Journal of Archaeology in the Low Countries 2-1 (May 2010)Chrystel R. Brandenburgh: Early medieval textile remains from settlements in the Netherlands. An evaluation of textile production

3 Textile production

3.1 Fleece processing

next section

The end product of a woven fabric is determined from the start of the production process by selecting the proper fibres from a fleece. Analysis of the fibres in archaeological textiles provides information about the natural colour, type and fineness of the wool selected for specific types of textiles. The earlier fleece type analyses were conducted using a chronological model of fleece evolution from hairy to fine and evenly distributed fleeces (Ryder 1964). Modern critics have argued that this approach is not useful for archaeological textiles. Ryder’s model does not take into account that the fleece from one single sheep varies greatly depending on which part of the body it came from. It also assumes that wool was used straight from a sheep when it is more likely that it was prepared, sorted and selected to create a better yarn (Christiansen 2004, Rast-Eicher 2008). Fleece analysis will therefore not provide chronological information but may provide evidence for the preparation of the fibre before it was spun into threads.

Recently the fleece type of several Dutch samples was analysed in order to compare these textiles to those from northern Germany (Walton Rogers 1995), Norway and Denmark (Bender Jørgensen & Walton 1986; Walton 1988) and Anglo-Saxon England (Walton Rogers 2007, 10-14, 62-64). Twenty-eight samples of Early Medieval textiles were selected for a quick scan of animal coat and natural pigmentation (table 3; analysis by P. Walton Rogers, the Anglo-Saxon Laboratory). All except two textiles proved to be made from sheep’s wool. A textile from Beetgum was almost certainly made from the undercoat of goat. Only four textiles were made from white fleeces. Among these were unusual fabric types, not necessarily locally produced (a felt from Ferwerd and a gauze-like textile, or Schleiergewebe, from Leens).^[4] The hat found at Oostrum (fig. 3) was also made from naturally white wool. Dark brown or black was the most common colour for twills and diamond twills.

Several fabrics were woven in a colour pattern using threads of different colours and thus creating stripes or blocks within the fabric. Macroscopically, this technique has only been observed in five textiles, but based on results of microscopic analysis, we may assume that it was applied more frequently. Fibre analyses show that six out of 28 textiles (21%) were woven with naturally dark wool in one thread system and originally light wool in the other thread-system.^[5] Among these were four textiles that had not been macroscopically recognised as such.

Seventeen samples from seven different textiles were selected for further analysis of fleece type (table 4). Warp, weft and any sewing thread or pile yarns were analysed separately. In order to identify the fleece type, 100 fibres were measured and the results plotted as a histogram.

According to the range, mode, mean and degree of skew of the measurements, the samples were allocated to one of seven fleece-type categories: Hairy (H), Hairy Medium (HM), Medium (M), Generalised Medium (GM), Fine (F), Semi-Fine (SF; previously called Shortwool) and Fine/Generalised Medium (F/GM). Eleven of the samples were HM, one was GM, two M and three samples (all from the same textile, Beetgum lab.nos. 46-95) were goat fibre.

Only two textiles show evidence for the special preparation of wool. The white wools in warp and weft of the gauze-like tabby, or Schleiergewebe ^[6] have been allocated to the M category because of their symmetrical spread of the fibre diameters and their means between 30 and 40 microns, although the maximum diameter for the M type should be 60 microns and both yarns include a single fibre thicker than 60 microns in diameter (table 3). The wools in a similar textile from Hessens, Stadt Wilhelmshaven, in north Germany (He33a, and possibly also in He31c), were similarly difficult to categorise and it was suggested that they may represent an HM fleece from which the hairs had been stripped out. The same may be true of the Leens example (Walton Rogers, unpublished). These textiles are also found in Anglo-Saxon England and in Viking Age Denmark, Britain and Ireland, and it is possible that they represent trade goods produced in a specialist workshop (Walton Rogers 2007, 68-69).

The textile from Beetgum is made from goat fibres. The outer coat hairs are absent in this case, which may indicate that the underwool was combed directly from the animal during its spring moult.

The Dutch textiles resemble those from northern Germany. Previous research on 27 samples from mainly seventh to ninth century sites in northern Germany shows that the textiles were made from fleeces that are categorized as Hairy or Hairy Medium. Similar results were obtained from samples of raw wool found at the same settlements. This may indicate that the wool was processed in the settlement, making the woven textiles a local product. Many fibres were originally of a brown or mottled brown colour. White fleeces were only observed in 22% of the threads. In contrast to the Dutch textiles, all textiles were woven with wool which was originally the same colour in both warp and weft (Walton Rogers 1995, table 3). Some of the same fabrics and pigmented fleece types are also found in Anglo-Saxon England, although the English material has a larger share of white wool and a wider range of fleeces. The terpen evidence contrasts with the material from Norway, which shows a much more precise method of selecting and processing wool.

The raw materials from the textiles from the terpen, found in the north German and Dutch settlements show a wider range of fleece types and a lack of carefully sorting which makes them closer in type to the Hessens-Elisenhof type textiles excavated in southern Scandinavia (Walton 1988, 153; Bender Jørgensen 1984, 130-1, Walton Rogers, unpublished).

3.2 Spinning

To spin yarns from fibres one needs a spindle whorl and a distaff. Depending on the direction the spindle whorl rotates, the threads are twisted either clockwise or anticlockwise resulting in z- or s-spun thread (fig. 4). Right-handed spinners generally spin clockwise (z-spun treads), but accomplished spinners can change the direction of spin when needed. In order to make an even stronger yarn several threads may be twisted together, resulting in plied yarn (for example a 2zS-yarn is made out of 2 z-spun threads, plied S-wise). Among the woven textiles single twisted threads are most common. Plied yarn has been used only in a few cases (table 5). This small share of plied yarns is a great contrast to the (Roman) Iron Age when plied threads were used in the majority of textiles (Bender Jørgensen 1992, 49).

3.2.1 Quality of spinning

The quality of spinning can be ascertained by studying the thickness of the threads and the degree of regularity of the spinning. Presumably every woman in the Early Middle Ages could spin with a considerable degree of skill. Recent spinning experiments have shown that even nowadays one can, with a little practice, easily spin quite regular thin threads of about 0.5 mm with practically any type or size of spindle whorl.^[7] We should however take into account that the breeds of sheep kept in the Early Medieval period had fleeces that were not as ideally suitable for spinning thin yarns as those bred nowadays. Thickness of threads has therefore been divided into 4 classes: 1.5 mm. The first category consists of very fine threads that needed careful spinning and more time to be woven into fabrics. The second category 0.5-0.75 mm may be regarded as the thickness that could easily be spun by any experienced spinner. The third and fourth categories are coarser or thicker yarns.

Wild (et al. 1998) and others have stressed the great advantages of digitally analysing the degree of twist in yarns because it enables researchers to distinguish the hand of individual spinners in a dataset and also the degree of experience of the spinner (Cork et al. 1996, Wild et al. 1998). Unfortunately this method was not available for this dataset, therefore the degree of twist has been classified as low, medium or high or, in the case of irregularly spun threads, as a combination of these. The finer threads (up to 0.5 mm) are generally spun medium to high regularity, making a very strong yarn. The 0.5-1.5 mm thick threads show a wide range of twist, from low to very high. The very thick yarns are often barely spun, making a very soft thread.

Table 6 shows that most fabrics from well-dated sites were woven with threads of 0.75-1.5 mm thickness. The finest, 1.5 mm is represented only in small numbers. There are some differences between the three major textile sites, Dokkum, Westeremden and Leens. In Dokkum there is slightly more fine spinning, whereas Westeremden has a larger share of the 0.75-1.5 class. Leens shows significantly less fine threads and more very coarse yarns of the >1.5 mm class. Middelburg shows a completely different pattern with a dominance of fine spinning, 37% in the <0.5 mm and 29% in the 0.5-0.75 mm class.

As stated before we may assume that it was possible to produce threads as thin as 0.5 mm despite the lower quality of the wool. Therefore there is no doubt that it would technically be possible to spin even finer threads. However, spinning and weaving these fine threads into equally fine fabrics would be more time consuming than spinning coarser threads and fabrics. These textiles consequently must surely have been more valuable than the coarser fabrics. The fact that more than 20% of the yarns are of this fine and time-consuming quality indicates that the people in these settlements either had time to produce this quality textile or sufficient wealth to purchase it, which may imply a certain level of craft specialization.

The majority of the spinning is of a much coarser quality. Instead of considering the technical limitations of the craftspeople, there may be other ways to explain this distribution. One very practical assumption may be that the settlements have yielded not only remains of clothes but also a large proportion of furniture or household textiles, which were not required to be of a very fine quality. Many pieces from the coarse section of the dataset however show traces of sewing and may have been primarily used as clothing. It may therefore be assumed that time was also an important factor in the choice of a certain quality of thread. As stated earlier, spinning and weaving fine threads is time consuming and had to be combined with numerous other tasks. Therefore an overrepresentation of rather coarse fabrics may indicate that the people involved generally did not have the time to put more effort into textile production. Neither would it suggest they had the means to let others do this work for them. Comparing the sites, Dokkum may perhaps be seen as an exception where there is an even distribution of threads up to 0.75 mm and thicker ones. Middelburg seems to be an important exception, but in reality it is not. The examples of fine spinning in the textiles from this settlement are in most cases only present in the warp. These fabrics have been made with a very thin and strong warp but very often with a much thicker weft. The production of these textiles is thus no more time consuming than for other rather coarse fabrics.

3.3 Dyeing

Many dyes were used in early historic times. A red colour was obtained by dyeing wool with dyestuff extracted from different species of dyer’s madder. Archaeological evidence for the cultivation of Rubia tinctorum L. in the Netherlands is not available until the seventeenth century. There is historical evidence that dyer’s madder did occur in the terpen area in the northern part of the Netherlands in the Early Middle Ages (Van Haaster 2001). Wild madder (Rubia peregrina L.) is known in the southern part of Great Britain and in the Mediterranean. Dyer’s woodruff and bedstraw were also used to produce red colours as far back as the fifth century AD (Cardon 2003, 120-128). Red could also be obtained from the insect kermes creating a very strong and colourfast dye. This precious dye was produced in the Mediterranean and valued greatly in north-western Europe as a symbol for kings (Cardon 2003, 618). Lastly shades of crimson could be obtained using a dye extracted from the insect of the Porphyrophora species. Evidence of this dye has been found in a sixth century context in Germany.

Blue was obtained from woad (Isatis tinctoria L.). Woad has been known in the Netherlands since the Iron Age (Cappers 1994). Yellow dyes were extracted from weld (Reseda luteola L.), which was widespread in western Europe (Cardon 2003, 170). Remains of this plant have been found in Roman forts in the Netherlands (Pals 1997, 35). Another source for yellow dye is the plant Dyer’s broom which has been identified in ninth century finds form York (Cardon 2003], 177). Purple was obtained from lichens of the genera Ochrolechia and Umbilicaria. This dye has been identified in ninth and tenth century finds form York, Dublin, London and Scandinavia (Cardon 2003, 501). Purple could also be extracted from marine molluscs but this was a very expensive way of dyeing. In the Late Roman period purple was associated with imperial majesty and in later periods it remained the colour for kings and synonymous with wealth. The prestige of the colour purple increased with its scarcity. The dyestuff was not locally available and had to be traded form the Mediterranean or Brittany (Cardon 2003, 574). Different hues of brown would have been obtained using natural dyestuffs from bark and nuts that were readily available in any wooded area.

Only the wealthy could afford to wear certain colours because they were expensive to produce. To wear them was thus a social signal to the wearer’s contemporaries, that they could afford this level of luxury (Hedeager Krag 1993). Analysing textiles for dyestuffs may therefore result in an indication of the wealth of the original wearer. There are, however, a few hurdles to overcome in relation to dye analysis. Natural dyestuffs deteriorate over time and will very often have disappeared entirely during the period the textile was buried. Consequently a negative result in dye analyses does not necessarily mean that a fabric was not originally dyed. Many dyes that were locally available, like the brown colours from nuts and bark, would also be hard to detect. It is generally very hard to discern the chemicals from these dyes from those naturally present in the soil because of their similarity to material found in the natural environment. It is therefore difficult to know exactly how colourful Early Medieval clothes were.

