An Early Iron Production Site with a Highly Developed Technology

Dehigaha-ala-kanda (KO. 14) at Alakolavava: An Early Iron Production Site with a Highly Developed Technology

SVANTE FORENIUS AND ROSE SOLANGAARACHCHI

This is a brief presentation of the preliminary results from the archaeological excavations at KO. 14, one of the sites chosen for archaeological excavations in the 1990 field season and continued in 1991. The Alakolavava, or Dehigaha-ala-kanda site (KO. 14) is one of the major sites where iron slag was found during the survey for sites in the Kiri Oya valley, in 1988. Different quantities were found at about twenty other sites, but slag from iron production was not distinguished from slag whichoriginates from forging. (This has been done later.) Further­ more, only one or a few pieces of slag were found at many of these sites. KO. 14 is also one of a cluster of sites (mostly pottery sites, but also including another iron production site, KO. 12, and a monastic site, KO. 17) grouped around the confluence of the Dehigaha-ala with the Kiri Oya. It is situated about 8.5km south-east of the Sigiriya rock and about 1.5km from the present Alakolavava village. A preliminary report on the site and a sketch map of the area has been published (Manjusri 1990). Among the important site indicators were a ’’large ac­ cumulation of iron slag spread so intensively over a vast area”; two rock surfaces containing a series of conical holes (KO. 14A, KO. 14L) and a sequence of rock shelters (KO. 14C, KO. 14E, KO. 14G, KO. 14H and KO. 141) lying on the eastern side of the site; and possible structural remains associated with these caves. As a result of the observations made during this survey, this area was earmarked as a future excavation site: "The site marks itself out as a priority for further studies in the future. Was this an iron production site? Was it a monastic site? Did it serve both purposes at one and the same time? Where did the ore come from? These and many other ques­ tions must be answered in the not too distant future" (Manjusri 1990:118). Archaeological excavations were planned to com­ mence in 1989, but could not be implemented until the next year (see Mogren: "Objectives" this volume).

Figure 10:1 Site plan of Dehigaha-ala-kanda.

Figure 10:2 Map showing location of Dehigaha-ala-kanda.

Particular attention was paid by the SARCP research group to the function of the conical holes. One of the first persons to report these "ancient rock-cup-marks" was Henry Parker (1984 (1909):221-234) who had assumed them to be con­ nected with the activities of stone masons, engaged in the construction of monasteries. However, Parker had not ob­ served or reported the existence of iron or iron slag associated with such conical hole sites, nor did he in any way connect these with the iron production process. It was the SARCP team that first conjectured them to be in some way used to grind iron ore before smelting, or perhaps to support the grinding machinery (Mogren 1990:58-59). Significantly, in discussing the possible function of these holes, Parker had recalled the chekku, or oil mill. It was the heavy slag deposit at the Alakolavava site that seemed to point to the conical holes being connected with iron production; and in the adjoining soil layer west of KO. 14A, small pieces of quartz, iron ore and potsherds were observed. Mogren points out that there are other sites in the Kiri Oya valley showing the same spatial connection between iron slag sites and conical holes (Mogren 1990). In 1990-1992, more sites showing traces of iron production have been found in the region, in a setting that seems to strengthen the supposed connection further. Their spatial context also seems to imply a monastic connection, even though KO. 14 itself cannot be considered monastic any more. This issue will get further in-depth treatment in a forthcoming monograph on the KO. 14 site, due to be published as soon as all the analyses are fin­ ished. By that time it will also be possible to discuss Dehigaha- ala-kanda in an international context. The present article should be regarded only as a brief preliminary report.

Figure 10:3 Plan of the trench with excavated furnaces.

The setting

The Dehigaha-ala-kanda site is named after the low mountain on which it is located. The site is surrounded by chena land in the north, west and south-west. To the south is a small, peren­ nial stream, Dehigaha-ala, which is a tributary of the Kiri Oya. The confluence is situated between one and two hundred metres east of KO. 14. The iron production site covers an area of irregular shape, about 100m from north to south and 70m from west to east. The ground slopes towards the west and south-west. On the surface of the site, iron slag is spread extensively and con­ centrated in at least half a dozen large slag heaps, especially in the north-western, western and south-western sections. Some slag pieces are strikingly massive, indicating an advanced stage in the bloomery smelting technique, others are porous, with grass impressions, indicating a tapping of slag on to a wet grass bed some time during the process (Noreus 1990 un- publ.).

