In the Melsvik Stone Age chert quarries near Alta in Northern Norway there are dozens of extraction marks that are difficult to explain by other ancient techniques than fire setting. Hence within the Melsvik archaeological project, run by the University Museum of Tromsø, last week we experimented with fire in order to substantiate that it actually formed an important method of breaking loose small and big pieces of stone. The idea was that it is not necessary with big fires and high temperatures, but that small, controlled “bonfires” are enough to create high shear stress and cracking. In this way high temperatures greatly reducing the quality of the chert for tool making are avoided. It works! Here’s a preliminary report with video.
A little video of the fire setting experiments
The Melsvik chert quarries are so far the most important ones discovered in Northern Norway. They may be dated all the way back to the so-called “pioneer phase” around 9500 BC, not long after the ice sheet of the last Ice Age retreated up north. But the quarries were particularly in use in the Early-Middle Mesolithic (7-8000 BC), providing material to knives, arrowheads, scrapers and so on for the region. Archaeological excavation was initiated in 2012 following plans for building a new highway right through parts of the site. Though preservation of the whole site would certainly have been the best, the parts that will soon disappear offer a great opportunity to undertake experiments with ancient extraction techniques, in particular fire setting.
The chert deposit is situated on a small hill, forming thin layers (usually less than half a metre thick) above and within Precambrian dolomite (from stromatolites). This peculiar geology, where chert is often “draped” around dissolving dolomite (“karst”), implies that there are often hollows between the chert and the dolomite. Thus, at places the chert seems to stand under quite some tension, which we can test by banging it with big stones: The feeling is that they “come back” to you (recoil) and hence breaking bigger pieces loose with such a brutal technique is basically out of the question. In addition, the chert is extremely hard, compact and tough. Though intensive cleavage in several directions characterise the deposit, there are barely natural, open cracks and fissures, making it almost impossible to extract stone by using hammerstones and bone (and stone) wedges. On repeated blows with all kinds of hammerstones, which we have tested at length, what you usually get is pulverised stone and the one or the other small piece, mostly unsuitable for tool making. Stone Age man was thus bound to employ more efficient techniques.
Evidence of fire setting
On looking at the horizontal to slightly sloping, excavated quarry faces there are at least 30, probably 50 or more, round, shallow depressions measuring about 0.5 to 1 metre across. They usually have convex surfaces at the bottom, which is a strong indicator that fire was used in their creation. They are certainly not a natural phenomenon. In addition, beside and below zones of such depressions there are rather thick layers of broken-up chert pieces and chert gravel, all with sharp edges, but without traces from direct, man-made working, such as bulbs of percussion. Such layers ought to represent the waste from fire setting. Charcoal from burning is difficult to find, but soot layers are not unusual. In this respect it should be recalled that Stone Age charcoal is generally not very common at these northerly latitudes due to poor preservation conditions.
An additional indicator of elevated temperatures is a range of sharp-edged chert fragments with a dull, white colour. The natural colour of the chert ranges from shiny greyish white to bluish and purplish. It becomes dull white upon exposure to fire above, say, 300-400 centigrade, which we have tested by burning it in bonfires. This is a result of microcracking leading to increase in surface area, which alters the light reflection properties of chert.
Fire setting experiments
With all these indications of Stone Age fire setting, we set out to replicate the method. What we had at hand were:
- Birch wood for burning. Pine and mountain ash were also available in the Mesolithic up north, but not oak, which give higher burning temperatures.
- Hammerstones, stone wedges and fire-hardened bone and antler wedges – all in different sizes – to aid in removal of cracked stone. All tools were found or produced specifically for the project; we did not use any of the masses of hammerstones found during excavation.
- A thermocouple to measure temperature. Unfortunately, it didn’t arrive in time, so we were only able to control the temperature in one of our experiments.
Over four days we made five experiments, each lasting from about one to three hours:
- One single fire at a slightly sloping chert surface.
- First two, then four fires around a small, elevated chert outcrop.
- Two fires along a vertical face.