Recently, seven textiles were selected for dye analysis (table 7). Dye could be identified in the hat from Oostrum (fig. 3), the body of which was made from a white fleece and had decorative stitching in fawn wool. The same madder type dye was present in both the textile and the sewing thread but it was much more concentrated in the stitching, making it likely that the ground fabric was light red, salmon or peach and the needlework a deep dull red. Chemically, the dye was dominated by purpurin but there was a trace of alizarin, which suggests that the dye came from the roots of Rubia tinctorum L. (Walton Rogers unpublished).

There appeared to be a tannin-based brown or black colorant in the headdress or hat from the site of Berg Sion (Dokkum)(fig. 5).^[8] This is fairly exceptional since the headdress was made out of naturally brown wool, which in most cases would not have been dyed. Tannins are widely distributed in nature, especially in material from trees, and it is not always possible to recognise tannins deliberately applied as dye. In the case of the Berg Sion headdress, however, the colorant was detected in the main fabric of the hat but not in the needlework, which suggests that the tannins were present in a dye applied to give a solid black to the already naturally dark fleece colour of the headdress. The dye could have come from barks, nuts or oak galls.

No dye was detected in the Leens Schleiergewebe-tabby. This does not mean that the textile was not dyed. Other textiles of this type have proved to be dyed black, blue or purple (Walton Rogers 2007, 69).

Previous research on Anglo-Saxon textiles has pointed out that naturally white wools were often dyed. Analyses of naturally brown or black samples nearly always had a negative result, meaning that in most cases these textiles were not dyed at all.

The textiles from the Netherlands and Germany (Walton Rogers 1995) are very similar in this respect. Dyestuffs have been detected in only a few textiles. Those fabrics that had certainly been dyed come from hats that had been sewn with great care (see 4.2 Needlework) and must have been valued for their appearance. The rest of the textiles were probably either originally (mottled) brown or black.

3.4 Weaving

3.4.1 Looms and their characteristics

The process of weaving large pieces of cloth was generally conducted on a warp-weighted loom (fig. 6, left). This type of loom would have stood slightly at an angle against the wall of a building. The vertical threads of the fabric, the warp, were hung onto the upper crossbeam of the loom and put under tension by attaching loom weights. These loom weights can be found in abundance at Dutch sites.

Another type of loom, known from the countries surrounding the Netherlands, is the two-beam vertical loom (fig. 6, right). This loom type was in use during Roman times and must have remained in use in parts of France during the Merovingian and Carolingian periods, re-emerging in more widespread use around the end of the ninth century (Henry 1998, 2005). The change of loom is associated with a predominance of a weave-type which hitherto had not been very popular, the 2/1 twill (fig. 6b). Moreover the shift from one loom type to the other may be related to a change in the organisation of textile production from a domestic basis to a more organized and centrally controlled production (Henry 2005). If and when this loom type was actually in use in the Netherlands is not certain.

From the tenth century onwards, historical texts mention a third loom type, the horizontal treadle loom (Cardon 1999, 412). The oldest finds in northwest Europe associated with this loom type are dated to the tenth century. In the beginning the width of the cloths produced on this loom was not very large. When weaving cloths of more than 1 metre width, one needed two weavers to operate this loom and it was only later that this became custom.

While the warp-weighted loom was very suitable for weaving broad cloths up to a length of 10 m, the horizontal loom was most effective when weaving narrow fabrics longer than 10 m (Cardon 1999, 415). The warp-weighted loom has no reed or batten, which may have affected the regularity of the thread systems. This irregularity is visible in the woven fabric in variable spacing of the threads and curving lines (Hammarlund et al. 2008).

No research has been conducted so far into the specific weaving tools that are associated with the various loom types, therefore the distribution of weaves presented below cannot yet be related to a type of loom.

3.4.2 The fabrics from Dutch settlements

Before discussing signs of specialization in weaving, a brief overview is required of the characteristics of the textiles from the Dutch settlements. This discussion will focus on the different techniques observed and their distribution across time and space.

