The Fire in The Flint: Arrowhead Production and Heat Treatment

In 2004, David Clark wrote of the concentrations of arrowheads found at a number of sand dune sites in Scotland that 'at present we have no narratives that seek to explain, or even acknowledge this situation' (Clarke 2004, 47). This article seeks, through an exploration of the manufacturing processes involved in producing a particular arrowhead from Luce Sands, Wigtownshire, to provide some of the detail to allow the construction of a wider narrative.

Figure 1

Figure 1: Sketch map showing the location of Luce Sands

Luce Sands is an area of dunes situated at the head of Luce Bay on the south-west peninsula of Scotland (Fig. 1). Within recorded history an area of rough grazing, it has in recent years been under the aegis of the Ministry of Defence. It was, prior to this, a favoured collection area for antiquarians, with large quantities of pottery, lithics and metalwork being exposed by the shifting of the sand and the erosion of the soils.

Undertaking work on the lithic assemblage from Luce Sands involved examining the archaeological assemblages from the dunes and making an assessment of the resources locally available to prehistoric flint knappers. Pebble flint was collected from Luce Bay, less than a mile in distance from the dune area where the concentrations of prehistoric material occur; 1318g of material was collected in 10 minutes and, in less than an hour, 3281g of material had been collected. This collecting was repeated twice on subsequent occasions, with very similar results. The material was shown to the flint knapper John Lord, who very kindly agreed to test the pebbles for their potential as a raw material.

The results indicated that 80% of the flint pebbles were of sufficiently good quality to manufacture a wide range of artefacts. From one 252g pebble were obtained: one large round scraper, two medium side scrapers, two medium end scrapers, two thumbnail scrapers, one small side scraper, one notched tool, one backed blade, two arrowheads and five other useable flakes. Together these weighed 116g, or 46% by weight of the pebble.

A notable quality of this flint was its extreme toughness. It required hard hammering and indeed one quartzite hammer broke during the manufacturing process. Pressure flaking was extremely difficult.

The debitage produced paralleled closely that found on the sands, where the number of split pebbles of good-quality flint, from which only a piece had been removed and the rest discarded, indicated that flint was not in short supply during the prehistoric period.

Also examined was a small assemblage of lithics from an excavation at Knocknab, an area on the Luce Bay dunes. Among the artefacts were two pieces of a leaf-shaped arrowhead, found separately, which proved to refit perfectly (Fig. 2). This arrowhead was very similar to many in the National Museum of Scotland (NMS) collection.

Figure 2

Figure 2: Arrowhead from Luce Sands

As part of the research into the Luce Bay assemblage, a programme of experimental knapping was undertaken to produce analogous arrowheads from local material.

It is necessary to employ different techniques in order to produce different kinds of leaf-shaped arrowhead. An edge-retouched only arrowhead may be produced fairly simply by striking a suitably thin flake and retouching both edges to create the size and shape required. To produce a finely knapped arrowhead with invasive retouch across both faces, on the other hand, it is necessary to strike a fairly thick, large blank from the core. The arrowhead will then be produced as a core tool with the blank being reduced on both its faces (John Lord pers. comm.). The initial reduction will be by hard or soft hammer. Once the blank has been suitably shaped and thinned, pressure flaking will be used to finish the piece. It is reasonable to assume that prehistoric flint knappers would have used very similar techniques.

The majority of the leaf-shaped arrowheads recovered from the Sands were bifacially retouched with invasive flakes rather than simply edge-retouched. It was decided to undertake experimental work to attempt to reproduce the thinness and fineness of these arrowheads using the locally collected pebble flint, and taking as a model a finely made arrowhead of translucent flint from the Knocknab excavation (see below) which was available for a direct comparison. This further work was intended to answer two specific research questions:

  1. how much more work was involved in producing an arrowhead with invasive retouch on both faces?
  2. was it actually possible to produce arrowheads of this quality using pebble flint and antler pressure flaking tools or had imported flint been used to produce these high-quality articles?

John Lord was asked if he would produce firstly, arrowheads made on flakes with bifacially retouched edges, and secondly an arrowhead as close as possible to the archaeological model.

The edge-retouched arrowheads were produced in something around ten minutes from the first selection of the flint nodule. They were small and elongated, one measuring 28 x 12mm with a thickness of 3.5mm and the other measuring 23 x 12mm with a thickness of 3.5mm. This put both arrowheads within Green's typology of very small arrowheads 4C r and u, types that are typically found in areas of poor flint availability (Green 1980). In other words, they conformed to Neolithic arrowhead types although they are markedly different from the broader types that dominate the Luce Sands assemblage.