- First two, then four fires at and around a small, elevated outcrop with a hole down to underlying dolomite.
- Two fires at a slightly sloping surface.
The short-lived, small fires
Each fire had a diameter of about half a metre across, generally consuming some 10-20 small birch logs. The objective with two or more fires beside each other was to create shear stress in the area in between and simultaneously keep the temperature low in these areas. Such fires performed the best, both at sloping surfaces where we knew the rock was under significant natural tension and around slightly elevated outcrops. The chert usually started to break at the surface after only 5-10 minutes, whereas it usually took some 45 minutes to one hour to create deep, lateral cracks more than 15 cm below the rock surface, for example between two fires.
Temperatures reached and material achieved
The temperature just below the centre of the fire reached 4-500 degrees centigrade after 20-30 minutes, whereas it was in the range of 40-60 degrees at depths of 15-20 cm after about one hour. We could control this by placing our hands on the chert after removal of broken-up material. After the initial surface cracking, the chert typically developed parallel, lateral cracks further down, breaking loose flakes with a thickness of 3-5 cm, weighing from a few hundred grams to a kilo or more. However, in one experiment (2) we were able to break loose a block weighing some 20-30 kg and with a maximum thickness of about 15 cm. Of course, a lot of smaller, sharp fragments were also produced.
Sounds of cracking stone
In cases where the rock was under significant natural tension, cracking proceeded with much sound. We could literally hear how the rock came alive. Surface cracking sounded like making popcorn, with small chert fragments jumping up to two metres into the air. Formation of deeper cracks was followed by lengthy “krrrrks” as tension was released. But the video give a better indication of the sounds!
Fire setting at vertical walls
Fire setting at vertical chert faces (3) was not very successful. One reason is that it is difficult to transmit heat to a vertical face without building protective walls around the fire, the other that the places we selected did not seem to stand under much natural tension. We were able to crack the rock, but we only obtained a few pieces useful for tool making, the rest being too small.
On looking at the distribution of fire setting marks in the quarry, it is evident that they are most common at horizontal to slightly sloping surfaces. However, it may well be that the method was also employed at vertical walls; we just haven’t figured out how yet. What we probably can say, is that protective walls were not built in the quarry in the Stone Age. This is because we have not found much fieldstone with clear fire marks during excavation. Such would have been used to build walls.
Cooling by water and removal of material
We also tested the effect of rapid cooling pouring water over the hot surfaces. Generally, it didn’t seem to have much effect, other than giving us swift access to the fire set places for removal of flakes and blocks. Removal was undertaken using stone and bone/antler tools only, and it worked excellently. Banging a little bit here and a little bit there helped in freeing cracked stone, and when this was not enough banging in simple wedges (stone/bone/antler) did the job.
Good and bad stone
Clearly, the uppermost cracked stone was totally unusable for tool making. However, the deeper layers that had been exposed to temperatures of less than perhaps 200-100 degrees – often even less – seemed perfectly suitable for making the kind of relatively small tools that are found during the excavations in and near the quarry – and at other Mesolithic sites in the region. However, we cannot be entirely certain of the quality of the raw material produced before it is tested. Replication of tools will be undertaken at a later date.
With our limited knowledge and experience, as compared to Stone Age people, the most important result was perhaps not that it is indeed possible to produce seemingly suitable stone for tool making by fire setting, but that the final forms of our fire set surfaces greatly resembled the old ones in the quarry. Moreover, waste from our experiments – sharp-edged fragments of all sizes up to several kilos – very much resemble what is found in the old waste heaps. These are thus crucial indications that fire setting was actually used at the Melsvik chert quarry 10.000 years ago. Stone Age people used their brains in extracting stone – not brute force!
Anja Roth Niemi, Tromsø University Museum, project leader at Melsvik, for wonderful days at Melsvik, and not least to archaeologists Audun Berg Selfjord and Rudi Mikalsen for great help with the experiments. Also thanks to all other members of the project for support.
About the Melsvik archaeological project (in Norwegian):
Blog entry at my website from last year, hypothesising the use of fire setting at Melsvik, with many references: