samples, BM1 and HO2, is not diagnostic of a specific provenance.

Table 5 – Five Tropical African taxa with vestured pits, exclusively uniseriate rays and storied rays

Family	Subfamily	Genus	Species	Local Name(s)
Fabaceae	Caesalpinioideae	Brachystegia	cynometroides	Ekop; Naga
Fabaceae	Caesalpinioideae	Brachystegia	nigerica
Fabaceae	Caesalpinioideae	Cynometra	alexandri	Muhimi
Fabaceae	Papilionoideae	Pterocarpus	angolensis	African bloodwood; Muninga
Fabaceae	Papilionoideae	Pterocarpus	soyauxii	African padauk; padouk

Appendix 4: Charcoal identification (C.R. Cartwright)

Techniques of charcoal identification

Standard techniques of identification and terminology determined by the International Association of Wood Anatomists (IAWA) are usually adopted for the identification of modern wood (Wheeler et al. 1986; Wheeler et al. 1989). For each sample, the key features are compared with reference collection specimens and textual descriptions. This IAWA protocol may be applied to archaeological or historical wood, providing it is modified to accommodate the effects of the conditions of preservation such as charring or desiccation (Cartwright 2005). Normally, samples are prepared to expose transverse, radial longitudinal and tangential longitudinal sections or surfaces for identification (TS, RLS and TLS respectively). Charcoal is fractured by hand to expose fresh TS, RLS and TLS for examination. Fracturing rather than cutting exposes the clearest view of the cellular structure; cutting with a scalpel or microtome blade creates fine debris that penetrates and fills the cells, obscuring key anatomical structures. However, in the instance of the Olokun Head sample, which contained many charcoal fragments, the aspects of the anatomical structure available for examination were entirely serendipitous inasmuch as planes of the charcoal were randomly exposed when thin sectioned. Consequently, the usual controls could not be exercised over the search for diagnostic features. Furthermore, because the charcoal had not been deliberately fractured to expose precise TS, RLS and TLS, a number of the fragments straddle the RLS/TLS, which makes identification very problematic as the cells are distorted and misaligned. The Olokun Head polished thin section was examined using the Hitachi S-3700N variable pressure scanning electron microscope (VP-SEM).

Results of the VP-SEM examination of the charcoal in the Olokun Head

Despite the constraints mentioned above it was possible to discern that at least three woody taxa are represented by the charcoal fragments in the Olokun head sample and the following levels of identification are suggested:

The first category of fragments is characterised by these associated diagnostic features: vestured pits, exclusively uniseriate rays (fig. 21) and storied rays. Five Tropical African woody taxa meet these criteria, but the identification cannot be narrowed down any further as there are no other identifying features visible.
The second category of fragments only shows the features: rays 1-3 cells wide, and larger rays 4-10 cells wide (fig. 22). These are very common features in hardwoods worldwide (though obviously absent in softwoods). In Tropical Africa and in Europe there are at least 329 taxa with the criterion of rays 1-3 cells wide and at least 345 taxa with the feature of larger rays 4-10 cells wide, so in the absence of any other associated features, all that can be suggested is that there is (at least) one other hardwood present which is different from category 1.
The third category comprises thin lath-like organic voids or narrow radial files of plant cells. Insufficient diagnostic criteria are present in most cases to narrow the identification down further than to the monocotyledon palm family, Arecaceae.

Conclusions

At least three different woody taxa are represented by the numerous charcoal fragments in the clay core from the Olokun head, one of which is Tropical African in origin and two of which may or may not be.

Haut de page

Notes

1   The Olokun head, in common with other heads from Ife, has been numbered and re-numbered since it was initially recorded by Kenneth Murray in the 1940s. These original KM numbers have been superseded in more recent years by other museum accession numbers based on various year and date or consecutive number combinations. Some heads have received multiple numbers which continue to be used interchangeably and some appear to have been given sample numbers. In addition, during analysis at the British Museum in 1947-8 the Olokun and other Ife heads were allocated sample numbers (BM sample numbers) and in the most recent database of Ife material by Frank Willett in 2004 consecutive numbers were allocated according to material (M = metal, T = terracotta, S = stone). These are referred to in the text as Willett numbers. The Olokun head is currently numbered NCMM 38.1.2. However it is also referred to in the literature as KM number Ife 16, BM sample number 16 and Willett number M4. In this paper the most current NCMM numbers will be used within the text, although the BM sample numbers have been retained in Table 1.