Objectives and realization

The aims of the excavation in 1990 were to try to clarify what the slag represented, to determine the state of the site and date the remains. It had earlier been decided that the slag had to originate from iron production in bloomery furnaces. A temporary bench mark (TBM) with an assumed height of 200m above mean sea level, was established on a small rock to the south of the southernmost visible slag heap (Mound A) and the site survey was carried out in relation to this TBM. A centre point of a free-standing grid system was marked at a designated spot to the south-west of the same slag heap. Initial attention was focused on this southernmost, visible slag heap. It seemed most likely that if there were any furnaces to excavate, they ought to be found on its eastern upslope. The ground was flat and covered by a topsoil deposit. A small depression in the ground looked to be a convenient place to start the search for furnace remains. A 4x4m area (later ex­ tended) was chosen for the first test excavation. In the middle of this trench, the remains of two well preserved furnaces (named 6A and 6B) of surprisingly large dimensions, were unearthed. It was also decided to dig a trench through the slag heap in the slope (Trench 1) to get a rough estimate of the amount of slag. In 1991 the excavated area was extended to the south, where three more furnaces were found. Only two of them (labelled 26 and 27) could be excavated. The third furnace (labelled 25) was left unexcavated, as a big tree was standing on top of it. In this second season of excavations, two more trenches were dug through slag heaps, one at the northern end of the slag heap trenched in 1990 (Trench 2) in order to study the degree of soil deposition over the slag heaps of the site (a process that has made estimates of production quantities very difficult); and another one through a very prominent slag heap

Mound B) in the north-western part of the site area (Trench 3). A few test pits have also been put down in the site. One of these is adjacent to one of the rocks with conical holes. This was done with the intention of obtaining data regarding waste materials, deposited in soil layers adjoining the rock, from the process of pounding or grinding iron ore, carried out on the rock itself.

Figure 10:4 Furnaces 6A and 6B seen from the west. Photo: Mats Mogren.

THE FURNACES

Position of the furnaces

The furnaces were placed above a slope where slag and other waste material had been thrown out. All the excavated fur­ naces were constructed in pits that had been carved out in the bedrock. The bottom end of the furnaces lay about one metre below the top of the bedrock. In front of each furnace there was a relatively flat area, about one square metre in extent. The pits were oval in shape, a total length of 1.5m (includ­ ing the furnace) and a little less in breadth. In the two pits excavated in 1990, the rock had been carved out in such a way that it almost looked like rounded steps, on the side towards the slag heap. The reason for this might have been to make it easier to move the bloom and waste material out of the pit.

Construction and shape

The furnaces themselves were constructed of clay and stone slabs. The back walls were almost vertical and embedded in the bedrock. Even the side walls were in places directly con­ nected with the carved-out rock. On each side of the furnace shaft, upright stone slabs, upto 0.8m in height and 0.15m thick, were placed directly on the rock at the bottom of the pits. The slabs were for stabilizing the walls, but could also have had the function of carrying the weight of the super­ structure. Clay lining was observed on the slabs, on the sides that faced the furnace shafts. In some cases, even the outer surface was covered with clay. Starting from the top of the stone slabs, the furnace shafts grew narrower upwards. The back walls had a slightly bent shape in plan and the change-over to the side walls was soft. No remains of the front walls were found in situ. This was not surprising, as those parts of the furnaces had to be broken to make it possible to get the bloom (spongy lump of iron and slag) out. As the furnaces were used over and over again, it was also necessary to remove the slag that collected at the bottom. Furthermore, the back and the sides of the shaft often had to be repaired before a new campaign could begin. In this connection, it could be mentioned that the stone slabs also functioned as a breakage limit, to protect the side walls when the front was torn down. In all the excavated furnaces the front walls, as mentioned, were missing. However, the slag that was left at the bottom of the furnaces indicated that the front walls had been rectilinear in plan. Like the other walls, the front seemed to have been vertical up to the top of the stone slabs while the upper part leant backwards, away from the smelter working in front of the furnace.

