Studies of the African Iron Age Judith A. Todd University of California

Berkeley, California

Primitive ironmaking is still practiced in certain parts of Africa; modern analysis techniques can relate the composition of slag inclusions in finished objects to the local ores. - - - - SUMMARY - - - The contribution the metallurgist can make to African Iron Age studies has been assessed by detailed investigation of the pre-industrial bloomery iron process still practiced by the Dimi people of the Gemu province in southwest Ethiopia. Metal objects and slags from this process were examined metallographically, and chemical analyses of the ores and slags have been obtained. Quantitative energydispersive x-ray analyses have shown that it is possible to relate the composition of the slag inclusions to the iron ores of the region. Such an analytical technique, when combined with archaelogical, linguistic, and historical evidence, may provide a valuable means of characterizing the ironwork of specific production centers and establishing trade networks of iron objects. In this context, theories of the African Iron Age are reviewed with specific reference to Ethiopia.

T

he gradual accumulation of radiocarbon dates for the African Iron Age has given rise to several theories concerning the development and diffusion of ironworking throughout the continent. Despite their increasing number, many of these dates are concentrated in West Africa, the Nile Valley, Eastern and Southern Africa (Figure 1); more recently, dates have also been obtained from West Central Africa. Relatively few iron age sites of the 1st millenium A.D. have been discovered partly because the sites are difficult to recognize as iron was probably smelted outside the village, and partly because iron objects may be scarce due to the re-use of scrap. Besides the dating of carboniferous material associated with excavated furnace sites, linguistic, historical, archaeological, and ethnographic evidence have contributed to our knowledge of the African Iron Age. This paper reviews the recent theories concerning the African Iron Age and considers the contribution which the metallurgist can make to these JOURNAL OF METALS. November, 1979

studies by discussing one of the few remaining pre-industrial bloomery iron processes still practiced in a remote part of Ethiopia. This process is a very valuable ethnographic and metallurgical record because imported scrap iron has killed local smelting industries in many regions of Africa, particularly in coastal areas. Hence, studies of ancient smelting practices are often based either on metallurgical reconstructions suggested by existing archaelogical evidence or on reconstructions of the process in parts of Af.rica where the practice ceased many years ago.

This paper also details the Dimi bloomery process, still practiced in 1973 by up to forty smiths in Dimam, Gemu Gofa Province, S.W. Ethiopia. The Dimam region is a mountainous area 70 miles from the nearest source of scrap iron; and as the only means of transport is to carry iron by head-load, the Dimi continue to smelt iron in the traditional way. It will then be shown how metallurgical examination of the smelting products, particularly analyses of slag inclusions trapped in the metal, can characterize the ironwork of a specific production center.

.Do DI •• i€IC

€ AD

SIC

Figure 1. Carbon-14 dates for early ironworking in Africa.

39

THEORIES OF THE AFRICAN IRON AGE The oldest hypothesis suggests that a knowledge of ironworking spread from the Nile Valley southward to Meroe in the Sudan, where the presence of slag mounds indicates the production of iron on a large scale. From here, various routes have been suggested for the diffusion of ironworking to W. Africa and also through E. Africa to the south. However, Trigger' has pointed out that iron objects are rare in Nubia before 400 B.C., and Tylecote' has found no trace of iron smelting at Meroe before 300 B.C. Tylecote claims that there is no technical evidence for the spread of the Meroitic tradition of ironworking and that the slag mounds could be accounted for by the needs of the Meroites. That ironworking is believed to be late (possibly 400-500 A.D.) at Daima near Lake Chad, situated very near to the likely corridor from the Nile Valley to W. Africa, is an argument against Nigerian iron working originating from the East. The earliest West African evidence for ironworking comes from Taruga in Nigeria,' where smelting furnaces ranging from 440 to 280 B.C. have been found. Tylecote considers these furnaces similar to European Iron Age types and suggests that the same form was also present in Carthage. He thus derives the furnaces at Taruga from N. Africa. According to Williams,' similarities between the Celto-Roman furnace and ex-