2  In the conclusion of their 1949 paper Fagg and Underwood state that ‘We would make it very plain that it is not our purpose in any degree to apportion or imply responsibility, much less blame, for what took place’ (that is, the supposed forgery of the Olokun head). However this apparent disinterest slipped a few months later when in their response to Moss’ paper they suggested the true composition could have been determined through analysis of the fragment brought back by L. Frobenius, thus implicating him. Drewal and Schildkrout (2010: 26-7) also refer to L. Frobenius, suggesting that an experienced caster could have made a copy using Frobenius’ detailed notes and the fragment. Platte (2010: 42) goes even further and states that ‘It is quite possible that he took the head and left the replica behind’, but without attempting to explain how the replica could have been produced in Ife in 1910. L. Frobenius (1913: 103) had suggested that an electroform copy could be made. The head is most certainly not an electroform and the technology was not available in Nigeria at that time. Willett in 2004 M 4, although accepting Fagg and Underwood’s interpretation of their examination and their conclusion on the authenticity of the head, clearly states that ‘there is no evidence that Frobenius perpetrated the substitution’.

3  In Ife Olokun is often referred to as a goddess. Olokun is more frequently viewed as a god in Benin where he is viewed as the source of the Oba’s coral.

4  Everyone assumed that the heads were of bronze, that is, copper and tin, until Moss’ analyses (see Table 1) showed they were of copper or brass, that is, copper and zinc.

5  Note that location, discovery and subsequent history were very different from the other heads found later in the Wunmonije enclosure. Frank Willett (1967: 23-6) observed that artefacts from both areas showed signs of having been reburied.

6  The Colonial Secretary, Mr L.C. Moorhouse, sent Charles Partidge, the Resident at Ibadan, to investigate.

7  With sincere thanks to Yusuf Abdallah Usman, Director-General of the National Commission for Museums and Monuments, Nigeria.

8  The process of hollow cire perdue casting as described in the catalogue accompanying the British Museum exhibition (Drewal & Schildkrout 2010: 42) is incorrect.

9  This is described correctly by W. Fagg and L. Underwood (1949) but in his book on West African bronzes, L. Underwood (1949: 5) stated incorrectly that ‘a model is made in wax about a quarter of an inch thick’ but without explaining how such a free standing hollow wax was to be created, as the core was only mentioned at a much later stage. It seems probable that L. Underwood was subconsciously thinking of the indirect cire perdue technique (see below) with which he would have been more familiar. Beeswax alone is too soft and has to be mixed with between 10 and 20% of resin to harden it. Denis Williams (1974: 183-4) noted that in some parts of Nigeria latex obtained from the Euphoria kamerunica plant is used in place of wax. It is not known whether this material could have been used by the Ife craftsmen, although F. Willett (2004: 1.4) pointed out that Euphoria kamerunica did not flourish in the Ife area.

10  On head NCMM 79.R.11 (Willett 2004) the crest has broken and the iron armature is exposed. Testing with a magnet from the side showed strong ferromagnetism. Testing the complete crests on the BM and Olokun heads in the same way failed to detect the presence of an iron armature, strongly suggesting they are not present.

11  The micrographs were prepared by Dr. Godrichs of the Central Laboratory of the Belgian Museums. Moss produced an apparently unpublished metallographic report dated 1948 (but mentioned in Willett 1967: 31, 54). Three of the four heads examined were of copper which has an appreciably higher melting temperature than leaded brass and in addition is quite viscous when molten and for these a white hot mould held at elevated temperature for some time would have been necessary to ensure the metal filled the mould before setting

12  ‘The copy in the present Oni’s possession was suspected by me in 1945 of being a reproduction’ (Underwood 1949: 3).

13  This included metallographic examination and analysis. The conservation comprised partial cleaning and removal of the dirt followed by waxing.

14  Almost immediately after the publication of Fagg and Underwood’s paper Moss (1949) cast doubt on the conclusions of that publication. In a very short note reporting the composition of the Olokun and British Museum heads, Moss lamented that he had not been present to examine the head before its return to Nigeria, and thus he could offer no alternative interpretation, instead tersely stated that ‘I consider the conclusion that the head was sand cast in a four or five piece mould is not supported by the evidence offered’.

15  As the head had just been moulded in the British Museum together with the other Ife heads (Platte 2010: 43), presumably they, or at least Underwood, wrote from direct experience.

16  Although L. Frobenius (1913: 310) specifically described the head as being finely chased.

17  W. Fagg and L. Underwood also explored alternatives, stating that they ‘cannot be explained by accretions on the wax… before investment … since this would produce hollows in the casting’. This is incorrect; only if the solid accretion was in the wax surface and then became attached to the clay mould would it appear as a hollow in the casting. A putative solid accretion on the wax would merely become incorporated in the moulding clay and thus not contribute to the casting at all. However, if this accretion was the result of a build-up of wax turnings from carving the crest then this could have been moulded and then melted out together with the rest of the wax leaving a space to be filled by the incoming metal. Such carelessness on the part of the original craftsmen seems inconceivable.

18  See Rinne & Friel (1975) for a detailed description of the production of an after-cast of fragments of a lifesize Greek statue.

19  W. Fagg and L. Underwood did not attempt to explain how the core in their postulated sand casting could have been made to replicate all the internal features of the head.

20  The 1913 English translation of Frobenius’ original German publication stated that the head was 14.5 inches tall, that is approximately 5% larger than the present head. However, in the original German edition the height is given as 35.5 cm which equates to only 14 inches, very similar to the measurement made on the head as part of this project. If the present head really had been 5% smaller than that measured by L. Frobenius this would have been strong evidence that the head was indeed a copy. It is surprising that although W. Fagg and L. Underwood gave the erroneous figure in their 1949 article they apparently did not realise the significance of what would have been the most convincing piece of evidence to support the contention that their head was an after-cast.

21  Chrome green was used to patinate the plaster casts of the heads made at the British Museum in 1948 and this is very likely the innocent explanation of the presence of very small quantities on the head (it is significant that X-ray fluorescence analysis of the patinated surface failed to detect chromium, showing it is present only in very small amounts).

22  Specifically the malachite should have been decomposed at temperatures above about 200⁰C (Mellor 1967: III, 272), driving off the carbonate and hydroxyl fraction and leaving the copper oxide, tenorite.

23  Major analytical programmes have been undertaken on West African copper alloy castings, usually in the hope of establishing the source of metal. These studies commenced in the late 1950s with elemental analysis, joined from the 1970s by lead isotope analysis. The principal papers reporting elemental analysis include: for Ife – Moss (1949) and Barker (1965) wet chemical and emission spectrographic analyses; Willett (1959), Werner (1970), Werner & Willett (1975), further emission spectrographic analyses. For Igbo Ukwu – (Shaw 1970) emission spectrographic analyses; Craddock et al. (1997), atomic absorption analyses. For Benin and related metalwork- Werner & Willett (1975), emission spectrographic analyses; Craddock & Picton (1986), atomic absorption analyses. The papers reporting lead isotope analysis include: General – Goucher et al. (1978); Joel et al. (1995); Willett & Sayre (2006); for Ife – Willett & Sayre (2006); for Igbo Ukwu – Craddock et al. (1997); Willett & Sayre (2006); for Benin – Goucher et al. (1978); Joel et al. (1995); Willett & Sayre (2006).

24  These can have very strange compositions with several percent of arsenic or antimony in addition to being very heavily leaded (Craddock & Hook 1995).

25  The bronze was a massive neck ring that carried the marks of recent sampling. This raised the interest and suspicion of Bernard Fagg, brother of William Fagg, and the then Director of Antiquities in Nigeria (as reported in Willett 1964). The donor, one Adejumo, vicar of St. Philip’s church, Aiyetoro, Ife, explained that it had been assayed to determine the gold content. This seems a reasonable explanation; however B. Fagg would have none of it. Clearly those wishing to make an exact copy of the Olokun head had somehow carried out a full quantitative analysis in Nigeria to ascertain if it was of similar composition, and on discovering that it was bronze instead of the required leaded brass, discarded it.

26  In practice most metal forgeries even now use contemporary modern alloys that are often anachronistic to the supposed period of the object. There are a very few instances of reusing old metalwork of the correct period, but no recorded instances of making up an alloy so precisely to match the original (Craddock 2009: chapter 7).

Haut de page

Table des illustrations


		URL	http://journals.openedition.org/aaa/docannexe/image/266/img-1.jpg
		Fichier	image/jpeg, 708k

		Titre	Figure 1 – The Olokun head NCMM 38.1.2
		Crédits	© National Commission for Museums and Monuments, Nigeria
		URL	http://journals.openedition.org/aaa/docannexe/image/266/img-2.jpg
		Fichier	image/jpeg, 564k

		Titre	Figure 2 – Sketch of the Olokun head showing the positions of particular features and from where samples were taken
		Crédits	© M. van Bellegem & Tony Simpson
		URL	http://journals.openedition.org/aaa/docannexe/image/266/img-3.jpg
		Fichier	image/jpeg, 608k

		Titre	Figure 3 – Simplified sequence of direct hollow cire perdue casting
		Légende	A: a core in the form of the sculpture is made in clay;B: the sculpture is covered in beeswax over the clay core. Iron rods are inserted through the wax into the core to prevent movement during firing;C: fine details are sculpted in the wax. Tubes of wax, known as runners (a), are applied at the top. Separate wax vents (b) are inserted to allow gases to escape during casting;D: layers of clay are applied directly to the wax surface, enclosing the vents (b) and runners (a) to form a mould;E: the entire mould is heated, melting the wax which is drained away through the runners;F: molten metal is then poured through the runners into the cavities left by the wax;G: after it has cooled, the clay mould is removed and the runners and iron rods are cut off to reveal the completed sculpture;H: the sculpture is polished to produce a smooth surface
		URL	http://journals.openedition.org/aaa/docannexe/image/266/img-4.jpg
		Fichier	image/jpeg, 304k

		Titre	Figure 4 – Oxidised remains of an iron armature that would have run up the centre of the of wax rod forming one of the runners, now chiselled-off (cf. fig. 2)
		Crédits	© P. T. Craddock
		URL	http://journals.openedition.org/aaa/docannexe/image/266/img-5.jpg
		Fichier	image/jpeg, 1,6M

		Titre	Figure 5 – Remains of a riser behind the crest on the Olokun head, showing that the crest must have been inclined to necessitate a separate riser at the baluster (cf. fig. 2)
		Légende	NB the crest itself must have acted as the principal riser with another one probably at the back of the head
		Crédits	© P. T. Craddock
		URL	http://journals.openedition.org/aaa/docannexe/image/266/img-6.jpg
		Fichier	image/jpeg, 1,2M

		Titre	Figure 6 – Detail of the scarification on the Olokun head where the cast lines have been accentuated by scorping probably with a steel point (cf. fig. 2)
		Crédits	© M. van Bellegem
		URL	http://journals.openedition.org/aaa/docannexe/image/266/img-7.jpg
		Fichier	image/jpeg, 1,8M

		Titre	Figure 7 – A: Casts of chisel marks cut in the original wax of the neck of the Olokun head; B: Detail of marks showing their diffuse nature (cf. fig. 2)
		Crédits	© P. T. Craddock
		URL	http://journals.openedition.org/aaa/docannexe/image/266/img-8.jpg
		Fichier	image/jpeg, 760k

