Editor’s Note: This article first appeared in the February 2006 issue of the Norwegian Textile Letter (Vol. XII, No. 2). Katherine Dickerson took a deep dive into a technical analysis of the Flesberg three-shaft bound rosepath technique, comparing it to the more commonly-known krokbragd technique. 

By Katherine Dickerson

Bench cushion woven by Katharine Dickerson in a variety of Flesberg patterns. Photo: Katharine Dickerson

During the summer of 2003, I traveled to Norway with a textile study group sponsored by the Vesterheim Norwegian-American Museum.  During this tour we were taken into a school gym and shown a group of beautiful, visually sophisticated coverlets.  These coverlets came from the southwest Buskerud area.

The Flesberg Farm Women’s Organization has facilitated the creation of a booklet of patterns documenting these coverlets.  Anna Bakken, from Kongsberg, borrowed the coverlets, “plucked” from their patterns, attempted to find the original colors, and wove them anew.  The patterns were then drawn up and written out on computer by her grandson, Håvard Strand.  

The Flesberg coverlet project was begun in 1991 and concluded in 1994.  The coverlet patterns were taken from blankets woven in the period from 1850 to 1920.  Each farm had its own combinations and variations of patterns.  Some farms produced more than one pattern.  There was a total of 30 patterns in the resulting documentation.

Anna Bakken used a black, 12/6 cotton warp on all the samples.  The warp was set at about 6 epi.  The original pieces used a variety of warp materials and home spun and dyed 2-ply wool weft.  Their epi varied depending on the warp and weft used.  

Anna Bakken stated: 

“In our day the coverlets were used for the most part as well hangings.  They are woven in one width, often with fringes.  Earlier they were always the length of a bed, woven in two pieces then sewn up the middle.  This required the weaver to beat the weft evenly so that the patterns matched when the two pieces put together. The two lengths were sewn together by hand with tight stitching. At each end was a narrow, double fold of approximately 2 – 3 cm, also sewn by hand.

These textiles were used as coverlets on the bed.  They were also used as wall decorations for large family affairs like weddings, christenings, and funerals.  As coverlets became worn, they were used for less decorative purposes, for example as an under layer to sit upon in a horse drawn sleigh.  Finally, they became horse blankets, laid over the horses’ backs on cold winter days when they had earned a well-deserved rest, warm and sweaty, after drawing the farm folks to church, town, or on Christmas visits.  Some of the coverlets we have woven were in fact found in the stable.

Upon returning from Norway some of the tour participants formed what we call the Flesberg Study Group.  Through the efforts of Katherine Larson and Laurann Gilbertson, a copy of the Flesberg booklet was made available to this group.  Threading, tie-up and treadling drafts accompany photographic reproductions of the samples.

Flesberg pattern sampler woven by Katharine Dickerson. Photo: Katharine Dickerson

Re-constructing the historical coverlets is one approach to gaining an understanding and appreciation of the coverlets.  However, I have a different approach to perpetuating historical woven structures.  This approach comes out of my more than 40 years of doing research regarding indigenous weave structures and patterns.  It must be stressed that with any historical creative property there is the question of cultural ownership.  This really comes into question when re-constructing any piece thread by thread.  These Flesberg patterns and their documentation are owned by the Flesberg Farm Women’s Organization.  I feel strongly that the study of their woven heritage has to be done with sensitivity and respect for these generous women and their cultural heritage.  Therefore, the information I will be giving out will not be specific to the Flesberg Farm Women’s Organization coverlets and their resulting designs. Rather, the information will be of a general and conceptual nature that will hopefully allow a deeper understanding and appreciation of their weaving heritage. 

It is often easier to understand conceptual information related to a previously unknown woven structure when it is compared and contrasted to a more known woven structure or system.  I have thus chosen to compare and contrast krokbragd, or bound rosepath weave, with the Flesberg system as they have many similarities.  Krokbragd was known in the Nordic countries at least as early as the middle Ages.  It is covered in some detail in pages 86 through 93 in Katherine Larson’s book,  The Woven Coverlets of Norway.  

Sketch by Katherine Larson

Krokbragd is a very simple three harness weave structure.  There is only one threading possibility for single Krokbragd.  The even numbered threads are in one group and the odd numbered threads are alternately divided into two groups.  This grouping of the warp lends itself to weaving on a warp-weighted loom.  The even numbered threads are placed so that the natural open shed of the warp-weighted loom has them in the down position so that they can be covered directly.  The odd numbered threads would be hand picked after raising all the even numbered threads carried on the heddle bar. The division of the threads into 3 groupings instead of 2, expanded the weavers design possibilities significantly from basic 2 block patterns to 3 block patterns.  The desired width of 4 to 5 feet, could be achieved in one piece.

Krokbragd coverlet from Norway (1830-1870), owned by Vesterheim Norwegian American Museum. (Details)

The Krokbragd weave structure could be woven much faster using three harnesses on a 4- harness horizontal loom.  The weaver did not have to hand pick when covering the odd numbered warp threads.  Also, the weft could be shot across and beaten into place more quickly.  The treading repeat was made up of threading harnesses 1,2,3,2 consistently.  The coverlets woven on the horizontal loom generally were woven in two pieces with a seam joining the two pieces at the center, as the horizontal looms do not have the weaving width of the upright warp weighted looms.  

Krokbragd draft

There is a 4-thread repeat to this type of Krokbragd threading.  The largest under shot was under three warp threads.  This would occur on the backside when covering each of the odd numbered warp groupings threaded on harnesses 1 or 3.  The resulting backside of the cloth has a 2-block visual pattern since the undershot rows formed by the two groupings of odd warp threads cover the “tabby” row of the even warp threads.  This undershot structure on the backside allowed the blankets to be pliable and in addition, the undershot structure trapped air, making them much warmer than the 2-harness weft faced weave.

Krokbragd basic draft

DOUBLE KROKBRAGD

There is a second type of threading which is called double Krokbragd.  As with the single Krokbragd, every second thread is threaded on harness 2.  The odd numbered warp threads are distributed in a slightly different manner.  Instead of alternating consistently between harnesses 1 and 3, harness one is used twice and harness three is used twice.  Thus, the threading repeat is made up of threading harnesses, 1,2,1,2, 3,2,3,2 consistently.  With double Krokbragd there is an 8-thread repeat.  The under shots alternate between 1 and 5 threads instead of under just 3 warp threads consistently as with single Krokbragd.

Double krokbragd, also showing floats on the reverse side. (Weaving and photo: Robbie LaFleur)

The dent setting, or ends per inch, varied depending on the use of the single or double threading.  Single Krokbragd is usually set at between 5 to 8 epi and woven with thicker 2-ply weft wool.  Because of the increase to 5 warp threads included with the under shots on the backside, the Double Krokbragd was usually set with a closer epi of 8 to 10.  This kept the resulting undershot to a functional length.  The 2-ply wool weft had to be finer to cover the dense warp easily.  The resulting coverlet was much finer in weave structure, lighter in weight, and more visually complex.

Double krokbragd draft

The structure of Krokbragd gives real limitations in terms of block pattern designing.  Three blocks are used, but since the threading repeat in both instances uses harness two at every other thread, the resulting structure is not visually flexible.  It is human nature that we see color before we see structure. Thus, weavers of Krokbragd have to rely on very sophisticated color usage to make the resulting pieces visually fluid and enlivened.  This is why most Krokbragd woven pieces use so many colors.  I personally always felt a bit limited or cramped when weaving using the Krokbragd structure.  

So now let’s begin to compare and contrast the Krokbragd and Flesberg weaving structure and their conceptual and technical underpinnings.  I will begin with the similarities between the two.  It is hoped that this comparison will serve as a bridge to facilitate a conceptual understanding of the Flesberg weave structure. 

• Both weave structures use a strong warp, made out linen, hemp, cotton or tightly spun wool.
• Both have a weft made out of softer 2-ply wool.
• Both are a type of boundweave, for the most part covering the warp.
• Both divide the warp into three groups and thus when woven on the horizontal loom require three harnesses.
• Both require that a weft shot travel the full width of the piece, from selvage to selvage.
• Both have a tie-up that results in 1 harness at a time being in the down position.
• Both require 3 shots to cover all three groupings, or the entire warp.
• Both have a smooth topside that has a three-block pattern.
• Both have a backside that has undershots of the weft to increase warmth and pliability.
• For both, the back undershot side of the weave is visually less intricate than the smooth topside because it is visually a 2-block pattern. 
• Both depend on the sequencing of color changes to develop their visual imagery.
• And finally, both have an even number of threads in their total threading repeats.

And now for the differences between Krokbragd and Flesberg:

Flesberg basic draft

The Flesberg visual patterning is far more complex and varied than Krokbragd, even though both are based on a 3-block pattern.  As a result, the Flesberg does not have to rely on color and weave to create visual interest.  This is why the Flesberg blankets do not have the same amount of color variation.  As well, they often have large areas of a solid background color.

Flesberg threading is a type of 6-thread point twill adapted to threading on a 3-harness loom.  This type of threading works quite evenly on a four-harness loom with the threading repeat using harnesses 1,2,3,4,3,2.  We are all familiar with this type of basic point twill treading. However, what is interesting to me is that instead of using all 4 harnesses, which would seem to be the most logical approach since they were using a 4-harness counter balance loom, the Flesberg coverlets are woven on 3 harnesses.  The threading instead of the 1,2,3,4,3,2 place the warp thread that would have been threaded on harness 4 on harness 1.  The resulting Flesberg threading is a sequence of 1,2,3,1,3,2.  This peculiarity really interested and excited me.

Why would the weavers choose this 6-thread point twill threading repeat and use only 3 harnesses on a 4-harness counter balance loom?  The answer lies in the fact that this use of a 6-thread point twill threading repeat, using an odd number of harnesses, results in far more design possibilities than the Krokbragd threading.  But at the same time, it maintains the 3-block, 2-block structure with the under shots on the backside.

4 harness point twill

This would not be the case if 4 harnesses were used.  A 6-thread point twill threading with tie-up using pairs of harnesses would result in a 3-block pattern on both top and bottom.  But since these blocks overlap, the resulting imagery would be blended or blurred.  If the tie up were done so that only warp threads placed on 1 harness were covered with each shot, as in Krokbragd, the result would be a 4-block pattern on the top, but the underside would have floats that are functionally too long and again the visual imagery would be blurred.

There are six possible threadings that fulfil the requirements of the 6 treads distributed over the 3 harnesses in a point twill-treading pattern.  These 6 threading possibilities can be broken down into 3 possibilities paired with their opposites.

6 threading possibilities

The Flesberg samples were all woven with the threading 1,3,2,1,2,3.  However, this does not mean that all the blankets were woven with this threading.  The following are the six possible threadings with the even numbered possibilities being the opposite of the odd numbered possibilities:

Generally, because weaving has a binary basis, even numbered repeats remain the same or visually consistent.  Odd numbered repeats, however, flip or change visually.  This is caused by the numerical fact that an even + an even = an even = an even = an even.  The sum of the repeats always remians a consistent even.  Whereas, with odd numbers, and odd + an odd = an even + an odd = an odd + an odd = an even.  An easy way of understanding this basic nature of repeats and resulting visual imagery is through simple addition.  With even numbers, 2 + 2 = 4 + 2 = 6 + 2 = 8 + 2 = 10, the resulting sum remains even.  However, when we add odd numbers, 1 + 1 = 2 + 1 = 3 + 1 = 4 + 1 = 5, the resulting sum flips between even and odd numbers.  It is this flip between odd and even threading sums that result in a real expansion of the visual design possibilities of the 3-block pattern structure of Flesberg as compared to Krokbragd.

I will use the Flesberg Farm Women’s Organization samples threading of 1,3,2,1,2,3 as the basis for my following discussion:     

• Harness 1 carries every 3^rd warp thread. (An odd numbered threading repeat of 3 within the 6 thread even repeat).     
• Harness 2 is used with either 1 warp thread or 3 warp threads between (for an offset even-numbered repeat of 2 and 4).
• Harness 3 is used with either 3 warp threads or 1 warp thread between (for the opposite offset even-numbered repeat of 4 and 2).

This internal combination of odd and even threading repeats results in more visually diverse pattern or imagery possibilities.  This is only one part of what makes Flesberg fantastic.

Flesberg

Since we naturally see color and pattern before structure, I would now like to investigate the structural differences between Krokbragd, Double Krokgragd, and Flesberg woven pieces as a result of their different threading.

