Edith Sherwood Ph.D.

The Composition of Ife Heads - A Numerical Analysis

Among the most famous of the West African brass sculptures are the eighteen naturalistic portrait heads discovered in Ife Nigeria, in 1938. Recently, an exhibit of these heads together with other brass, terracotta and stone artifacts from Ife, toured selected museums in the UK and USA(I). This exhibit caused many viewers to reconsider their opinion that African art is the work of a primitive people. These remarkable artifacts are equal in workmanship and artistry to sculpture made during the European Renaissance.

  • Leo Frobenius, during an expedition to Nigeria from 1910-12, was shown the brass Olokun head, the Obalufon mask and various terracotta heads. He was astonished by the quality and artistry of the objects and questioned their African origin. He wondered whether he had found evidence of an African Atlantis(III).
  • Frank Willett proposed that the Ife heads were used as funeral effigies for various deceased Onis of Ife, and were the product of different sculptors working over several centuries. This hypothesis is still accepted by many African art historians(IV).
  • Kenneth Murray, suggested that “only one artist, working for a couple of years, made the majority of the heads.” His reason for this conclusion was based on the heads’ stylistic similarities(V).
  • William Fagg pointed out that all the Ife heads had a couple of features in common; the mouth was displaced by about 6 millimeters from the midline of the face and the corners of each mouth ended in a depression or dimple. He considered that these “two artistic peculiarities are a mannerism of the artist rather than fidelity in portraiture.” Portraits painted by some Renaissance artists have the same artistic peculiarities(VI).
  • Stefan Eisenhofer is of the opinion that the Ife heads can only be dated to some period before 1700 AD(VII). Thermoluminescent analysis of core samples taken from the interior of two of the heads gave the following approximate dates: 1440 +/-65 A.D and 1490 +/-85 A. D. possibly dating the Ife heads to the latter part of the 15th century(VIII).
The Composition of the Heads

The Ife heads fall into two groups; six were cast from almost pure copper (data not given), the remaining twelve from brass. Chemical analysis of the brass is presented in Table I(IX). The sample numbers are assigned by Underwood(X).

Table I Composition of the Ife Brass Heads. (Baker 1965)
Head No.% Copper% Zinc% Lead% Tin% IronTotal
168.811.615.90.90.998.1
271.213.311.80.60.799.6
774.015.35.61.20.296.3
871.815.011.10.70.699.2
974.39.611.61.30.699.2
1174.111.911.31.30.898.4
1277.59.39.91.10.698.4
1373.49.714.50.00.998.5
1476.211.06.41.70.498.6
1579.811.84.40.90.697.5
1674.212.311.01.00.699.1
1873.413.111.40.80.699.3
  1. The analytical results do not add up to exactly 100%, indicating that some oxidation of the metal surface may have taken place.
  2. Head 7 contains 2.6% gold in addition to the other metals.
Table II Physical Properties of the Metals
Metalm.p. °Cb.p. °CDensity gm/cm3Source in Nature
Copper108325628.92Free metal
Zinc419.59077.10Salts only
Lead327.5171811.3Free metal
Brass900-940 8.4-8.75
  1. Zinc, due to its low boiling point, is volatile and is readily oxidized to zinc oxide. The density of brass varies according to its composition.
  2. Zinc ores, although abundant in nature, only occur in the form of a salt like calamine (zinc carbonate). The free metal was isolated in Europe in the 18th century, but may have been made in India at an earlier date.
  3. Brass was first produced during Roman times using a cementation process.
  4. When molten brass containing less than 35% zinc solidifies, the zinc and copper atoms bond in a face centered cubic lattice structure known as alpha brass.
  5. Brass loses approximately 10% of its weight when re-melted due to some of the zinc atoms vaporizing(XI).
  6. About 2% by weight of lead is added to molten brass to make it easier to pour. As the brass cools, the lead separates into small droplets. The size of the droplets depends on the rate of cooling. A slow rate produces larger droplets. Metallographical examination of the micro-structure of Head 7 indicated a slow rate of cooling(XII).
Manufacture of Brass Using the Cementation Process