amples from Upper Niger make it difficult to avoid the conclusion that African designs reflect a mediterranean influence on native metallurgy at an early date. Such theories are now being strongly questioned. The distribution of rock paintings and engravings of horse-drawn wheeled chariots in the Sahara indicates two possible trans-Saharan routes in the mid-first millenium B.C., one from Morocco to Mauretania and the other from the Tunisian area acorss the Tarsili and Hoggar mountains to the middle of Niger. A knowledge of ironworking could, therefore, have come from the Carthaginia area to Nigeria and/ or in the trail of copper workers to Mauretania and then spread east and south. As more early Iron Age sites are excavated, it is becoming evident that ironworking had probably spread over much of W. Africa by 300-600 AD. Similarly, recent dates from W. Central Africa show with more arid more certainty that the early Iron Age was fully established in central and southern Rwanda by the 3rd century AD. In fact, this region seems to be a very old metallurgical center from which Iron Age technology could have spread to the east and south. Linguistic evidenceS has associated . the diffusion of iron technology with the spread of the Bantu from the Niger-Cameroons border area (1000-500 B.C.) eastward across sub-Sahara Africa, and Phillipson· has traced two routes southward through E. Africa. Current radiocarbon dates confirm an Eastern

• Alllm Lake Tana

o

• Addis Ababa

Omo Valley

Site

Figure 2. Map of Ethiopia showing location of the Omo Valley site, Gema Gofa Province. S.W. Ethiopia. 40

African ongm for the S. African Iron Age rather than the Congo Basin-Sambia area as suggested by Oliver" on linguistic grounds. Recent work by Schmidt and Avery7 has provided archaeological verification of the oral tradition of ironworking among the Haya of northwestern Tanzania and has dated iron production to 2400-2550 years ago. Although iron objects have been discovered at Haoulti in N. Ethiopia as early as the 5th century B.C., the lack of archaeological evidence means that it is possible only to speculate on the links between the Aksumite empire, which reached its peak approximately 300 A.D., the Meroitic civilization, and the East African Interior. Increased iron usage in Ethiopia appears to be the result of prolonged contact with S. Arabia, the links being more clearly defined than those with Meroe. As no evidence suggests that the Ethiopian process is derived from the interior of Africa, it is also possible to take the view that ironworking may have developed indigenously in Ethiopia, as several scholars believe is the case for W. Africa. THE DIMI BLOOMERY PROCESS Although there are no excavated ironsmelting sites in Ethiopia, information concerning the traditional process can still be gained by observing presentday practices. In 1973, fieldwork was carried out among the Dimi, who inhabit a mountain range in the northern part of the Omo Valley, S.W. Ethiopia (Figure 2). The manufacture of iron is still an important local industry, supplying most of the agricultural tools, weapons, and ritual objects used by the Dimi and their immediate neighbors. In the past, Dimi ironwork was traded extensively in the surrounding regions, mainly for salt, but the introduction of scrap metal to a town 70 miles away has severely restricted this trade. Dimi smiths, however, find such scrap difficult to work and, in view of transport problems, prefer to manufacture their own iron. The main ore used for smelting was limonite (Fez03'2HzO) but magnetite was also smelted. Iron ore was dug from surface pits and crushed so that gangue could be removed by hand before sun drying. No prior roasting of the ore was observed. The ore, plus a small quantity of recycled slag, was reduced with charcoal fuel in the Dimi furnace, which is a forced-draft, conical, nonslag-tapping shaft furnace, one meter high, with a pit extending 50 em below ground level. At the beginning of the smelt, the furnace was filled with dried grass (Figure 3) which was lit and allowed to burn for a few minutes before charcoal was added. The smith continued setting up the furJOURNAL OF METALS • November, 1979

two-part tuyere

Figure 3. (left) Filling the furnace with dried grass. Note the repaired cracks in the furnace walls and also the Bowl of cow dung, used to line the inside of the bellows pots.

nace and periodically added layers of ore and' charcoal until the furnace was full. Six sets of three holes were cut in the base of the furnace for 18 tuyeres linked to six bowl bellows. The tuyeres (Figure 4) consist of a two-pipe arrangement, a total of 36 pipes, the larger pipes of approximately 25 cm length being sealed with clay into the base of the furnace, and the smaller pipes sealed to holes on the outside of the bellows pots. The bellows pot was a clay bowl set into the ground and sun dried. The diaphragm, not a permanent fixture, consisted of a goatskin or sheepskin tied to the bowl with rope. The center of each skin contained a hole which acted as a valve. It was operated by inserting the thumbs in the hole and raising the skin to admit air (Figure 5) then closing the hole with the palm of the hand to force the air into the furnace . The average blowing rate ~as 47/min to a rhythm called by the leading smith, but was increased to 60/min for short periods after the addition of fuel and charcoal. The bellows were operated for four hours, then work ceased, the tuyeres were removed, and the furnace left to