Approximate dimensions of the furnaces:

Width at the bottom: 0.8-0.95m

Width 1.5m above the bottom: 0.50m

Depth at the bottom: 0.40-0.60m

The back wall in the best preserved furnace was 1.6m in height. The total height of the furnaces had probably been about 2m. The side walls were 20-40cm thick (including the stone slabs) at the lower half and about 5cm at the top of what remained of the furnaces. The furnaces were relatively broad, compared to the depth, i.e. the distance from the front to the back wall. Obviously the ironmasters at Dehigaha-ala-kanda had tried (and succeeded) in achieving optimal conditions in the furnaces. The arrange­ ment of eight parallel tuyeres (see below) made it possible to control the temperature in order to reach the highest level of accuracy in the smelting process.

Figure 10:5 Furnace 27 in profile.

Figure 10:6 Reconstruction drawing of furnace 6B at Dehigaha-ala-kanda. Drawing: Malinga Amarasinghe.

Air supply

In the first excavation season (1990) a number of pipes for the air supply, so-called tuyeres, were found. The tuyeres, made of burnt clay, are cylindrical in shape. The outer diameter is about 9cm and the hole through which the air was blown into the furnace is about 3.0-3.5cm. The length is unknown, as the fronts are partly burnt away on all the tuyeres that were found. The length would probably have been about 25cm. Analyses of the clay in the tuyeres and clay from the furnace walls prove that different types of clay have been used. The tuyeres clay is more heat-resistant than the furnace walls (Dag Noreus, personal message). Tuyeres similar to the ones found at KO. 14 have been observed at least on one other site in the Kiri Oya valley (KO. 26 Kosgaha-ala). The uniform design and the distin­ guished properties suggest that the tuyeres were made of a carefully selected clay, perhaps in a factory where this man­ ufacture was a speciality. During the second excavation season (1991) some details were found that were crucial for the understanding of the furnace construction. In the previous season, some of the wall

Figure 10:7 Profile drawing of excavated furnaces seen from the west.

pieces found had imprints of two tuyeres, hinting that these had been attached to the furnaces in pairs. In front of two of the furnaces excavated in 1991 were found rather large pieces of the lower parts of the front walls. These finds were very interesting, as they had similar imprints and were found toge­ ther with tuyeres ‘in situ’. The imprints proved that the tuyeres had been attached parallel and close to each other (like organ pipes). It could also be proved that the front wall had been filled up with tuyeres, from left to right. A total of eight tuyeres had been used at the same time in each furnace. This seems to be a very special arrangement for the furnaces at Dehigaha- ala-kanda. Unfortunately these finds did not disclose at what angle the tuyeres entered the furnace. A high yield and evenness in the quality of the produced iron among other things, depend on the possibility of control­ ling reducing conditions in the furnace. The temperature is one of the most important factors. Both the amount of air and the airflow have an influence on the temperature. The excavated furnaces have shown that great attention had been paid to solve the difficulties connected with air supply. The eight parallel tuyeres gave the ironmasters at Dehigaha-ala-kanda the ability to distribute the air equally to all parts of the hearth and to avoid cold zones in the furnace. That natural draught had been utilized for the air supply is out of the question. The damage on the lower parts of the back walls indicated that the pressure with which the air had been blown into the furnace could only have been achieved by using bellows. The bellows were probably placed in or beside the pits in front of the furnaces. No remains of the bellows were expected to be found, as they must have been made of organic material. Neither the design of the bellows, nor whether they were foot or hand powered, is known. On the left side of each of the pits in front of the furnaces, there was a cut in the rock a couple of centimetres deep. A guess is that these cuts probably served as holders for some construction that would have made the pumping easier.

THE ORE AND ITS PREPARATION

Occurrence

Iron ore has been found more or less all over the site. The main type is magnetite (Fe3O4. NH2O). One of the largest concentrations of ore was found in the north-eastern part of the site lying below the rock with the five conical holes (KO. 14A). Pieces of ore were also found in the filling around the furnaces and on the ground surface. No thorough exploration for ore deposits has taken place yet. As the magnetite is a type of rock ore, one would assume that the ore has been extracted by mining somewhere in the neighbourhood. There could, for instance, have been ore dep­ osits close to the caves on the east side of the mountain.