The Europeans used the cementation process to manufacture brass until metallic zinc was isolated in the 18th century. During the Roman era, cementation involved smelting copper with a mixture of zinc ore (calamine/zinc carbonate) and charcoal. The charcoal reduces zinc ions to vaporized zinc. The molten copper absorbs the vapor. This process produced a brass containing less than 10% zinc. In the 15th century, the calamine was first roasted in a wood fire to zinc oxide. Crucibles, packed with layers of ground zinc oxide mixed with powdered charcoal and copper granules were heated in a furnace to a temperature of 1000 °C +/-100 °C(XIII)(XIV)(XV)(XVI). This temperature is sufficient to reduce the zinc ions to zinc vapor, but not high enough to melt the copper (melting point 1083 °C). The vaporized zinc permeates the surface of the copper granule. The color of the fumes emitted from the crucibles were used as a guide for controlling the temperature of the furnace. Finally, the temperature was raised sufficiently to melt the copper, and the brass was poured into molds. A brass containing 28% by weight in zinc is the highest concentration of zinc this method can produce(XVII).

One drawback to making brass using the cementation process was that the uptake in zinc could not be easily determined(XVIII). However, the 15th century metallurgists had observed that during the initial heating of the copper granules with charcoal and zinc oxide, they gained weight, reflecting the uptake in zinc. The composition of the brass may be calculated by weighing the copper granules before and after heating. Therefore brasses of lower zinc to copper ratios may be produced by the addition of an appropriate weight of copper to the molten alloy whose uptake in zinc had previously been determined. This would require using a crucible with a lid and covering the molten alloy with melted glass to prevent loss of zinc. Whether this was actually done is not known.

The Source of the Lead

The brass Ife heads have an unusually high lead content. Werner and Willett suggested that the ores used to cast these heads came from the Harz region of lower Saxony. They postulated that various leaded zinc ores when cemented with copper, would produce the percentages of copper, zinc and lead given for the brass Ife heads in Table I . However, the presence of 2% lead prevents the absorption of zinc by copper by 4%. Head 8 contains 71.8% copper, 15% zinc and 11.1% lead(XIX). A leaded zinc ore containing 11.1% lead would produce a brass containing about 6% (28 – (4 * 11 / 2)) zinc. This head was therefore made by adding refined lead to the molten brass in the melting pot. The same argument holds for the other brass heads. Figure 2 in Werner and Willett’s paper, gives a plot of the percentages of lead vs zinc given in Table I. This plot shows graphically that the percentages of lead fall into three groups, with mean values of 5.47%, 10.94% and 14.45% respectively. These means are approximately in the ratio of 1:3:4. Some heads received 1 bar, others 2 bars and others 3 bars of lead. Lead isotope analysis showed that the lead in heads 1, 8, 9, 12, 15 and 16 were from the same source and heads 2 and 11 from a different source. The source of the lead would not matter, provided that all the bars weighed the same. If lead bars were available, it is reasonable to assume that copper bars were used to cast the six heads made from almost pure copper.

Hypothesis for the Casting of the Ife Brass Heads

The only refined source of zinc in the 15th century was brass. I assumed that the zinc in each head came from a brass containing the same amount of zinc as the brass head containing the highest percentage zinc, namely head 8. If the percentages of lead and the minor impurities are ignored, the composition of the brass present in head 8 can be calculated. Head 8 contains 15.0% zinc and 71.6% copper. This head may have been cast from a brass containing 17.3% zinc (100*15.0 / (71.6+15.0). This represents a brass with the lowest concentration of zinc that can be used to cast all the heads. If 17.3% brass is substituted for the zinc in the other heads, then additional copper is required. Table III shows the percentages of 17.3% brass in column C, and copper in column D, required for each head to have the same percentages of copper and zinc as in Table I.

I experimented with brass containing higher concentraions of zinc, but did not obtain the same uniform pattern of results that 17.3% brass produced.