cool. The following morning, the bloom was quenched with water, broken with digging sticks, and then passed out through the top of the furnace (Figure 6) . I t consisted of iron, slag, and charcoal fritted together; the majority of slag was broken away and scattered into the bush. Later, the iron was taken to the forge, heated in a charcoal fire, and forged with stones (or a blunt-ended iron hammer for smaller items) into larger pieces for tools. The iron yield was low in the observed Dimi smelt, but it was reported that sufficient iron to make a pair of digging sticks (approximately 6 kg) could be produced by similar operation of the furnace . During the Dimi smelt, samples of ore, charcoal, furnace slag, cinder, furnace material, and a discarded tuyere were collected. Unfortunately, the smith would not part with more than a few fragments of iron. However, samples of a bloom (Figure 7) and more iron ore were later obtained from a second site. A series of iron artifacts were also collected from the surrounding region for examination; Figure 8 shows a digging stick, spearhead, hoe, black iron bangle, sickle, cowbell, and a sacrificing knife with an I8-inch blade.

Figure 8. Examples of Dimi ironwork, including digging stick, spearhead, hoe, iron bangle, sickle, cowbell, and sacrificing knife with an 18-inch blade. JOURNAL OF METALS • November, 1979

Figure 5. Bellows in operation, showing goatskin diaphragm.

Figure 6. Removing the bloom from the furnace.

Figure 7. Fragment of bloom, 2.5x. 41

ANALYSES OF BLOOMERY PRODUCTS

The iron produced in the Dimi process forms as a solid sponge or "bloom" of metal coated with furnace slag. Although most of this slag is squeezed out of the bloom during forging, small fragments or inclusions inevitably remain in the iron. Provided that no flux has been added during the reduction process, these slag inclusions represent an original furnace product and their compositions may be related to the iron ore used for the reduction. By determining the compositions of slag inclusions and iron ores, it may be possible to relate iron objects to ore sources, and hence determine the provenance of iron objects. The Dimi ore and slag samples were first analyzed qualitatively using a D.C. arc Hilger Large Littrow Spectrograph to determine the elements . present in major, minor, and trace quantities (Table I). (Sulfur is not detected by conventional spectrography, and detection of phosphorus is only moderate.) Having identified the elements present in the ore and slag, the major and minor elements (greater than 0.1%) were determined by standard wet chemical analysis. Analysis was focused on the major and minor elements rather than trace element analyses since the latter involve larger errors arising from analyses at very low overall concentrations and also from heterogeneity in sampling the material. Analyses of the ores (Table I) indicate that the gangue material consists mainly of alumino-silicate clay, with very low lime content and manganous oxide varying from negligible to 3.7%. The high alumina and silica content of the ore is reflected in the analyses of the slags (Table II), although the silica content of slag 54 may be high due to reaction with the forge pipe. The lime content of the slag is higher. than would be expected from the ore analyses, and this is attribu ted to the absorption of lime from the charcoal (Table III). Many reports concerning bloomery products contain relatively few analyses because detailed chemical analyses can take considerable time. Comparative analyses were made using the stereoscan with energy-dispersive x-ray techniques, as the time required to make a single measurement of all elements with atomic numbers above sodium is only seconds for a given region. The xrays, generated by a high-energy electron beam exciting a small volume of material, are detected, amplified, and displayed as shown in Figure 9a for rapid identification of the elements. As no computer attachment was available for immediate processing of the data, the background counts were subtracted by fitting peaks to the spectrum (Figure 9b). The integrals of the corrected peaks were then compared with those from standard materials to obtain an apparent concentration, 42