		Titre	Figure 8 – Detail of the surface of the headdress on the Olokun head (cf. fig. 2)
		Légende	The diffuse appearance of the casting is because it was never fully cleaned after casting, with some of the original mould material still present in the interstices. The remains of offerings and ancient iron oxide paint together with accumulated accretions of dirt and corrosion combine to obscure the detail ()
		Crédits	© M. van Bellegem
		URL	http://journals.openedition.org/aaa/docannexe/image/266/img-9.jpg
		Fichier	image/jpeg, 2,1M

		Titre	Figure 9 – Base of the crest (cf. fig. 2) where Fagg and Underwood erroneously claimed that the corrosion and dirt were in fact casts, thus demonstrating that the head must be an after-cast
		Crédits	© P. T. Craddock
		URL	http://journals.openedition.org/aaa/docannexe/image/266/img-10.jpg
		Fichier	image/jpeg, 1,3M

		Titre	Figure 10 – Detail of metal fins at the base of the crest (left and bottom left hand side; cf. fig. 2)
		Légende	These had run into a crack in the mould during casting and had not been removed in the subsequent cleaning operations
		Crédits	© M. van Bellegem
		URL	http://journals.openedition.org/aaa/docannexe/image/266/img-11.jpg
		Fichier	image/jpeg, 1,6M

		Titre	Figure 11 – Detail of the surface of the proper right depression behind the headdress (cf. fig. 2) still covered by refractory material showing that it cannot be the cut away remains of a runner or riser from the present casting
		Crédits	© S. La Niece
		URL	http://journals.openedition.org/aaa/docannexe/image/266/img-12.jpg
		Fichier	image/jpeg, 1,7M

		Titre	Figure 12 – Diffuse chisel marks where a runner on the outside of the head has been removed (cf. fig. 2)
		Crédits	© P. T. Craddock
		URL	http://journals.openedition.org/aaa/docannexe/image/266/img-13.jpg
		Fichier	image/jpeg, 1,7M

		Titre	Figure 13 – Fresh chisel cuts where one of the two runners inside the head (cf. fig. 2) was detached after the casting (cf. the diffuse marks where a riser channels on the outside of the head was removed) (cf. fig. 12)
		Crédits	© P. T. Craddock
		URL	http://journals.openedition.org/aaa/docannexe/image/266/img-14.jpg
		Fichier	image/jpeg, 1,3M

		Titre	Figure 14 – A (cf. fig. 2): proper left side of the head showing scarification with extra line running down. B: detail of the line and also a second shorter line of scarification to the left
		Légende	The first line was claimed by W. Fagg and L. Underwood to be a mould line on their postulated sand casting, the second line was not mentioned. They are clearly contemporary with the other scarification lines, mould lines would have run through them
		Crédits	© P. T. Craddock
		URL	http://journals.openedition.org/aaa/docannexe/image/266/img-15.jpg
		Fichier	image/jpeg, 1,7M

		Titre	Figure 15 – The 1910 break on the proper right hand side of the head (cf. fig. 2) showing the rather diffuse nature of the edge suggestive of it being the cast of a break, but is in fact due to the extensive corrosion of the metal (cf. fig. 16)
		Crédits	© S. La Niece
		URL	http://journals.openedition.org/aaa/docannexe/image/266/img-16.jpg
		Fichier	image/jpeg, 1,6M

		Titre	Figure 16 – Surface of the small fragment which became detached from the edge of the 1910 break (cf. fig. 15) showing massive intergranular corrosion, particularly the red cuprite and the yellow of the brass metal (width of field of view c. 2 mm)
		Crédits	© S. La Niece
		URL	http://journals.openedition.org/aaa/docannexe/image/266/img-17.jpg
		Fichier	image/jpeg, 1,6M

		Titre	Figure 17 – Interior of the head showing the remaining core material behind the eyes (cf. fig. 2)
		Crédits	© P. T. Craddock
		URL	http://journals.openedition.org/aaa/docannexe/image/266/img-18.jpg
		Fichier	image/jpeg, 1,2M