Table III – A Numerical Analysis of the Data from Table I
Head No.% Cu
A
% Zn
B
Calc % brass
(Zn*100/17.3)
C
% Cu left
D=A+B-G
D
Brass/Cu
C/D
E
No. bars
Brass/Cu
F
Pot No.
Brass/Cu
G
168.811.667.0513.355.025:120:4
271.213.376.887.6210.0910:120:2
774.015.388.440.86No CuAdded24:0
871.815.088.710.09No Cuadded24:0
974.29.655.4928.311.982:116:8
1174.111.968.7917.214.004:120:5
1277.59.363.7633.041.633:215:10
1373.49.756.0727.032.072:116:8
1476.213.980.359.758.248:124:3
1579.811.868.2123.392.923:118:6
1674.212.371.1015.404.629:218:4
1873.413.175.7210.787.037:121:3

Table III shows the following:

  1. The ratio of percentages of 17.3% brass to free copper is given in column E. With the exception of heads 12 and 16, this ratio is approximately an integer. It is actually the ratio of two integers, i.e. the ratio of the number of bars of 17.3% brass and copper that may have been used to cast each head.
  2. The ratios of 1.63 and 4.62 for heads 12 and 16 may correspond to 3:2 and 9:2 bars of brass to copper respectively.
  3. Column F gives the minimum number of bars of brass (17.3% zinc) and copper, required to satisfy the ratio calculated in the previous column.
  4. Column G shows the total number of bars of brass and copper that may have gone into the melting pot.
  5. If Table III is correct, the Ife heads were cast from brass, copper and lead bars. When brass is remelted approximately 10% of its weight is lost due to the zinc vaporizing.
  6. The brass used to cast these heads may have had an initial composition of approximately 19.2% zinc.
  7. Brass rods, containing 20% zinc and less than 0.5% lead, were found in 1969 at Ma’den Ijafen, Mauritania. They were shipped from a Mediterranean country to North Africa in the 12th century A.D.

Table IV, makes the assumption that a total of approximately 24 brass and copper bars combined, went into the melting pot to cast each Ife head.

Table IV – A Numerical Analysis of the Data from Tables I and III
Head No.% brass%Cu left%PbRatio brass/CuNo. bars brassNo. bars CuNo. bars Pb% wt brass bar% wt Cu Bar% wt Pb bar
161.713.415.95:120433.43.35.2
276.97.6213.810:120233.83.84.6
788.40.865.6No Cu24013.75.6
886.70.0911.1No Cu24023.65.5
955.428.313.62:116833.53.54.5
1168.817.211.34:120523.43.45.7
1253.833.09.93:2151023.63.35.0
1356.127.014.52:116833.53.44.8
1480.49.756.48:124313.33.36.4
1568.223.34.43:118613.83.94.4
1671.110.811.09:218424.03.95.7
1875.723.311.47:121323.63.65.7
Mean       3.63.55.3
  1. The % weight of each bar was obtained by dividing the % weight of each metal by the number of bars used to cast the head.
  2. The brass and copper bars appear to weigh about the same amount.

I was only able to find a weight for Head 11, 20 lbs or 9.08 kg. If nearly all the alloy in the melting pot was used to cast this head, then 86% of the weight of the head is copper and zinc, i.e. 25 brass and copper bars had a combined weight of 7,809 gms. The weight of each bar was about 312gms (7809/25). The weight of a lead bar was about 466 gms (312*5.3/3.55). From about 1480, the Portuguese used brass and copper manillas to buy gold, pepper and ivory and later slaves from the West African tribes, including those in Nigeria. Brass manillas, recently recovered from a 1524 shipwreck off the Basque Coast of Spain, weigh an average of 306 gm. The Portuguese purchased brass and copper manillas from Erasmus Schetz of Antwerp. I have been unable to find an analysis of the brass used to cast these manilas.