Table I: Chemical Analyses (wt.%) of Ethiopian Ores

Fe z0 3 SiOz MnO

AIP3

CaO MgO % moisture loss % combined HP,COz PzOs S TiO z Total CaO: SiOz

01 Chalco

02 Gerfa

03 Gerfa

04 Garo

05 Wocho

06 Balcha

68.67 6.70 3.43 5.75 nd nd 2.49

75.73 7.10 2.06 7.42 nd nd 2.18

77 .37 6.70

55.79 17.71

8.53 63.92

1.29 nd nd 2.52

11.18 nd nd 3.35

18.31 nd nd 10.08

53.40 12.30 3.70 8.20 0.50 tr

12.12

3.10

11.72

10.77

3.40

0.73 0.01

2.05 0.01

0.27 0.01

0.64 0.01

tr 0.01

99.89

99.28

99.87

99.44

104.24

0.97 <0.10 2.50 0.041

Table II: Chemical Analyses (wt.%) of Ethiopian Furnace Slags and Clay

Sl Gerfa Fez03 FeO SiOz MnO Alz03 CaO Pz Os TiO z % moisture 1055 S Total

S2 Garo

41.37 28 .73 2.01 21.94 5.73 ns ns 0.49 0.02 100.27

45.00 34.50 1.25 14.30 1.40 1.29 1.90 ns nd 99.64

S4 Gerfa forge

Clay

47.11

11.52

10.60

32.19 nd 16.67 nd ns ns 0.47 0.01 96.44

63.19 nd 11.16 6.05 ns ns 0.46 0.02 92.38

67.20 0.07 14.10 1.30 ns 1.10 ns

S3 Garo

~

94.37

Table III: Analysis of Ethiopian Charcoal Spectrographgic Analysis Considerable amount - AI, Ca, Fe, Mg, Mn, Si Trace amount - B, Cu, Pb Slight trace - Mo, Ni, Ag Chemical Analysis, wt. % 0.010 0.01 0.026 3.7 Ti . very slight trace Total ash at 800°C: 1.53 wI. %

Figure 9. a.) Energy-dispersive x-ray spectrum, and b.) corrected spectrum, as used to analyze slags and inclusions.

which was then further corrected for absorption, fluorescence, and atomic number effects using the program of Duncumb and Jones. 11 Both furnace slags (Figure 10, Table IV) and inclusions (Figure 11, Table V) were analyzed this way. It was also possible to obtain analyses which agreed well with the bulk composition of the slag by scanning the electron beam over JOURNAL OF METALS • November, 1979

Table IV: Analyses (wt.%) of Individual Phases in Four Inclusions in Ethiopian Furnace Slag S2 (EDAX) Dark Phase

Light Phase FeO MnO CaO Si0 2 K20 Al 20 3 Ti0 2 P20S MgO Total CaO:Si0 2

Inclusion 1

Inclusion 2

Inclusion 3

Inclusion 4

Inclusion 1

Inclusion 2

Inclusion 3

Inclusion 4

66.0 2.2

66.4 2.4

67.7 2.5

27.7

27.1

0.3 0.8 1.4 98.4

1.1 1.3 98.3

63.9 1.8 0.3 29.1 0.3 3.6 0.9 1.0

27.3 1.1 2.7 41.9 2.1 15.7 6.7 2.1

21.6 1.2 4.2 47.9 2.3 15.2 1.6 3.7 1.0 98.7 0.088

36.1 0.9 1.0 38.6 1.7 18.7 2.9 1.7

21.2 1.2 3.9 50.0 2.5 16.0 1.7 3.1

101.6 0.026

99.6 0.078

100.9 0.010

27.1

0.9 1.8 100.0

99.6 0.064

Table V: Analyses of Single-Phase Inclusions in Ethiopian Knife (EDAX)

FeO MnO CaO Si0 2 K20 Al 20 3 Ti0 2 P2 0

S

MgO Total CaO: Si0 2

1. 55.6 1.1 4.5 24.0 1.4 6.4 0.5 5.5

2. 55.3 1.2 3.2 22.0 1.1 6.8 0.8 11.9

3. 54.5 1.3 4.2 26.5 1.5 7.5 0.5 7.7

4. 54.7 1.3 3.1 23.3 1.9 6.9 0.9 5.6

99.0 0.188

102.3 0.145

103.7 0.158

97.7 0.133

5. 54.3 1.4 2.6 23.8 1.3 7.0 0.8 11.3 1.2 103.7 0.109

6. 53.3 1.2 2.6 21.7 2.5 5.1 0.8 12.8

7. 53.4 1.3 4.0 26.6 2.3 8.3 0.6 4.4

8. 61.8 1.3 2.9 22.0 1.0 5.0 1.1 4.2

9. 50.6 1.3 3.3 27.0 1.7 7.8 1.0 11.3

100.0 0.120

100.9 0.150

99.3 0.132

104.0 0.122

10. 48.6 2.2 3.5 18.4 0.3 5.8 0.7 22.1 2.2 103.7 0.19

Figure 10. Furnace slag.