To produce arrowheads of finer quality required the striking of a blank, and the arrowhead was made using this as a core tool. It took about 30 minutes to produce an arrowhead rough-out of the length and width of the archaeological examples. However, when it came to the thinning process, problems emerged. Despite John Lord's best efforts, the pebble flint was too unyielding to allow invasive retouching. Using antler tools the maximum length of flakes that could be removed by pressure flaking was around 4mm. A copper point was tried and this increased the maximum length of flake to about 6mm. This gave the arrowheads produced a plump profile and none were less than 4mm thick at their point of maximum thickness. Further, there were large areas on their faces that remained unretouched, in contrast to the archaeological example. Several attempts were made using a number of flint pebbles. However, the problem remained that the flint was persistently too tough to allow pressure flaking across the surface of the arrowhead (Fig. 3).

Figure 3

Figure 3: Arrowheads made from pebble flint

A number of possibilities were considered:

  1. it was possible that the prehistoric knappers were using particular flint pebbles of a finer flint than had so far been discovered;
  2. they may have been utilising imported flint which was more amenable to the thinning process; and
  3. the pebble flint may have been heat treated – a process that reduces the tensile strength of the flint, so rendering it less serviceable as a functioning object but making it easier to work.

The possibility that the raw material had been imported was considered. However, a large number of the Neolithic arrowheads from Luce Sands known from the various collections are not a match for either of the two kinds of flint, Antrim and Yorkshire, known to have been imported into the region. Furthermore several of them exhibit a slightly pinkish tinge, which was not found in any of the 200 beach pebbles investigated.

It was noted in Cowie's excavation report that 'a number of the flints appear to exhibit the colour and texture change characteristic of thermally treated flint'. These flints came from a sealed soil context and had not been subjected to sand polishing so that it was possible to detect the characteristic gloss imparted to the flint by heat treatment. They were, furthermore, found in association with a number of calcined flints and it was considered that a dump of 'heat fractured angular sandstone fragments' might have had a part in the heat treating process (Cowie 1996, 30).

As a result of all these considerations it was decided that heat treatment was a strong contender for the reason why Neolithic people were able to produce such finely knapped artefacts. It was therefore decided to examine what effect heat treatment would have on the Luce Sands pebble flint.

The experimentation undertaken thus far indicated that only pebbles weighing 100g or more were likely to yield flakes of a suitable size for the manufacture of arrowhead blanks. Smaller pebbles, while often of good-quality flint that could be used for small blade cores, would not produce the broad flakes required. Fifty of the larger pebbles, including two very large examples of over a kilogram in weight each, were utilised. These were knapped to produce a number of flakes that could be used as arrowhead blanks.

Ten of the pebbles, that is 20% – notably the smaller examples – although of good-quality flint, proved very difficult to work. Seven of these were included for heat treatment to ascertain whether this made the flint easier to access, without any further preliminary working beyond splitting open the pebble. Twenty-nine (58%) yielded one or more flakes suitable to act as blanks. The remaining 11 pebbles (22%) were either thermally damaged or of too poor-quality flint to be worked.

Of the two very large pebbles, one proved to be badly thermally fractured and useless for knapping purposes but the other yielded a number of good-quality large flakes. A number of these were included in the sample for heat treatment. Others were worked and indicated that these would yield only very small flakes. The pebbles showed a variety of colours, falling broadly into grey or brown. These results were very much in line with those from the earlier testing of pebble material (Fig. 4).

Figure 4

Figure 4: Blanks and flakes of pebble flint

The aim of this experiment was solely to investigate whether by using simple techniques that would have been easily available to prehistoric craftsmen it was possible to alter the material sufficient to render it suitable for making finely knapped arrowheads. Work undertaken by previous researchers has indicated that direct heat on flint is liable to have a deleterious effect on the raw material, with the flint cracking or crazing in a way that makes further knapping impossible (Rick and Chapell 1983 inter alia). Lee has demonstrated that the most successful heat treatment using early technology involved placing the flint within a good conductive medium and using a low, wide fire kept well fed to provide a concentrated heat (Lee 2001).