		Titre	Figure 18 – Taper section viewed with polarizing filters, showing developed cuprite growth at the grain boundaries. Note also the large grain size similar to those observed on some of the other heads examined for Moss in 1948. (Width of field of view c. 4 mm)
		Crédits	© S. La Niece
		URL	http://journals.openedition.org/aaa/docannexe/image/266/img-19.jpg
		Fichier	image/jpeg, 1,8M

		Titre	Figure 19 – A: Petrological thin section from the core of the Olokun head showing a vitrified fabric, with abundant poorly-sorted quartz and micritic calcareous fragments infilling the voids of the paste. (XPL, 5.4 mm field of view). B: Petrological thin section from the core of the British Museum head showing a non-calcareous and slightly micaceous fabric with poorly-sorted quartz and abundant opaques and iron oxides. (XPL, 0.87 mm field of view)
		Crédits	© M. Spataro
		URL	http://journals.openedition.org/aaa/docannexe/image/266/img-20.jpg
		Fichier	image/jpeg, 976k

		Titre	Figure 20 – SEM-EDX elemental X-ray map for calcium showing the penetration of calcium salts (red) into the core material in the Olokun head suggestive of prolonged burial
		Crédits	© M. Spataro
		URL	http://journals.openedition.org/aaa/docannexe/image/266/img-21.jpg
		Fichier	image/jpeg, 928k

		Titre	Figure 21 – VP-SEM image of two charcoal fragments in the Olokun Head core sample. The fragment on the right is a TLS (Tangential Longitudinal Section) with exclusively uniseriate rays
		Crédits	© C. R. Cartwright
		URL	http://journals.openedition.org/aaa/docannexe/image/266/img-22.jpg
		Fichier	image/jpeg, 544k

		Titre	Figure 22 – Micrograph of Olokun Head showing charcoal fragment in TLS with rays 4 (or more) cells wide (0.87 mm field of view)
		Crédits	© M. Spataro
		URL	http://journals.openedition.org/aaa/docannexe/image/266/img-23.jpg
		Fichier	image/jpeg, 1,2M

Haut de page

Pour citer cet article

Référence papier

Référence électronique

Paul T. Craddock, Janet Ambers, Maickel van Bellegem, Caroline R. Cartwright, Julie Hudson, Susan La Niece et Michela Spataro, « The Olokun head reconsidered », Afrique : Archéologie & Arts [En ligne], 9 | 2013, mis en ligne le 29 juillet 2015, consulté le 03 décembre 2024. URL : http://journals.openedition.org/aaa/266 ; DOI : https://doi.org/10.4000/aaa.266

Haut de page

Droits d’auteur

Haut de page

Sommaire - Document suivant

Index

Numéros en texte intégral

Tous les numéros

Présentation

Informations

Lettres d’information

La Lettre d’OpenEdition

Revue soutenue par l’Institut des sciences humaines et sociales (InSHS) du CNRS, 2023-2024

The Olokun head reconsidered

Résumés

Entrées d’index

Mots-clés :

Keywords:

Index géographique :

Plan

Texte intégral

Bibliographie

Annexe

Appendix 1: X-ray diffraction (XRD) analysis and metallographic examination (S. La Niece)

Appendix 2: Raman Spectroscopy (J. Ambers)

Appendix 3: Petrographic and chemical analysis of the cores (M. Spataro)

Petrographic analysis

SEM-EDX analysis

Elemental map composition

Discussion and conclusions

Appendix 4: Charcoal identification (C.R. Cartwright)

Techniques of charcoal identification

Results of the VP-SEM examination of the charcoal in the Olokun Head

Conclusions

Notes

Table des illustrations

Pour citer cet article

Référence papier

Référence électronique

Auteurs

Droits d’auteur

Navigation

Index

Numéros en texte intégral

Présentation

Informations

Suivez-nous

Lettres d’information

Affiliations/partenaires