Conclusion

An initial survey of the results obtained from the chemical analysis of the brass Ife heads (Table I) suggested that refined metals were used to cast these heads. The only assumption I made was that the zinc in all the heads was replaced by a brass whose composition is that of the head with the highest percentage of zinc. The other heads required the addition of copper to the brass in order to duplicate the percentages of copper and zinc in Table I. Sufficient quantities of brass, containing the same percentage zinc had to be available to cast all the heads. The cementation process may have been adapted to produce brasses with different percentages of zinc.

It is difficult to believe that the data presented in Tables III and IV is meaningless. It fits together too well and supports some of the hypotheses presented by many African art historians. The results from Tables III show that the copper and zinc percentages in Table I, may be duplicated by using an appropriate combination of brass (17.3% zinc after remelting) and copper bars. The means were obtained by averaging the three clusters of lead in Table I. These means fall approximately into the ratio, 1:2:3, indicating that lead bars of similar weight were added to the pots containing molten brass and copper. Table IV shows that approximately 24 bars of copper and brass, may have been used to cast these heads and that both the copper and brass bars weighed about the same amount. Based on the weight of a single head, the brass and copper bars weighed approximately 312 gms and the lead bars weighed approximately 466 gms.

The Ife heads, with the exception of head 13, all have the same fingerprint of trace metals, indicating that the copper came from the same mine. This supports Kenneth Murray’s hypothesis that the Ife heads were made in the same workshop at about the same time. William Fagg’s observation that all the heads show the same modeling characteristics indicates that the same artist modeled all the heads. Since Nigeria has no copper deposits, the metals would have been mined and refined in Europe then transported to Nigeria. As there is no archeological evidence that Ife had any kilns, the artist who produced these flawless castings using the difficult and exacting lost wax process probably had a workshop elsewhere.

References 
  1. ↑ back Drewel, H.J., & Schildkrout, F., 2010, Diversity and Divinity Ife Art in Ancient Nigeria, University of Washington Press.
  2. ↑ back Blier, S.P. http://www.collegeart.org/pdf/artbulletin/Art%20Bulletin%20Vol%2067%20No%203%20Blier.pdf, p. 12
  3. ↑ back Frobenuis, L., 1913, The Voice of Africa, London Hutchinson and Co, vol I, pp. 319-322.
  4. ↑ back Willett, F., 1967, Ife, McGraw Hill, New York, pp. 18-30.
  5. ↑ back Murray, K.C., 1963, Ancient Ife (Letter to the Editor), Odu, ix, pp. 71-80.
  6. ↑ back Fagg, W. & Pemberton, J. 3rd, 1982, Yoruba, Alfred A. Knopf, New York, p. 10.
  7. ↑ back Eisenhofer, S. African Art, Taschen, p. 46.
  8. ↑ back Willett, f. & Fleming, S.J., 1976, Archaeometry, xviii, pp. 136-37.
  9. ↑ back Baker, H., 1965, Man, 10, January-February, p. 23.
  10. ↑ back Underwood, L., 1968, Bronzes of West Africa, Tiranti, London.
  11. ↑ back Cradock, P.T., 1978, Journal of Archeaological Science, 5.
  12. ↑ back Baker, H., 1965, Man, 10, January-February, p. 23.
  13. ↑ back Bourgarit, D. & Bauchau, F., 2010, JOM. Vol. 62, No.3, pp. 27-33.
  14. ↑ back Pollard, A.M. & Heron. C. 2008, Archaeological Chemistry, p. 195.
  15. ↑ back Bourgarit, D. & Thomas, N.,
    http://www.academia.edu/2634806/David_Bourgarit_Nicolas_Thomas_From_laboratory_
    to_field_shared_experiments_in_brass_cementation
    .
  16. ↑ back Martinon-Torres, M. & Rehren, T., 2002, Agricola and Zwickau thgeoty and practices of Renaissance brass production in SE Germany, HistMet 36, 17 pages.
  17. ↑ back Cradock, P.T., 1985, Archaeometry, 27 no.1, 23-25.
  18. ↑ back Werner, O. & Willett, F.,1975, Archaeology, vol 17, No. 2.
  19. ↑ back Cradock, P.T., 1978, Journal of Archeaological Science, 5, p. 12.
Website design by Erica Sherwood