31 July 2015

An Old Band-Woven Bookmark

A while back, a private book collector in the Midwest sent me a textile to examine.  It had come to him as a bookmark in a sixteenth-century bound manuscript.  With his permission, I am sharing my analysis here.

The bookmark is what's known as a "portable register" type.  Rather than being attached to the binding or a page of the book, it is removable.  Specifically, it's a multiple-strand bar anchor bookmark.  Several textile strips are secured to a bar (the "anchor") that is designed to rest on or at the top of a book.  See this article for an introduction to medieval bookmarks, and see the sources at the bottom of this post for more information.


This bookmark is composed of what appears to be a hand-whittled wooden bar anchor around which are sewn five lengths of band-woven textile.  The strands vary in condition from bright, flat, and flexible to crusty, dirty, curled, and stiff.  Only one of the ends has any finishing treatment; most are quite ragged and torn. 

The anchor is is about 2-1/4" (56mm) long by 1/4" (6mm) at its widest diameter.  The carved away portion of the bar is about 1-1/4" (32mm) long.  The wood is a light-colored, fine-grained hardwood, slightly bowed along its length.  Wear has chipped and pitted the surface at the two ends, which were originally smooth.  The collars at either side of the carved away area are still sharply cut into the wood, as if by a small but very sharp knife.

The five individual lengths of band-woven textile have been looped around the anchor for an original count of ten hanging tails, the strands.  A line of coarse stitching worked with a naturally colored linen thread holds together the entire set of looped bands.  The sewing thread is wrapped several times around each of the two outside bands. 

Six of the original ten strands survive in a usable length.  Both sides of the middle strand have been torn away leaving nothing but frayed ends next to the anchor.  One entire side of each of two strands has also been torn away, again leaving frayed ends next to the peg.  Two strands are substantially complete although one end of each is shorter than the other end; both the shorter ends are also frayed. 

The two longest remaining ends are approximately 22" long.  One terminates in an overhand knot.  The other has some broken warps and comes to what looks like a blind end.  It may be that these 22" lengths reflect the original lengths of the strips.  The book in which the bookmark was found is about 12.5" tall, which corresponds to the area of each long length that is comparatively clean and bright in appearance.

The textiles appear to be five strips cut from a single long narrow band.  The band is of warp-faced tabby, i.e., either rigid heddle or inkle woven.  The warp is two-ply linen of approximately 0.4-0.5mm diameter.  The band varies between 4mm and 5mm in width, and the beat is very regular.

one of the best preserved pieces of the band
The eleven warp ends are in four colors—bleached, golden yellow, light greenish blue, and dark blue—and the weft is more of the light blue linen.   The blue threads are finer than the gold and white ones.

weaving draft for band


There doesn't seem to be a good solid hook from which to hang a specific date on this bookmark.  There is no reason to assume it was necessarily as old as the manuscript.  While some portable register bookmarks found in medieval books are likely to have been original (Swales & Blatt 2007, 167), others are not.  This bookmark could easily have been an afterthought, added to the book centuries after it was bound.  Perhaps someone retrofitted an existing anchor by giving it new textile strands in a later century.  Or perhaps the present bookmark might have completely replaced an earlier marker whose anchor was in poor repair but whose existence was helpful.

Although some parts of it are worn and dirty, proving that it's been around for a while, the linen for the most part is in extremely good shape, supple and not deteriorating.  It is hard to imagine linen strands would be in that fine a shape after hundreds of years closed between two layers of vellum; over so long a period of time surely the pH of the vellum would prove damaging to the linen.  Also, the use of several colors in the linen tape is, as nearly as I am aware, more likely to be a product of the eighteenth or nineteenth century than of the sixteenth or seventeenth.  I could not find any examples in Swales & Blatt's catalogue that were even remotely similar to this one save #2, "plain-woven red/blue/white wool and linen tape," which the authors could not examine (Swales & Blatt 2007, 148).  A mixed wool and linen tape seems even less likely to date back that far than the one at hand.

Accordingly, I'm going to hazard a guess that the textile portions of this bookmark are nineteenth century, at the earliest.  About the anchor, I've no idea, but it seems to have been fairly amateurishly produced, perhaps by someone bookish who needed a replacement anchor for a special book.

I appreciate the opportunity to have examined this piece as well as the challenge it posed to my knowledge of historic bookmarks.  I learned a lot!


I created the weaving draft to match the original colors using KXStitch* and the Anchor color palette.  The exact Anchor colors I used for this draft are as follows.
  • gold:  295 Jonquil
  • white:  2 White
  • light blue:  1062 Peacock Blue
  • dark blue:  131 Blue


Janzen, Jenneka.  "Mark Their Words:  Medieval Bookmarks."  https://medievalfragments.wordpress.com/2014/07/11/mark-their-words-medieval-bookmarks/.  Accessed 31 July 2015.

Kwakkel, Erik.  "Smart Medieval Bookmarks."  http://medievalbooks.nl/2014/09/22/smart-medieval-bookmarks/.  Accessed 31 July 2015.

Swales, Lois, and Heather Blatt.  "The Bookmark."  In Das Hainricus-Missale. Vollständige Faksimile-Ausgabe der Handschrift Ms. M. 711 der Pierpont Morgan Library New York. Kommentar, hrsg. von Hans Ulrich Rudolf, 165-175.  Graz, Austria: Akademische Druck- und Verlagsanstalt, 2010.  (See also pages 179-181 in the German translation, which have the informative schematics and photos.)

Swales, Lois, and Heather Blatt.  "Tiny Textiles Hidden in Books:  Toward a Categorization of Multiple-Strand Bookmarkers."  In Medieval Clothing and Textiles, vol. 3, ed. Robin Netherton and Gale. R. Owen-Crocker, 145-179.  Woodbridge:  The Boydell Press, 2007. 

* KXStitch is an open source cross-stitch graphing program for KDE.  It comes pre-loaded with the Anchor, DMC, and Madeira color palettes.  When using KXStitch for graphing textile patterns, I habitually use the Anchor palette rather than the DMC or Madeira one; I find it better represents naturally dyed textile colors, which are the colors I am accustomed to using.

01 July 2015

2/1 Twills: Rippenköper

© Carolyn Priest-Dorman, 2000

The “rippenköper,” or “repp twill,” has no standard equivalent name in English. The weave was first identified, named, and explained in 1967 by Hans-Jürgen Hundt in an archaeological publication. Rippenköper is a catch-all term for those twills, usually 2/1, whose basic structure alternates bands of warp-faced twill with bands of weft-faced twill, “usually after every third pick” (Bender Jørgensen, p. 14). Although it is a very simple
weave, I haven’t found it yet in a modern book.

28 June 2015

Two Asymmetrical Pavy Weaves

© Carolyn Priest-Dorman, 2003

A year ago my article “Some More Medieval Linen Weaves” presented a number of multishaft medieval liseré weaves of pavy design. One important design feature common to all these textiles was that the wales of the pattern all lined up and met perfectly. Accordingly, when I was attempting to draft an 8-shaft version of a pavy liseré weave for that article I focused a lot of attention on getting the diagonals to line up perfectly.

Last summer, however, I ran across two historic pavy weaves that are markedly irregular; their float arrangements are not perfect, and the wales do not line up perfectly. Further, the structure of these particular two textiles is not a liseré; it is a gebrochene. That is, it is an “Ms and Ws” structure with twill floats in both warp and weft systems, not just in the weft system as with a pavy liseré.

Middelburg-Nassau-Grimbergen, draft no. 1

One is part of an antependium from Middelburg-Nassau-Grimbergen, now in the Koninklijk Instituut voor het Kunstpatrimonium in Brussels. It dates to the first third of the 16th century. A complete analysis was not available to me, but the drawdown by J. Vynckier (de Jonghe, p. 70) was. The other is a large fragment of linen (Tx 63) in the collection of the Abbey of St. Truiden in Tongeren, Belgium. It was analyzed by Daniël de Jonghe (pp. 270-272), who assigns it a date similar to that of the Middelburg piece. Although the piece is a fragment, the complete width of 33.6 cm survives, including both tabby selvedges. Like most medieval ecclesiastical linens, it is woven with Z-spun singles line linen. The thread count is about 60 ends and 45 picks per inch, and the cloth is bleached.

If the structural analyses by de Jonghe and Vynckier are correct as printed, then the two textiles are curious inversions of one another. Both textiles are identically drawn in, yet their tie-ups are exact opposites. If, however, the structural analysis by Vynckier is drawn using a different convention than that by de Jonghe, then the two textiles may be closely related.

Middelberg-Nassau-Grimbergen draft no. 2

De Jonghe’s textile analysis can be checked against the photo of Tx 63; as always, he represents the warp with black and the weft with white. I was not able to check Vynckier’s analysis against the antependium because I do not currently have access to a photo of the actual textile. But if Vynckier’s drawdown uses white to represent the warp and black the weft, then the two textiles could be woven on the same warp using the same tie-up by simply changing the treadling sequence. Because de Jonghe dates them together due to their commonalities, it’s worth considering that they might be closely related, perhaps from the same production center. Accordingly, I give two different versions of the Middleburg draft, for those who’d like to try weaving them both on one warp without switching tie-ups.

St.-Truiden Tx 63

The draft called “Middleburg 1” is the one I first derived from the drawdown. It assumes the black-warp, white-weft CIETA convention that de Jonghe uses. The “Middleburg 2” draft I based on my hunch that the two textiles are related, and that Vynckier might have represented the textile “backward” from the CIETA convention. Instead, it is predicated on a white-warp, black-weft convention. The draft for Tx 63 is cut down and reworked from that of de Jonghe (p. 272), whose drawup and draft present more than a complete repeat and are tied up differently than I would do it.


Daniël de Jonghe, “De Textieldocumenten uit Sint-Truiden: Technologische Bevindingen,” pp. 63-105 in Stof uit de Kist: De middeleeuwse textileschat uit de abdij van Sint-Truiden. Leuven, Belgium: Uitgeverij Peeters, 1991.

Daniël de Jonghe, catalogue entry for Tx 63, pp. 270-272 in Stof uit de Kist: De middeleeuwse textileschat uit de abdij van Sint-Truiden. Leuven, Belgium:  Uitgeverij Peeters, 1991.

*  *  *  *  *  *  *  *  *  * 

This article was originally published in Issue 35 (March 2003) of Medieval Textiles, the newsletter of the medieval textiles study group of Complex Weavers.  The drafts and drawdowns I submitted were created using the freeware drafting program, WinWeave.  Using GIMP, I have turned the original drafts into graphics for republishing here.

25 June 2015

Jorvík 1307: This Project Grew in the Planning

Or, "what I did with the rest of that handspun warp."

It all started when I realized I'd spun a good bit more warp than I needed for my yard of  Jorvík 1307.  I thought I should probably try some other weft with it so I'd have a sample for teaching aids.

I should say here that I have a good-sized collection of samples I use as visual aids when discussing various weaving concepts, structures, treatments, and colors.  They're mostly offcuts from larger projects.  Once in a while, though, I weave something specifically for use as a teaching sample.  Jorvík 1307 was turning out to be the occasion for doing that again.

I looked through my stash of handspun worsted yarn in appropriate fibers—Icelandic, Manx Laoghtan, Shetland.  I noticed I had a fairish quantity of Shetland singles yarn in the 0.6mm diameter range, some Z-spun and some S-spun.  The yarn size was in the zone for Jorvík textiles, although not dead-on accurate for any of them when combined with the pre-existing 0.4mm warp.  But the Z-Z versus Z-S textile divide has been on my mind since I first read Lise Bender Jørgensen's work on spin direction nearly 25 years ago.  Here by happenstance I had an opportunity to experiment with it at period-correct setts using period-correct yarns.  I decided I'd weave some of the warp off with each of the two types of yarns so I would have an example of two cloths sharing the same sett, with the same size and fiber type of weft yarn but differing in their direction of spin.

Partly as a relief from the attentive work I'd put into making the first part of the textile as correct as I could, I decided to beat this part of the textile by instinct, until it looked and felt "right."  Purely subjective!  The 17-18 picks per inch of Jorvík 1307 was too loose for this smaller weft.  I didn't count my picks, but I worked hard to beat evenly, especially when it came to matching the beat I'd used in the S section when it became time to weave the Z section.  Overall the weaving went speedily and without incident.  I did not notice anything to differentiate the ways the two yarns behaved as weft.

It was difficult to get a good representative photo of the section woven with S-spun weft.  This washed-out shot (taken on the loom) shows the texture more clearly than any of the others I took.  You can make out the lozenges, but they're indistinct.

S-spun weft

When I switched to the Z-spun weft, the structure was immediately more clear.  Here's a shot taken off-loom that shows mostly the Z-spun weft area.

Transition from S-spun weft (below) to Z-spun weft (above)

Here's a better shot of them both together.

Z-spun weft at top, S-spun weft at bottom

After finishing, the S-spun section of the cloth has about 27 picks per inch, while the Z-spun section has only 25 picks per inch.  I am at a loss to know whether this result stems from my having failed to beat precisely across different portions of the warp or from some property of the weft yarns themselves.

I have let some other string geeks (mostly spinners) play with the finished cloth to see what they thought.  Everyone agrees that the Z-Z textile shows the weave structure most clearly.  That could explain why so many broken lozenge twills are woven with Z yarns in both systems.  If you're going to go to the trouble of knitting heddles for and then weaving a broken lozenge, I should think you'd want your work to be noticed!

But if that's the case, then why would Z-S broken lozenge twills even exist?  The answer to that question may have something to do not with appearance, but with handling.

The Z-S textile, like the Jorvík 1307 one, gives a thick and cushy impression.  Everything about it seems like it's smooshed a little more together, from the pick count to its appearance and handling.  It's more limp, less dynamic than the Z-Z textile.  It reminds me of a tablet-woven band with alternating threading:  the twists cancel each other out, leaving the textile neutral.  The Z-Z textile responds more quickly to movement, which gives it a more lively hand.

Refreshing my memory about the numbers and distributions of Z-S broken lozenge twills will be a research pleasure.  Perhaps it will even lead to some practical conclusions, or at least a testable hypothesis.  But for production purposes I will probably stick as much as I can to Z-Z spun twills, since I enjoy the look and dynamism of them.  Also, my Z yarns are much better spun than my S yarns!

22 June 2015

Jorvík 1307: Finishing and Conclusions

Previous posts here and here.

After I finished the second part of the weaving (see next post), I cut the web off the loom.  I washed it by hand in warm water with gentle soap, then spun it out and hung it up to dry.  (Warp-weighted looms serve many useful purposes in the modern home; mine is also good for hanging string things up to dry.)  I gently ironed it, and then I played with it a bit.

Because this structure is so weft-faced, the weft yarn exerts a lot of control over the handling of the cloth.  The texture is supple, but thick and cushy owing to the soft weft yarn.  It isn't exactly felted, but it fulled together very nicely.

After I played with it for a while, I carefully measured it. 

Take-up and finishing gave me a 6% shrinkage rate in the warp and a 4% shrinkage rate in the weft.  I had expected a little more shrinkage in the weft direction, maybe 7%-8%, but at 26 ends per inch I was pretty close to my target warp count of 27.5 ends per inch.  As for my target weft count, it was 17.5 picks per inch.  In most spots I exceeded my target by a bit, sometimes getting as much as 20 picks per finished inch, partly because I underestimated warp take-up and partly because I incompletely compensated for my linen weaving habit of a firm beat.

Next it was time for the photo session!  With bright indirect sunlight and a grey background (yes, I know it looks blue, but it's grey), I got a pretty representative couple of photos.  Here's the whole five feet of cloth.

the two-yard piece
And here's the diva closeup shot.


Conclusions and Other Thoughts

I'm really happy with how well this turned out; given my inexpertness with a spindle, it could have gone much worse.  I'm especially proud of how well my first foray into warp spinning succeeded; it's given me courage to reach for a larger project.  But I'm even more happy about what I've learned about relative yarn sizes.

Many factors led to me choosing Jorvík 1307 as my target textile.  One of those factors was definitely the size disparity between the yarns used in the two systems of this textile.  I wanted to examine the common premise that Viking Age textiles are best recreated with yarns of similar grist in warp and weft.  I already knew that most of the Jorvík wool textiles are woven with differently sized yarns in warp and weft.  With Jorvík 1307 I would be working with an extreme example of that phenomenon.

Jorvik 1307 displays a warp yarn is among the very finest of the Jorvík, i.e., Viking Age York, wool finds.  Of the 31 wool textiles from Viking Age York in Walton's catalogue, only four (1261, 1299, 1308, 1382) have warps as fine as Jorvík 1307.  Only one textile has a finer warp, Jorvík 1300, which has warps in the 0.3mm size range but is an unevenly spun warp with other yarns up to the 0.7mm size.

Differences also come to light when considering the weft yarn.  Among those 31 Jorvík textiles there are four (1257, 1258, 1296, 1297) coarse tabbies with thicker wefts than 1307; another (1379) with a thicker weft is a coarse 2/2 twill.  The herringbone twills 1264 and 1302 have weft the same size as 1307 but their warps are not nearly as fine, making them more evenweave.  Jorvík 1300 has wefts in the same size range as 1307, but again it is uneven, with the wefts varying between 1.0mm and 1.5mm. 

In other words, Jorvík 1307 has a finer warp than about 83% of the Jorvík textiles, and a thicker weft than about 74% of them.  The contrast between its warp and weft—with warp at the small end of the scale and weft at the large end—is remarkable even among its own peer textiles.  And that doesn't even take into account the question of sett; I'll have to take that up with another project down the line.

I will use the shrinkage and take-up factors from this textile to inform future projects.  I expect these factors depend to some extent on loom type, yarn diameter, and degree of spin.  I think I'll change yarn diameter first in order to see how that affects the overall picture.

I will definitely be spinning another wool warp.  Next time I'll pick a textile with a slightly heavier warp and less size disparity between warp and weft.  That still gives me plenty of Jorvík textiles to choose from even if I stick to my favorite period.

I will definitely be separating and spinning þel for weft again, too.  The uneven S-spun weft was, I think, the weakest part of this project; there's lots of room to improve that particular skill!  The processing of fine undercoat wool is somewhat of a mystery to me, since my big combs didn't work with it.  The cards I used are inappropriate tools for Viking Age textiles.  While I did not use them in the way that would have been typical in the Middle Ages, to produce rolags of coiled fibers for woollen spinning, it still bothered me that I was using them at all.  As an interim solution they worked pretty well to get the fibers straight, but I'd like to move to a less anachronistic tool.  Next time I will try smaller combs.  If need be, I will resort to hair combs to get the fibers straightened out a bit before I try spinning them.

I'll do at least one more piece on the table loom before trying my homespun on the warp-weighted loom.  I'd rather not push my luck.

In the final installment:  what I did with the rest of the warp.

18 June 2015

Jorvík 1307: Warping and Weaving

 Previous post here.

I drew up a weaving draft for Jorvík 1307 based on the graphic in Figure 137a in Penelope Walton's Textiles, Cordage and Raw Fibre from 16-22 Coppergate.  Jorvík 1307 has a section that is regular broken lozenge twill and a bit that is point repeat lozenge twill.  Walton puts that down to warp disarray based on "muddled or torn areas" (page 331).  I suspect it was a threading error.  But either way, I don't think it was an intentional part of a greater pattern of broken and point repeats across the cloth.  Accordingly, I decided to make the entire textile a broken lozenge twill.

weave draft and draw-up I used
The loom I picked to weave this textile on is a Louet Kombo four-shaft table loom.  I would have preferred to use my Icelandic style warp-weighted loom, but it's in storage in the basement.  That's because the place on the dining room wall where I would have set it up is currently taken up with the project on my 7' ancient Greek style warp-weighted loom (see other blog).  But I've had good results weaving small lengths of Viking Age type textiles on the Kombo; at least for loom type, I was on familiar territory.

I worked out the math for the warp based on a thread count of 25 ends per inch (11 ends per centimeter according to Penelope Walton's analysis).  This got me within a couple of ends of the correct count, 27.5 ends per inch.  I figured the take-up and wet finishing at the end of the project would likely adjust the count to right about where I wanted it to be.

The Jorvík 1307 fragment did not survive with an intact selvedge.  Since I knew the selvedge of my textile would become the edge of a hat, I wanted it to be as strong as it could be without being too obviously modern.  I looked at other wool textiles from the same time and place as Jorvík 1307 to see what selvedge treatments might be appropriate.  All of the selvedges Walton mentioned were "simple," i.e., had no additional reinforcing threads or crammed ends or any of the other tricks modern weavers use to keep them strong.  All the twill ones suffered from the usual problem of the weft not always binding with the outside thread.  It was fun to see the different expressions of that problem in Walton's drawings.  In the end I decided to ply some of my warp double and use one strand of that as a floating selvedge at each side.  The thick weft would obscure it pretty much completely, but it would add structural integrity.

When it finally came time to wind the warp, I freaked out a little bit.  I had purposely avoided thinking about what might happen if I had critical warp failure.  What if the yarn were too hairy?  I hadn't considered using sizing on this project; for that matter, I'd never used it whether on linen or wool.  What if I got a lot of breakage?  Was there enough extra warp that I could handle several broken ends?  I didn't know.  But when I handled the yarn, it seemed to be very stable even with the high degree of twist.  I tied up the cross more securely than I usually do, just in case it wriggled once I took it off the warping pegs.  But it didn't!  It lay cooperative in my hands without twisting.  It wasn't the smoothest yarn of that size I'd ever seen, but I'd managed to work with several hairier ones in the past.  I was optimistic, and so I set to work.