As stated previously, Krokbragd has a basic 4-thread repeat of 1,2,3,2, which looks like a twill threading.  But when we look at the resulting woven structure, without the distraction of color, we can see that the woven result is a point twill structure with only 3 threads in either direction.  To look at it another way, each repeat is only 1 warp thread away from basic tabby weave.  Because of the small number of threads in each direction, the slanting nature or the diagonal grains of the weave structure are not very pronounced.  It also means that the pliability associated with more expanded twill weave is not evident in the hand of the finished piece.  The structural woven result of this threading really is quite closely akin to a tabby weave since every other warp thread is always threaded on harness 2.

Krokbragd draft

Double Krokbragd has a somewhat more interesting resulting twill woven structure.  It alternates between a tabby structure and the short twill structure of Krokbragd.  When the treading is repeated on harnesses 1 – 1 or harnesses 3 – 3, with a warp thread on harness 2 between them, the structure is tabby.  When warp thread on harness 2 is between warp thread on harness 1 and harness 3, then a 3-thread twill results.  But just like Krokbragd, the resulting woven structure is very close to tabby or plain weave.  

Double krokbragd draft

Flesberg, on the other hand, has a woven structure of a 6-thread point twill with each direction of the twill being made up of 4 warp threads.  The increased length of the twill diagonal results from the warp that would have been threaded on harness 4 being placed on harness one instead.  This, in turn, results in two warp threads being between each warp thread placed on harness 1.  This increase in the number of warp threads placed between those that are threaded on harness 1 results in the twill having 4 threads in each direction.

Comparison: krokbragd, double krokbragd, flesberg

This increase in the number of threads in one direction or the other has significant ramifications on both the visual image and on the hand or drape of the finished product.  The more developed twill structure allows the Flesberg pieces to have more drape.  It also allows for a much denser weave with a finer warp setting of between 8 and 10 epi.  The undulating point twill structure is visually evident when looking at the solid areas of the blankets.  The more pronounced twill structure also affects the look of the visual pattern and has to be accounted for when developing treadling color sequencing.

With Krokbragd, because of the even threading repeats, the picks covering the threads placed on each harness remain visually in the same place.  This is not the case with Flesberg.  The warp placed on harness 1 alternate consistently on every third thread.  Because of the odd threading repeat, the first pick covering the warp on harness 1 looks visually attached to the bottom of the twill structure, while the next pick covering the warp on harness 1 looks visually attached to the top of the twill structure.  This visual flipping consistently repeats across the full weft shot.

The visual flipping of the picks covering the warp threads on harness 1 is heightened by the directional slant of the point twill structure.  The individual picks of the weft thread covering the harnesses adjacent to harness 1 will physically slant down, then up, then down consistently.  The increased twill length gives a physical grain to the piece.  This grain, or slant is especially important to consider when weaving any images that contain diagonals.  I think that this slanting was one of the reasons I was so taken with the coverlets when I first saw them in Norway.  Most of my previous research has been studying the twining of the indigenous peoples of the Pacific Rim.  Weft twining is all about understanding directional slants and resulting visual imagery.  I think that is why I was intuitively drawn to these coverlets.  

Flesberg

Another key to understanding the increased visual possibilities with Flesberg is that when a shape changes from an increasing to a deceasing shape as in a diamond, or from a decreasing shape to an increasing shape as with X’s, the treadling turns or has a directional change.  There are 6 possibilities for facilitating treadling directional change.

This diagram illustrates the treadling selection for directional change on the right and its pair, or opposite, on the left. 

Using treadle 2, then 3, then treadle 2 again; and its opposite using treadle 3, then 2 then treadle 3 again.

Using treadle 1, then 3, then treadle 1 again; and its opposite using treadle 3, then 1, then treadle 3 again.

Using treadle 1, then 2, then treadle 1 again; and its opposite using treadle 2, then 1, then treadle 2 again.

Directional changes

Since the treadles are tied up so that only the warp threads attached to one harness at a time are covered, it takes 3 weft shots to cover all the warp threads.  Straight draw treadling results in the weft covering a group of warp threads on specific harnesses being separated consistently by 2 weft shots.  This 3-shot treadling sequence is interrupted by the visual necessity of directional changes.  Each of these directional changes results in the warp grouping, that is not included in the turn, being exposed for 3 weft shots.  As well, the warp being used at the beginning and end of the turn results in its being separated by only 1 weft shot and visually looks like a double row.  These inconsistencies should be accounted for when considering a directional turn.  

I would now like to share with you my own research regarding Flesberg system of weaving.  I wove a futon cover for the couch in the middle of my yurt.  You see the coverlet at the back of the couch when you enter the yurt. 