Conical holes

The conical holes at KO. 14A are 15-20cm wide at the top, and 7-15cm deep, with a rounded bottom. The exact function of the holes is still unknown. Below some of the rocks, crushed quartz and pieces of iron ore were observed, together with potsherds. In 1988 the strong connection between the conical holes and the quartz flakes was not understood, and the quartz flakes were interpreted as remains of a prehistoric camp site (Manjusri 1990). As stated before, it can now be considered as ascertained that the holes were connected to grinding, or pounding of the ore, in order to separate quartz and other unwanted material from the ore, and to break up the ore into a size suitable for the smelting process. The grain size of the ore is one of the factors of importance in efforts to find the proper reducing conditions (Noreus 1990 unpubl.). How­ ever, it has not been ascertained whether the conical holes have been used as mortars, or formed by some kind of grind­ ing machinery. The latter is more likely.

Figure 10:8 Three tuyeres found with fragment of a tuyere stand. Photo: Mats Mogren.

Figure 10:9 One of the conical holes found at Dehigaha-ala-kanda. Photo: Mats Mogren.

Other activities connected to the smelting process

Iron production is a process that involves several stages aro­ und the smelting itself. Many of these stages did not take place at the iron production site. Ore, wood and clay for the furnaces could have been brought from sites at different dis­ tances from the furnace sites. Charcoal burning probably took place at the point where the wood was collected . In a land of chena cultivation, traces of these activities could be hard to find. Roasting and pounding of the ore could have been ex­ ecuted at the production sites. At KO. 14 pounding has prob­ ably taken place in connection with the conical holes (see above). Roasting is a method of eliminating impurities and reducing the water content in the ore. No roasting sites have been observed at KO. 14. It must be kept in mind, though, that the excavated area is very limited and that the research con­ centrated on the furnaces and slag heaps. After the smelting, the bloom must be cleansed from the slag. The bloom was usually placed on a stone that acted as an anvil, and then the slag was pressed out by hammering with a sledge-hammer. During this process the bloom must be heated up repeatedly, in a smithing pit. It is significant that neither smithing pits nor obvious anvil stones were found.

Yield and volume of the production

Without a more detailed knowledge of the process, reliable estimates of the iron production are not possible. From what we know today about the process and the amount of slag at KO. 14, it is impossible to calculate the amount of iron produced at the site. We know that the slag at the site has to be counted not in tens or hundreds, but rather in tens of thousands of tons. A problem is that a great quantity of the slag deposits are hidden under soil transported by erosion. The yield from the smelting process is dependant, among other things, on the properties of the ore, the charcoal and the flux (if added). Other important factors are the temperature and the possibility of controlling it, and the furnace construc- tion/type. The skill and experience of the iron smelters is also a factor to be taken into consideration. A prerequisite to calculating yield is that the ore/slag ratio can be settled. Considering the amount of slag, the good qu­ ality of the ore and the highly developed furnace construction at Dehigaha-ala-kanda, a very modest estimate of the output of raw iron could be placed at thousands, or even tens of thousands of tons. However, it must be kept in mind that as much as 80% of the iron could be lost during the transforma­ tion into finished implements. Even if this is taken into con­ sideration, the iron production at KO. 14 must be regarded as a very large-scale ‘industrial’ production, executed by profes­ sionals.

Locating factors

At least three concurrent factors could have determined the choice of Dehigaha-ala-kanda for iron production. Large am­ ounts of wood were used in different stages of the production, like roasting the ore, pre-heating the furnaces and as fuel (in the shape of charcoal) in the smelting process. The jungle provided the iron producers with wood (even though the landscape must have become quite barren after some time). The ore was probably collected in the vicinity. The iron oxide content in the ore that occurs in the area could have been a factor in determining why this site was preferred over other sites. The access to perennial water is another important fac­ tor. Clay for the construction of the furnaces should be men­ tioned as yet another necessary raw material. Clay deposits should be found rather close to the production site.