My Kombo reed is ten dents per inch; I alternated sleying it 2 and 3 ends per dent to get my 25 ends per inch.  I threaded the heddles, triple-checking at the end of every lozenge to make sure I'd done it properly.  I added a couple of ends at each side just to give a little padding to the pattern.  Since I fully expected I'd lose threads at the selvedge, I figured I'd give the piece a few disposable threads.

I wound it onto the back beam without incident and tied it off.  The warp behaved beautifully throughout this process, better in fact than some of the commercial warps I'd used.  It also wasn't casting off fibers, which I thought was a good sign.

I was in a hurry to get a look at my handspun interacting on the loom, so I wove a complete pass (18 shots, one lozenge tall) right after the header cord, to see what it would look like.  The warp was not grabbing at all!  I felt like the Spinning Queen!  (I'll probably have to blog separately about this some day, because it's really sort of a rant about how people tend not to put enough twist in their yarn when weaving to period spec.)  But the warp was still not spread out as well as I'd like it to be, and the dark colors made the pattern hard to see.

One of the commercial yarns I considered using for a weft on this project is an undyed singles tapestry yarn from Wild West Weaver.  It was the correct diameter for this piece (1.2mm), so I used some of it as a high-contrast weft to help me establish my beat and weft count.  It also spread the warp out a little more evenly for the beginning of the usable portion of the cloth.

the first couple of inches
Once I tweaked my beat to get the weft count I wanted, the rest of the weaving went quickly and well.  Let's face it:  after you've gotten past sourcing and creating the yarn, weaving it up is the fun, fast, and easy part of a project.  The problem I've had keeping my place when I treadle a broken lozenge twill didn't occur when I was lifting shafts by using the pegs on the Kombo.  One of these days, after I get decent at weaving twill on the Icelandic loom, I'll have to try a BLT and see how lost I get.

Over the entire two yards of warp, I only lost three ends to breakage.  All of them were in the group of the two outermost yarns on each side of the warp, and they all snapped within the first few inches of the weaving.  Since I'd added a few extra yarns to the warp at each side, these breakages didn't even impinge on the lozenge pattern, so I didn't bother to replace them when they snapped.

I found it almost impossible to take photos of the cloth on the loom.  My weaving space for this project was in a darkish room.  The best condition for viewing the cloth, the conditions that allowed me to actually see the lozenges clearly, was when it was late in the day and the lights were off.  Both the indirect daytime sunlight and any electrical lighting washed out the surface; flash photography was even worse.  Here's the best photo I was able to take.

on the loom

You can see the contrast between the warp and weft yarn sizes at the top of the fell.

During the weaving I realized I wasn't entirely happy with giving away every inch of this cloth.  I wanted to keep a bit of it for myself, so I kept weaving after my target length had been achieved.  I used up all the weft yarn I had spun, down to the last four or five inches that were left on the shuttle after the last possible pick.  That gave me another few inches of completed cloth that I could keep for my teaching stash.

In the next installment:  finishing and conclusions.

*  *  *  *  *  *  *  *  *  *

The weave draft and drawdown I created for this project were produced using the freeware drafting program, WinWeave.  I took a screenshot of the completed draft.  Using GIMP, I edited the screenshot to produce the .jpg for posting.

16 June 2015

Jorvík 1307: Fiber and Spinning

I recently had occasion to weave a reproduction textile.  Since I only needed about a yard of 8" wide cloth, I decided to challenge myself by spinning the warp and weft myself.  Even though I am a slow and not very accomplished spinner, with no experience of weaving from my handspun before this project, it seemed like the best decision.  I wanted to make this project very memorable, to try something new.  Spinning my own yarns would also place within my grasp a much greater range of textiles from my chosen time and location (Period 4B at Jorvík, i.e., the mid-tenth century Anglo-Scandinavian culture of York, England) than I would usually have to choose from based on the availability of commercial yarns in the right fiber content, fiber diameter, spin direction, and yarn diameter.

After looking through my stash of spun yarn and spinnable fiber, I chose Jorvík 1307 to reproduce.  It was from the right period in a classic weave structure for the period, the 20/18 (or 10/9, if you prefer thinking of it that way) broken lozenge twill.  Although I've woven other BLTs (that's my shorthand for broken lozenge twills), I'd never done a 20/18 one before.  It was also apparently an undyed textile, although brown in appearance.  I decided to do it in naturally pigmented wool because I thought the recipient would appreciate that more than plain off-white.