areas of the sample containing representative slag phases. Chemical analysis of the iron gave the following composition: 0.4 C, < 0.01 Mn, 1.01 P, < 0.01 S, 0.17 Si (balance Fe). Metallographic examination of a polished surface etched in 3% nital solution, revealed a carbon content varying from that of a eutectoid steel (0.8% C) to wrought iron. Samples of material were· also taken from a knife and a digging stick. I t is not always feasible to examine a complete microstructure, particularly in specimens required for museum display. However, a wedge-shaped piece of JOURNAL OF METALS • November, 1979

Figure 11. Inclusion in knife section.

material can be removed using a jeweler's saw, providing two edges which can be polished to reveal the structure at, say, 75° to the edge and which should reveal directional characteristics. Metallographic investigation and compositional analyses (using the electron probe micro analyzer or scanning electron microscope with ED AX attachment) can

be made and the section replaced using a suitable filler. This preserves the appearance of the artifact but provides a specimen which can be re-examined if, for example, new techniques become available for inclusion analysis; and, unlike wet chemical analysis, there is no substantial destruction of the removed section. 43

Figure 12. Etched microstructure of knife.

The knife section (Figure 12) and the digging stick microstructure were again variable in carbon content between approximate limits of 0.02-0.2%, that is, lower than the average carbon content of the bloom. The knife section contained a weld defect, indicating that the thickness had been formed by forging two pieces of iron together, and the direction of elongation of many inclusions confirmed that a principal direction of forging was from center to edge. Both sections were decarburized at the surface, apart from one small area of the knife where a maintained carbon content extended to the edge. The central region of the knife was ferritic with a large grain size and was surrounded by a series of alternate bands of higher carbon content and ferrite. Many inclusions were associated with pearlite, lying in the

higher carbon bands (Figure 11). Etching with Oberhoffer's reagent revealed that phosphorus was distributed in the large ferritic grains and that the areas immediately adjacent to the inclusions, which were associated with pearlite, appeared to be lower in phosphorous than the rest of the grain. This depleted zone was too small to be detected by the microprobe electron beam, but was detectable by a very fine beam on the stereoscan. It would appear that during forging, phosphorus had been oxidized from a zone of iron surrounding the inclusion into the. slag phase, the oxygen being supplied by the ferrous oxide in the inclusion. This phosphorus-depleted zone accounts for the association of pearlite with these inclusions, since carbon from the surrounding iron transforming first from austenite to ferrite (phosphorus is a ferrite stabilizing element), would diffuse to areas of lower phosphorous concentration along the most rapid diffusion path - the grain boundaries. Examination of the inclusion analyses (Table V) establishes that the phosphorous pentoxide content of the inclusions did not arise as the result of flux, such as bone, added to lower the free-running temperature of the slag, btlt is a direct consequence of the high phosphorus content of the iron ore and hence the iron. This raises the important point that classifying iron ar-

tifacts by their phosphorous content should be approached with caution. The inclusions in the knife (Table V) and the digging stick sections show values of manganous oxide in 0.1-2.0% range. Very strong reducing conditions are required for manganese to appear in the iron; and although this is achieved by solid state. reduction of magnetite ores in the Swedish Hoganas process, employing sealed saggers, Tylecote'z points out that manganese in ores smelted by the bloomery method rarely appears in the iron. This is supported by the low values of manganese found in Dimi iron. The manganous oxide content of the Ethiopian furnace slag and the inclusions examined are consistent with an iron manufactured from ores 03 and 04 (Table I), but are low in relation to the manganous oxide content of ores 01 and 06. To investigate this relationship further, laboratory reductions of 03 and 06 were carried out in an atmosphere of carbon monoxide at 1200°C and 1300°C. The results (Table VI) reveal that slags produced from 06 contained up to 18% MnO, whereas slag from reduced 03 contained less than 1% MnO (Table VII). It would appear from the different manganous oxide contents of the experimental slag products that the knife and digging stick were made from an ore similar in composition to 03 rather than 06. The high manga-

Table VI: Analysis (wt.%) of Four Slag Phases from Ethiopian ore 06 reduced at 1300°C Light Phase FeO MnO CaO SiOz KzO Al z0 3 PzOs CuO S03 Total

Dark Phase

1. 87.1 8.8

2. 83.4 9.6

3. 85.1 9.9

4. 81.8 8.2

1.8

3.4

3.1

2.7

1.3

1.3

2.3

1. 46.2 16.9 0.3 27.0 0.9 8.3 1.9

2. 47.8 17.4 0.2 27.0 0.8 7.5 0.3 2.1

0.4 97.7

0.6 98.3

99.8

95.0

C 2 0: SiO z

103.1 0.007

101.5 0.011

3. 47.0 16.1 0.5 26.2 1.0 11.2 0.6 2.2 0.3 105.1 0.019

4. 47.1 16.3 22.5 0.5 7.8 1.0 95.2

Table VII: Analysis (wt.%) of Nine Slag Phases and the Iron from Ethiopian Ore 03 Reduced at 1300°C (EDAX)

FeO MnO CaO SiO z KzO Al z0 3 TiO z PzOs CuO Total CaO: SiO z

1. 36.6 0.3 33.4 18.5 6.6 1.1 1.1 97.6 0.009

2. 37.7 0.8 0.5 35.5 0.2 18.3 6.6

3. 36.6 0.8 0.5 35.6 0.2 19.8 6.7

4. 37.8 0.7 0.4 34.8 0.3 20.5 7.7

5. 38.0 0.9 0.4 33.4 0.1 19.2 6.6

6. 38.7 0.9 0.5 33.8 0.2 19.4 7.1

7. 38.3 0.8 0.5 36.9 0.2 19.8 7.2

1.1 100.7 0.014

1.1 101.3 0.014

1.2 103.4 0.012

1.0 99.6 0.012

0.7 101.3 0.015

1.1 104.8 0.014

8. 38.5 0.9

9. 67.0 0.3

35.8 0.1 19.9 7.5 0.9 0.5 104.1

17.7 13.5 5.2 1.0 104.7

Matrix: 0.3 P, 0.8 Cu, 0.3 Si, balance Fe.

44

JOURNAL OF METALS • November, 1979

nous oxide slag phases of reduced 06 also confirm manganese present in the ore concentrates in the slag phase, so that manganese is important in establishing the relationship between inclusion and ore compositions. The correlation between the compositions of furnace slag, inclusion phases in the Dimi iron artifacts, and the experimental reduction products produced from a known ore composition showed that the knife and digging stick were consistent with the ores of the region. The sections examined are thought to have been manufactured from a low manganese ore such as 03 and 04. The Dimi artifacts have a characteristic microstructure of ferrite and low carbon bands; and, as there was no marked deviation from the smelting procedure along the mountain range, a typical Dimi product would contain inclusions of low lime/silica ratio and high phosphorus content. Therefore, although the work of many artisans may result in the same type of microstructures, the combination of microstructural studies with those of inclusion composition analyses are very valuable in distinguishing the ironwork of different production centers. The study of Dimi ironwork, and also further experimental reductions of various types of iron ore,I' has demonstrated that artifacts made from different ores can be distinguished by their inclusion compositions. However, inclusion analyses must be combined with typographical and metallographic information obtained from the object. The maximum information concerning a particular smelting process and its products can be found by detailed analyses of related ore, furnace slag, and metal samples. In view of the lack of analytical data, future studies of such archaeological samples

would be very valuable. Once the metalwork of a production center has been characterized, it would be interesting to sample the ironwork (dated to the same period) of the surrounding region to establish the probability of the two being related, and hence, determining the extent of the surrounding trade. Until more data concerning inclusion studies becomes available, the study of related ore, slag, and metal is more profitable than studies designed merely to test the similarity or dissimilarity of groups of inclusion compositions. Metallographic studies of historical iron objects have so far been primarily concerned with describing the artifacts and making limited comment on the microstructure and heat treatments. The metallographic techniques described, combined with in-situ studies of inclusion compositions, may enable a smelting system to be more closely specified and hence be a significant development in determinating the provenance of iron artifacts. ACKNOWLEDGMENTS This research was carried out while the author was a research student working with Dr. J .A. Charles in the Department of Metallurgy and Materials Science, University of Cambridge, England. The author would like to thank Dr. Charles for his guidance and encouragement and Professor R.W.K. Honeycombe for providing laboratory facilities. The research was supported by the Science Research Council, the Mary Ewart Travelling Scholarship from Newham College, Cambridge, and the British Universities' Student Travel Association, and the British Federation of University Women. References

1. B.G. Trigger, "The Myth of Merae and the African Iron Age," AI, His. Studies II, 1 (1969).

2. R.F. Tylecote, "Iron Working at Meroe, Sudan," His. Met. Soc. IV (1970) p. 62·72.

J.