Among the well-dated sites, nearly 50% of the textiles were woven in a diamond twill (table 7a & b). 2/2 plain twills are also present as a large group, followed by tabby, 2/1 twill, cross twill, herringbone or chevron twill and repp-effect tabby in small quantities. There are considerable differences between the major textile sites of Dokkum, Leens, Westeremden and Middelburg. Dokkum shows the largest variation of weaves, which is not remarkable since this site has yielded nearly twice as many textiles as Leens and three times as many as Westeremden. Dokkum has an equal number of diamond twills and 2/2 plain twills. Westeremden gives a very different picture with a large majority of diamond twills and very few 2/2 plain twills. In contrast, Leens shows considerably more 2/2 plain twills than diamond twills. Among the textiles from Middelburg (12 in total) we only see diamond twill and cross twill.^[9] These different ratios among the sites may point to preferences for specific fabrics that were not necessary or required to the same extent at every site. There are considerably more 2/2 plain twills in many sites than previously documented in the diagram by Bender Jørgensen (1992 48, fig. 58).

Diamond twills show many patterns (table 8a). Some sites, like Dokkum, show a considerable variation of pattern repeats. In Westeremden on the other hand, a large majority of diamond twills are woven in pattern repeat 20/18, which points to a certain preference for this pattern there. This preference is also present in settlements across the border, such as Elisenhof and Hessens (Stadt Wilhelmshaven) (Tidow 1995, 359).

Several fabrics are woven in a spin-pattern. These patterns are created using both z- and s-twisted threads in warp or weft. The different direction of the twist of the yarns gives a very subtle but clear pattern. This pattern is present in 10 textiles.^[10] All these textiles are rather coarse, the finest being spun in 10 x 8 threads per cm, but most are below 7 threads/cm. The pattern is present in diamond twills, 2/2 and 2/1 twill and tabby.

Borders or selvedges are observed in 25 textiles (table 9). Many of these borders are not reinforced at all, but are created by weaving the weft-thread immediately back into the fabric. This technique is, not surprisingly, mostly observed in rather coarse fabrics, but it is also present in a few of the finer textiles. Reinforced borders are present in 15 cases. These borders are made in tablet weave creating either a tablet woven band of three to six tablets or a tubular border (fig. 8). An example of a starting border in tablet weave was found at Hoogebeintum.^[11]

3.4.3 Signs of specialization in weaving

There are several ways to identify possible specialization in the weaving process. One is the estimation of the time and effort spent. A common way of estimating this is by comparing the thread counts of the weaves. Weaving a fine fabric with a large number of threads per centimetre takes more time than weaving a coarse fabric with only a few threads per centimetre. It is therefore useful to divide the dataset into groups ranging from coarse to fine. However, a focus on thread count alone would not do justice to many of the textiles. A cloth does not necessarily have to be of a high thread count to be valued. A coarse but regularly spun and woven fabric may be very pretty and equally valued for its craftsmanship. So besides this quantitative approach one can consider the regularity of the weaving. Relevant variables might be whether or not faults are visible and whether the appearance of the fabric is regular or not. This is a subjective way of classifying the textiles, but nevertheless gives an impression of the skill of the weaving. Lastly, there are fabrics that needed special skills or specific tools to produce. These most likely are the products of specialized workers and must have been valuable goods.

The fabrics may be divided by thread counts into five groups, ranging from very coarse to fine (fig. 9). The majority of the textiles have thread counts below 12 threads/cm. Only a small group may be considered as fine quality, but there are no fabrics finer than 28 threads/cm. There are slight differences between the sites (fig. 10). Leens has yielded more coarse fabrics, which may point to an overrepresentation of household textiles. In Westeremden and Middelburg this coarse group is missing altogether and both sites yielded considerable quantities of finer fabrics.

There are two examples of very fine spinning and weaving in the dataset. First, the so-called Schleiergewebe or veil weave found in Leens (fig. 11).^[12] This is a very fragile and open tabby, woven with z-spun threads of 0.2 mm and approximately 10 threads/cm. The fabric was woven out of naturally white wool^[13] and was possibly used as headdress. The other, finer textile is another tabby (repp-effect) found in Dokkum. This fabric is a very dense cloth woven with 28 x 15 threads/cm.^[14] Two colours of wool were used, white for the warp and dark brown for the weft. It is not clear whether the fabric was also dyed, since no dyes have been detected on the textile. Both these fabrics must have taken considerable time to produce.