The first hurdle was warp yarn.  According to Penelope Walton's book Textiles, Cordage and Raw Fibre from 16-22 Coppergate I needed tightly spun, smooth worsted singles yarn about 0.4mm in diameter of a hairy medium fleece type.  In mundane terms, that's a warp yarn about the size of a doubled piece of standard sewing machine thread, and I needed nearly 500 yards of it for my project.  (This project grew in the planning.)  The mode for fiber diameter in the Jorvík 1307 warp was 24 microns.   Luckily, I had plenty of Manx Laoghtan combed top which met that standard pretty well.  It's a heritage breed found only on the Isle of Man and believed to be period to the Viking Age.

Manx Laoghtan combed top

Although the Manx Laoghtan top was a little kempy, I succeeded in pulling out many of the kemps during the spinning.  Since I apparently like to spin small and tight, working to this fine a specification was significantly less of a problem than I expected.  I spun it to about 35 degrees of Z twist on a handspindle using a reproduction soapstone whorl; I set the twist with hot water and weighted drying.

warp yarn, twist set and ready to warp

So far, so good, and the project even managed to survive my energetic young cat's frequent "help" with the spinning.

I'm sure there's some cat hair in that yarn.
 Next I needed a really soft weft yarn.  The mode fiber diameter for wefts in the Jorvík 1307 textile was 20 microns, making them even finer and softer than the fibers in the warp yarns.  The original is very weft-faced, which would have accentuated the properties of the weft yarn.  Of the period-appropriate sheep breeds, only the þel (undercoat) from a purebred Icelandic lamb was likely to get me close to the micron count I needed.  After considering and discarding a number of possibilities, I bought an Icelandic moorit lamb fleece.

some of the washed locks
I washed some of it up, hand-separated the tog (guard hairs) from the þel, and spun only the þel into my yarn.
 separated tog (guard hairs) and þel (undercoat)

Manageable fiber prep was key to this part of the process.  The fluffy þel proved too unstructured for me to spin from the cloud.  I knew combing was the correct technology to use, but the þel was too short to mount properly onto my Indigo Hound Viking combs.  I gritted my teeth and tried using a flick carder on it.  That didn't work well.  Eventually I cursed and brought out my hand carders.  I used them to card the fibers parallel and doffed the carded fibers sideways.  I rolled the batts from the side edges too, making sure to keep the fibers parallel instead of coiling them into the more typical rolags.  This made reasonably manageable short lengths of not-quite-top for me to spin.  

The yarn needed to be 1.2mm in diameter, a "well-spun" S singles.  That's way outside the size range for the weaving yarns I am accustomed to use; it's more like a fingering weight knitting yarn (roughly a 4 ply size in the standard used at Ravelry).  I don't spin yarns as thick as 1.2mm by choice, and I rarely spin in the S direction either. I needed all the help I could get!  Fortunately I only needed about 150 yards of this weft.  I wound up doing a lot of park and draft spinning, and it took me about five times as long to spin as had the warp.  The result, while falling short of "well spun," was at least respectable enough to use for weaving.  I spun it to about 25 degrees of S spin, setting the twist with warm water followed by a light weighting.

I didn't take any photos of the completed weft yarn, and I used up every single scrap of it in the weaving, so I don't have any photos of just the weft to share here.

In the next installment:  warping and weaving.

14 June 2015

Viking Age Pick-Up Double Weaves from Sweden and Norway

© Carolyn Priest-Dorman 2005

 The pick-up double weave technique flourished in medieval Scandinavia. Some of the earliest examples are narrow bands, while later examples are generally large coverlet-sized textiles. From a love of the patterns on them I was drawn inexorably into  a consideration of their structure and production methods. This particular research project centered around three questions. Did pick-up double weave date back as far as the Viking Age? If so, which of the modern versions of pick-up double weave was most likely to be the one used in the originals? And which type(s) of loom might have been used?

Early double weave textiles strongly resemble early Scandinavian tablet-woven textiles in their geometric patterning. For years they had intrigued me. I saw no reason why they shouldn’t date to the Viking Age, but most early assessments of their age identified them as somewhat later. Happily, more recent radiocarbon dating has assigned new and, to me, completely believable dates to these textiles. Not only are these textiles Viking Age, but at least one–the Marby fragment–is from the Early Viking Age (see Nockert & Possnert, 112-114, for the English summary of the dating). That part of the research project, at least, was easy!

This article focuses on a group of five double weave textiles executed in the same weave structure. Four of them are securely datable to the Viking Age, and the fifth seems to belong in the same category even though it hasn’t been securely dated. Each has a double warp system, half wool and half linen, and is woven with a double weft system as well. Each warp system interlaces with a weft system of the same fiber type in tabby structure, giving two separate layers; the two layers  interpenetrate, or exchange places, to build up the patterns. One side is colorful wool with white linen patterns, while the other side is white linen with colorful wool patterns.