A circular pit about a metre in diameter and approximately 150mm deep was prepared and a bed of sand 30-40mm deep was laid in the base of the pit onto which the blanks and split pebbles were placed (Fig. 5).

Figure 5

Figure 5: Flint in the sand-lined pit

The flints were then covered over with a layer of sand 100-120mm deep and a fire of wood lit over the top of the pit. Sand provided a good medium for spreading the heat evenly. Moreover, within the dunes any hole dug would have been into a predominantly sandy soil, so this produced as nearly as possible the conditions in which the flints could have been treated in prehistory.

This fire burned for over 12 hours and then went out overnight. It was re-lit and burned for a further day. It went out and was allowed to cool naturally. A fortnight later the ashes were cleared away and the flint dug up (Fig. 6).

Figure 6

Figure 6: Flint after heat treatment

One of the pebbles had shattered. It was not clear if this was due to existing thermal fractures or solely as a consequence of the heat treatment. The rest showed some alteration. As flakes were removed, the scars had a marked lustre.

It became immediately apparent that the heat process had altered the knapping properties of the flint. It was far easier to set up platforms and the flint responded easily to flaking with an antler point. Flakes of 8-10mm were easily removed, and some flakes were detached that spanned the arrowhead and removed the opposing platform. The flakes also responded well to percussion with an antler billet.

John Lord observed that the flint now behaved like the Brandon series but was less brittle. The pebble flint retained the toughness that was one of its positive qualities. He felt its knapping capability was comparable with obsidian. Interestingly, however, one of the arrowhead blanks did break in the thinning process, snapping across as a result of the pressure, much as the archaeological example had done.

It was notable that the heat treatment had worked far more effectively on the blanks than on the nodules. Of the seven nodules fired, four shattered, one in the fire, another proved too tough to work and two produced good workable flint. The flakes from these were not consistent in their qualities. Presumably the differential intensity of the exposure to the heat as a result of where they were placed in the pit resulted in different results. Much of the flint appeared much paler as well as lustrous when the surface was removed, with several of the grey pieces appearing opalescent. From one of these originally grey pieces a near match was obtained for the Knocknab arrowhead.

As a result of this experiment a number of observations could be made.

Pebble flints from Luce Bay comparable in size and colour to worked pebbles from archaeological contexts could not be knapped to produce arrowheads of the fineness of archaeological examples prior to heat treatment. After treatment the flint became far more malleable and allowed the removal of long retouch flakes, making the production of fine flint work possible. There was, however, a far greater percentage of wastage, both from shattering pebbles and from broken blanks.

The heat treatment, as well as altering the behaviour of the raw material, also changed its physical appearance, making it rather more translucent and altering the colour. These changes were consistent with colour changes as noted by Lee (2001, 43)

It is not possible to ascertain by eye whether the archaeological examples have been heat treated. The colour range of the natural untreated flint is enormous and, as John Newberry has observed, 'a thick scraper and a thin leaf-shaped arrowhead made from the same homogenous single colour nodule may appear to be of different colours owing to the different light absorption and reflection properties of the two' (Newberry 2002, 5). The most obvious visual clue, particularly for flint, which is naturally rather dull, is the lustre that heat treatment imparts. However, this cannot be detected when the entire assemblage has been sand polished to a high sheen.

With reference to the flint from Cowie's excavation, none of the heat-treated flint showed the white calcification typical of burnt flint and it is doubtful whether any kind of treatment that involved the flint coming into direct contact with the heat source would have been effective (Cowie 1996). As has been seen, any over-heating, even when the flint was shielded from direct contact with the fire by several centimetres of sand, resulted in the flint shattering.

As heat treatment produced such a marked effect on the flint, a further programme of experimental work was undertaken to extend these findings. This had two main elements.:

  1. a more methodical examination of the beach material needed to be undertaken. To this end, further samples of pebbles of a suitable size were collected. Blanks were struck from these and clearly marked with a unique identifier. Half the blanks were heat treated in an identical manner to that outlined above while the other half were worked into bifacially flaked arrowheads without heat treatment. It was hoped that, given the quality variation from pebble to pebble, it would be possible to make a far clearer comparison of like with like in order to obtain a clear picture of exactly what difference heat treatment makes to the individual pebbles. The location of each piece within the pit was also recorded to ascertain whether there appeared to be a great difference as a result of where the pieces are placed; and
  2. a piece of Antrim flint provided by Alan Saville was treated in the same way, with a number of blanks being heat-treated and their knapping propensities compared to untreated examples. The knapping of the untreated flint would, it was hoped, establish whether Antrim flint could have been used to produce the very finely knapped arrowheads that are present in the Luce Sands collections without heat treatment having been needed. Of particular interest also was the colour changes within the Antrim flint that might be produced by heat treatment, and to what extent these would match with the colours in the archaeological specimens.

An innovation was now employed at the knapping stage of the experimental work. This was the use of a vice to hold the arrowhead while it was being worked. There had been discussion at the earlier stages concerning the very small size of some of the finely worked specimens, and of how difficult it would have been to maintain a grip on such small pieces while applying the necessary force to effect removals.

John Lord produced a simple vice to hold the arrowheads during the knapping process. This was simply a stick split at one end into which the blank was inserted. It was then held firmly, retaining the flint in place. This considerably improved the knapping, allowing long flakes to be produced that did not truncate in the way in which flakes do from an arrowhead held directly in the hand during working. Further, when held in the hand there was a tendency for the flint to twist in a way which made it impossible to maintain the correct angle for knapping, a problem obviated by the use of the vice. It was noted that this technological device would have been a simple adaptation of the technique used to haft an arrowhead onto a shaft (Fig. 7).

Figure 7

Figure 7: John Lord using a simple vice to hold the arrowhead blade

The results from Luce Sands pebble flint were very marked. Details can be given for the findings for one pebble which was, in its untreated state, particularly obdurate. This was a mottled brownish pebble. Heat treatment changed its colouring to a pale mushroom. The unheated portion was almost impossible to thin. Only short removals could be effected and after 30 minutes' work the arrowhead had been roughed out but showed step fractures across its surface. The first heat-treated blank responded far more easily to working. However, at an early stage of manufacture it broke, proving far more brittle than in the untreated state. A second blank was used and within 30 minutes a finished small arrowhead with bifacial flaking across the entire surface was produced.

The Antrim flint was much easier than the Luce pebble flint to work in an unheated state. It was possible to effect fairly long removals, the longest being 11mm. Nevertheless, John found it difficult to thin down the arrowhead enough to achieve an overall invasive retouch.

The heat-treated Antrim flint did not show much detectable difference in appearance. There was no marked colour change as a result of the heat treatment, although, when flakes were removed, the characteristic lustre was present. The flint was, however, much easier to work. Longer removals were possible, some of 14mm.

It would have been perfectly possible to produce bifacially flaked arrowheads from Antrim flint, but heat treatment did make the process both quicker and easier (Fig. 8).

Figure 8

Figure 8: John Lord holding the finished Antrim flint arrowhead

The outcome of this experiment seemed to be a confirmation that flint treatment was in use, certainly at Luce Sands. The majority of the local flint pebbles could not have been utilised to produce the finely worked arrowheads without this kind of preparation. Some of the colours present in the arrowhead assemblage would seem to be the result of heat treatment of the local material. They are not found in Antrim flint either before or after this material has been subjected to heat.

The arguments for the use of heat treatment are fairly compelling. The process is relatively simple and does not require much expenditure of effort. It would seem probable that it would have been discovered accidentally. It is extremely effective.

Furthermore there is, at present, no alternative explanation for the method by which obdurate pebble flint was rendered compliant enough to produce the many fine arrowheads that have been recovered from the sands. If heat treatment was being used, the flint users of Luce Sands would have had access to a source of attractive and highly workable material from which to produce highly accomplished examples of the knapper's craft.

However, even with the use of heat treatment, the production of small, finely worked arrowheads was not easy. Each would have required a considerable expenditure of time and effort. As a result of undertaking this series of experiments, John Lord concluded that an essential part of the operation would have been the production of the original blank. The flake struck would need to be fairly thin to avoid undertaking a long and laborious thinning process. The use of a vice to hold the arrowhead securely seems almost certain. Lastly, it seems very likely that a large number of the arrowhead fragments from Luce Sands may derive from breakages during manufacture, particularly snapped tips, which are present in some quantity. On several occasions during the experimentation process a tip was broken off. Usually the arrowhead could be recovered and a new tip worked. At other times the arrowhead split in half. It must be emphasised that this only happened when heat-treated flint, which had been rendered more brittle by the treatment, was being worked.


© Internet Archaeology/Author(s) URL:
Last updated: Wed May 27 2009