Comparing thread count and the regularity of the weave gives further information about the quality of the fabrics. The finer fabrics are often of a high and regular quality, as may have been expected, but this is also the case for most of the textiles in the middle group. This group, woven with approximately 10 threads per cm, was perhaps not necessarily of high value, but it may reflect the quality of work an accomplished weaver could achieve in normal circumstances. Using Olausson’s model for production it may also be possible to classify these textiles as the products of an independent specialist as they are characterised as efficient and standardised, requiring a minimum of production time and with little evidence of errors.

Another pattern emerges when the different weaves and their thread count are compared (fig. 12). 2/2 twills are generally coarser than their counterparts, 2/2 diamond twills. Technically, 2/2 twills are easier to weave than diamond twills and the fact that this bind is most often produced in low thread counts affirms its function as bulk product, which generally must have been used for general household needs. Diamond twills, on the other hand, were not made in coarse fabrics. The decorative pattern of this twill, combined with the higher thread counts and the technical difficulty, may point to a different value and use of this cloth type.

Finally there is one type of fabric that required extra technical skill to produce, piled weave. Piled fabrics are sparsely represented in the Netherlands. Examples were only found in Leens and Dokkum (figs. 13 & 14).^[15] These weaves are rather coarse, very thick and densely felted z/s 2/2 twills, with long strands of s-spun thread worked into the fabric and hanging from the surface. These threads had the same function as fur, causing water to drip down the threads instead of drenching the woven cloth beneath. This fabric was very suitable for cloaks, which has been confirmed by finds in England. There, this fabric has been mainly found in men’s graves as a cloak or body cover. In some cases piled fabrics had been dyed (Walton 1989, 336; Geijer 1938, 132) and the quality of the wool suggests that they were luxury goods (Walton Rogers 2007, 85). The production site of piled fabrics from the fifth to the seventh century (presumably contemporary with the Dokkum textile) is unknown. From the eighth century onwards (contemporary with Leens) piled fabrics were traded from Ireland and Iceland and the Frisians also seem to have had a share in this trade. Texts mention that they were trading in a cloth called villosa that may have been used for this type of cloak (Gudjonsson 1962, 70, Walton Rogers 2007, 85-86).

In summary, it is possible to conclude that the coarser weaves were generally made in z/s 2/2 plain twills. These twills were quick and easy to make and were used for household needs, bedding, sacks, etc. The largest group of textiles consists of regularly woven 2/2 twills or diamond twills that could have been produced by any able weaver, so presumably production took place on a domestic level. Applying Olausson’s model, the regularity and efficient production of these textiles may however also be interpreted as characteristic for the work of independent specialists. Only a small group of mainly z/s diamond twills are of a finer quality, which required more time to weave. It is not clear whether this production took place at the household level or at a specialist workshop. One can merely conclude that people did occasionally take the time to make these textiles or pay somebody else to spend their time weaving the cloth. The veil weave found at Leens and the two piled weaves from Leens and Dokkum are rare examples of textiles that were almost certainly objects of trade.

3.5 Felting

Felting is a process that takes place after weaving a fabric. It involves soaking the woven fabric in water and a fulling substance like soap or mud and then beating or treading it. The aim of this process is to make the fabric thicker, more dense and therefore warmer and waterproof. It is, however, not so easy to recognise whether or not a fabric was felted deliberately because a garment can get the same matted and felted surface when it is used in normal life through the friction of one piece of cloth onto another. On the other hand, the absence of a felted surface does not necessarily mean that a fabric was not felted. The matted surface can easily break away during excavation and finds processing, leaving a clean and unfelted appearance.

There are only a few textiles that show a felted surface. Several of these seem to have been primarily felted. The piled weaves (see 4.4), which were probably used as cloaks, must have been felted. These thick and dense fabrics were clearly meant to be waterproof and a felted fabric would greatly enhance the function of this garment. The mitten found in Dorestad (fig. 21), a thick mantle-like fabric from Dokkum and two pieces from Middelburg are also likely to have been felted.^[16] In the case of the Middelburg textiles, it has been suggested that they have a raised nap (Leene 1964). The technique of raising a nap involved roughening up the surface of the fabric with teasels and afterwards shearing the surface back with large iron shears. This technique had been in use since the Roman period (Wild 1970) and is considered a specialist activity in the Early Middle Ages.

• Created by the Amsterdam University Press i.c.w. Digitaal Productiecentrum