Pottery and 14C-datings

An attempt was made to date the site through the pottery found at Dehigaha-ala-kanda, by comparing it with that foundat Sigiriya and the Jetavana, Abhayagiri and Gedige sites in Anuradhapura. The pottery found at Dehigaha-ala-kanda can be categor­ ized into three main groups: Black-and-Red Ware, Black Wa­ re and Red Ware. The Black-and-Red Ware sherds are of a coarse ware, of a very poor level of production; while the Black Ware, in contrast, is a fine ware (Jayaweera 1990). The Red coloured pottery is in the form of goblets (kusalana), tray-bowls (tati), cooking utensils (atili, appalla and mutti) water carrying vessels (kalaya, kotala, mutti) and storage ves­ sels. These display a very high standard of manufacture. Among these, the tray-bowls occupy a very special place, having been turned out with a very high finish. They fall into two categories: those with and those without rims. The vessels without rims can be assigned to an earlier period than those with rims. They belong mostly to the Black-and-Red Ware familiarly associated with the megalithic tradition and the pro- tohistoric-Early Historic period (PHEH). Tray-bowls of this type, similar to those belonging to the early Anuradhapura period, discovered from Abhayagiri and Gedige (Kuna 1972) can be seen at Alakolavava. Potsherds of this type are stra- tigraphically connected to the gravel layer beneath the slag heaps. The rimmed tray-bowls are similar to those found at Ab­ hayagiri (Kuna 1987) and Jetavana (Ratnayake 1984). It seems clear that this Alakolavava pottery has some resemblance to that of the early Anuradhapura period. The fact that these types were not found at Sigiriya is noteworthy. It is possible to suggest, tentatively, that the pottery found at Alakolavava points to an evolution from the protohistoric - Early Historic (PHEH) transition to Phases 1 and 2 of the Early Historic period (EHP1, 2). Since the pottery found in these contexts does not extend to that of the period of the Sigiriya kingdom and does not appear after about the 4th century AC at Anuradhapura, it can be assumed that the fur­ naces also belong to the pre-Kasyapan epoch in the Sigiriya region. These assumptions based on pottery were later more or less confirmed when compared with the 14C-datings from the site (see fig. 10:10). Seven radiocarbon samples have been analysed at The Svedberg Laboratory, Uppsala University.

CONCLUSIONS

result of an indigenous development from a protohistoric stage? Who controlled the production, the forging and the distribution? What was the purpose of the production? Was the iron intended for a local, a distant, or even a foreign market? Was the iron reserved for special social groups in society? Was the development of the iron production caused by warfare or, for instance, by agricultural needs? What was the economic profit and who extracted the surplus of the production/distribution? What consequences did it have on the society?

Figure 10:10 Dehigaha-ala-kanda. Radiocarbon dates. Calibrated age ranges from cumulative probability, one sigma (68.26%). CalibETH 1.5b (1991).

The high technological knowledge that the ironmasters at Dehigaha-ala-kanda must have possessed, indicates that the experience of producing iron goes far back in time. To decide whether the entire technique, or parts of it, were imported from abroad, or developed within the country, much more research is needed.

The excavations at Dehigaha-ala-kanda could be regarded as a point of departure for a new era in Lankan archaeo-metal- lurgy, where archaeologists co-operate with natural scientists in the study of ancient iron production. The dimensions of the production, as well as the technique that have been used, are of the greatest interest in more than a Lankan perspective. There is no doubt that the widespread remains of early Lankan iron production are a valuable source material that could take the archaeo-metallurgy of the entire world several steps forward.

REFERENCES

Kuna, M. 1987. Local Pottery of Anuradhapura: A Way to its Classification and Chronology. Pamatky Archeologicke LXXVIII: 5-66.

Manjusri, M. 1990. Dehigaha-ala-kanda (KO. 14): a prehis­ toric habitation site, a monastic rock-shelter site and an iron production site in the Kiri Oya Basin. The Settlement Archaeology of the Sigiriya-Dambulla Region. Bandar- anayake, S., M. Mogren and S. Epitawatte (eds.). Colom­ bo: PGIAR.

Mogren, M. 1990. Project Strategies: Methodology and Per­ spectives. The Settlement Archaeology of the Sigiriya- Dambulla Region. Bandaranayake, S., M. Mogren and S. Epitawatte (eds.). Colombo: PGLAR.

Noreus, D. 1990. Assessment of iron production sites in the Sigiriya area and their importance in relation to archaeo- metallurgy worldwide; and of the capacity of the Institute of Fundamental Studies (IFS) in Kandy, for doing the relevant slag analysis. Unpublished Report. PGIAR Ar­ chive.

Parker, H. 1984 (1909). Ancient Ceylon. New Delhi: Asian Education Services.

Ratnayake, H. (ed.) 1984. UNESCO - Sri Lanka Project of the Cultural Triangle Reports: Jetavanaramaya Project, Anuradhapura. First Archaeological Excavation and Re­ search Report (January-June 1982). Colombo: Central Cultural Fund, Ministry of Cultural Affairs.