Photo: Katharine Dickerson

The rug was my starting point for dyeing the colors for the coverlet.  For the pattern areas I used some very refined tapestry wool left over from a 1970’s commission.  I over-dyed the yarn with natural dyes to give subtle color variations and more color depth as well as to harmonize with the colors of my rug.  For the solid areas I used a mat natural white yarn.  The contrast, in terms of color absorption and reflection, the difference between the natural white and the colored yarn heightened the visual power of the pattern areas.  

I wove the coverlet on my counter march loom in a double weave “C” curve.  I threaded 6 harnesses, three for each layer.  I used threading possibility 4 and thus my threading sequence was 3,6,2,5,1,4,3,6,1,4,2,5.  The top layer was threaded on harnesses 4 through 6 so that the bottom layer had the most open shed.  In this way I was able to achieve the double width without the visual interruption of the sewn center seam.

The coverlet is made up of 6 patterns that I classify as compound patterns as they are made up by combining two or three basic patterns.

Border 1

The first compound pattern band (Band 1) has borders of basic diamonds.  This is tricky because the top of the color for the top of the point carries over to the next pick that covers harness 1. The even picks are thus not visually connected to the diamond pattern and so careful color selection had to be used so that the even picks do not stand out.  I was also interested in developing a central figure that would echo the diamond and develop into a more rounded shape.  I wanted the border to ground the piece and so I made the overall visual to be dark in tone.

Border 2

The second compound pattern band (Band 2) was an extension of the curved nature of the center of the first pattern area.  I wanted to create contrast and so wove it in a lighter tone.  By carrying the natural white into the pattern, I increased the sense of light and movement of the visual image.

The third compound pattern band (Band 3) went back to dark tones.  It is made up of 3 basic patterns.   Here I was working on merging patterns within the internal structure of the pattern instead of having one follow the next.  Previously, I wove one simple pattern consecutively after another.  In this central area, I merged them by overlapping.

Border 4

The fourth compound pattern band (Band 4) is made up of 3 basic patterns.  The figure 8’s look simple, but required 3 directional changes to achieve the slant on the ends and in the middle.  With the central figure, I was interested in the alternating slant of the horizontals.  The draft of the central image I have previously shared with you.

Border 5

The fifth compound pattern band (Band 5) is also made up of 3 basic patterns but this time I used the natural white background color as the figure color in the center.  I wanted to highlight the central pattern and so the border patterns visually point towards the center of this band.

Border 6

With the final compound band (Band 6), I carried the background natural white into the border.  I also hit the center of the middle pattern with an intense orange to make it seem like it was internally lit.

In conclusion I would like to state that the internal combination of odd and even repeats of the Flesberg threading, as well as the increased twill structure length and the resulting physical slant of the weft threads, allow for far more variety and possibilities when designing the visual image.  These differences make me believe that this 3-harness weaving system is very sophisticated and technically developed.  It is the reason why I was so excited when we originally viewed all the coverlets.  It is also why I have been intellectually engaged in trying to conceptually understand this structure instead of just going through the motions of recreating the weaving samples step by step.  I hope that this talk has piqued your interest in Flesberg.  

I also hope that the 3-day designing and weaving workshop will give those participants a more complete understanding of the technical make-up and visual possibilities of the Flesberg weaving system.  I have learned through my work with indigenous people that revivals do not happen through reproduction of the past.  It is only when the techniques and processes are conceptually understood and then used in a contemporary context that they again become alive.  That is why I passionately believe that it is only when the Flesberg structure is used in a contemporary and original context that it will continue to develop and stay a part of our future Norwegian heritage.

Paper presented by Katherine Dickerson at the 3rd conference of the Norwegian Textile Guild, October 2005 in Decorah, Iowa.

Katharine Dickerson was born in Duluth Minnesota.  Her Norwegian heritage has become increasingly important in forming her aesthetic values.

She retired from the Alberta College of Art + Design after 30 years of teaching in the Textiles Department.  In 2009 she was awarded the designation “Lecturer Emeritus” by the ACA+D Board of Governors.

Katharine’s many years of research in Salish Twining gave her a unique understanding of the slanted pick orientation found in the Flesberg weaving structure.  She continues her Flesberg weaving and teaching at her home and studio on the Slocan Lake looking at the Valhalla Mountains in British Columbia.