3. M. Posnansky and R.J. McIntosh, "New Radiocarbon Dates for Northern and Western Africa," J. Af. His. XVII 2 (1976) p. 161·195.

4. D. Williams, Iron and Image, London, Allen Lane, 1974. 5. J.H. Greenburg, "Linguistic Evidence Regarding Bantu Origins,"'

J.

AI. His. XIII (1972) p. 189·216.

6. C. Ehret, "Bantu Origins and History: Critique and In-

terpretation," TransAfrican J. His. 2 (1972) p. 1-12. 7. P. Schmidt and D.H. Avery. Science, 201 (1978) p. 1085· 1089.

8. D.W. Phillipson, "The Chronology of the Iron Age in Bantu Africa,"

J.

AI. His. XVI, 3 (1975) p. 321·342.

9. R. Oliver, "The Problem of the Bantu Expansion," J. Af. His. VII (1966) p. 361·376.

10. H. de Contenson, "Les Fouilles De Haoulti en 1959," AnnaIIes d'Ethiopie, 5 (1963).

11. P. Duncombe and E. Jones, "Electron Probe Analysis and Easy to Use Computer Program for Correcting Quantitative Data," Tube Investment Technical Report, May 1970. 12. R.F. Tylecote, Metallurgy in Archaeology, Arnold, 1962, p. 244.

13. J.A. Todd, "Studies of Primitive Iron Technology," PhD thesis, University of Cambridge, England, June 1976.

ABOUT THE AUTHOR Judith A. Todd is a research engineer in the Department of Materials Science and Mineral Engineering, University of California at Berkeley, and is working on the design of low-alloy steels for thickwalled pressure vessels. She obtained her BA in materials science (1972) and PhD in metallurgy (1976) from the University of Cambridge, England. In 1976, she joined the Department of Metallurgy and Materials Science, Imperial College, London, as a research assistant and worked on an in terdisci p linary engineering-metallurgy program called "A Fracture-Mechanics Approach to Creep Crack Growth in Low Alloy Steels." In 1978, she was a research associate at the State University of New York at Stony Brook, teaching a National Science Foundation program for "Women in Science and Engineering," and carrying out research on aluminum hydride. She has been a project leader at the University of California, Berkeley, since January 1979.

The Metals Profession NRC Announces 1980 Programs for Postdoctoral Research The National Research Council announced its 1980 Research Associateship Programs to provide postdoctoral opportunities for scientists and engineers in the fields of atmospheric and earth sciences, chemistry, engineering, environmental sciences, life sciences, mathematics, physics, and space sciences. NRC Research Associates conduct research on problems largely of their own choice in selected federal research laboratories at various locations in the United States. The programs are open to recent recipients of the doctorate and, in many cases, to senior investigators also. Some programs are open to foreign nationals. The basic annual stipend is $18,000 JOURNAL OF METALS • November, 1979

for recent doctoral recipients. Higher stipends will be determined for senior awardees. Awards include relocation allowances and limited support for professional travel during tenure. Awards generally are for one year periods.

Applications must be postmarked by January 15, 1980. Contact the Associateship Office, JH 608-D3, National Research Council, 2101 Constitution Avenue N.W., Washington, D.C. 20418; telephone (202) 389-6554.

Journal on Surface Analysis Launched Heyden & Son, Inc., announced publication of a new bimonthly journal, Surface and Interface Analysis. The journal will provide a focus for research into the development and application of techniques for the characterization of surfaces, interfaces, and thin films, the firm said. Equal emphasis will be placed on the application of techniques

for materials characterization and relavant fields. Heyden & Son is offering complimentary copies of the first issue to all materials scientists and analytical chemists. Contact Heyden & Son, Inc., 247 South 41st Street, Philadelphia, Pa. 19104; telephone (215) 382-6673. 45