Historic Examples

At least four securely dated Viking Age double weave textiles survive: the Revsund border, Överhogdal IV, the Marby fragment, and the Kyrkås hanging. Two examples (the Revsund border and the Marby fragment) are narrow bands in the 6-7" range. Överhogdal IV is a frieze about a foot wide, part of the Överhogdal group of textiles, a wall hanging stitched together of five narrower friezes woven in various techniques at different times, which was preserved at a Swedish church. The fourth is the Kyrkås coverlet, surviving in four fragments of which the largest is rectangular and roughly one meter square (Franzén & Nockert, 108).

Although I couldn’t locate any secure date for the Rennebu fragment, I consider it likely to be a Viking Age textile as well; it is very similar in size to the Revsund and Marby pieces, and in technique to the entire set of four pieces. However, it does differ in one important respect from the Revsund and Marby pieces: in addition to geometric  pattening, the Rennebu fragment has human figures worked in the weftwise direction, i.e., it was meant to be displayed horizontally like a frieze. Narrow horizontal orientation such as this is paralleled in early pictorial textile work by the ninth-century Oseberg tapestry and the eleventh-century Bayeux embroidery. The figures on the Revsund border, on the other hand, were meant to be displayed vertically. The Marby piece is inconclusive since it is entirely geometric in patterning and can be viewed without prejudice from either orientation.

These five textiles share several technical features. Most early woven textiles from Scandinavia involved fine singles threads; in contrast, however, most of both the linen and wool thread used in these pieces was two-ply and more coarse. All save the Marby fragment feature a striped wool warp, a design feature quite remarkably uncommon among indigenous Viking Age textiles. In four cases, including the Marby fragment, the wool weft is entered in stripes. Three of the five are striped in both systems. The Kyrkås coverlet is especially remarkable in this regard, with three colors of wool used in each system, yielding the startling effect of white linen patterning on wool plaid! All display quadrate geometric patterning, often involving interlaced elements, and some also have pictorial elements.

Structural Issues

Engelstad discusses the apparent historical relationship between loom type and double weave structure (see page 133 for an English summary). While she does differentiate between the warp-weighted and the two-beam upright loom, it seems clear that the pick-up double weaves under consideration here were more likely to have been woven on the warp-weighted loom. However, she doesn’t go into the mechanics at all. A warp-weighted loom would easily permit the wool and linen systems of the warp to hang each at its own tension. A two-beam loom, by forcing all the warp to be wound onto the same warp beam, would have permitted the wool warp to stretch out of all proportion to the linen. These cloths, particularly the narrow ones, could easily be woven on either a standard warp-weighted loom or an upright loom with a weighted warp, i.e., a warp-weighted loom that stands vertically rather than leaning.

Structurally, it is readily possible to distinguish between two major types of pick-up double weaves, the reversable and the non-reversable. Textiles in reversable weave have the same appearance (with opposite coloration) on both faces of the cloth, and the patterns tend to be blocky in appearance. Textiles in non-reversable weave are one-sided, with the back side often not at all resembling the front side; however, the technique  permits more curvilinear patterning than in reversable. Some modern names for non-
reversable double weave are bohusväv, finsketäcke, finnväv, täkänä, and Finnweave. However, Finnweave is also used in some contexts as a name for reversable double weave, which can confuse and mislead. The specific quality of reversability is more important than the name an instructor might use for the weave structure.

The Viking Age double weaves are all of the reversable type. Back before I knew they dated to the Viking Age, and mostly out of structural curiosity, I blew up the best photos I found in books (Branting & Lindblom was great for this!) and tried to identify the weave structure from views of the front face only. Subsequently as I read Franzén & Nockert, Nockert & Possnert, and then reread Engelstad, I discovered I had correctly identified the structure, although I hadn’t figured out anything new and I still hadn’t figured out exactly how to weave one.

After the Viking Age but still during the Middle Ages, the technique of non-reversable double weave became popular in Scandinavia; it was executed on treadle looms.  However, weavers there did not entirely abandon the technique of reversable double weave. In Gudbrandsdal, a district of Norway, people were working it on the warp-weighted loom “until well into the ninteteenth century” (Engelstad, 133). Unfortunately, no records of the specific methods they used appear to have survived. The Technical Bibliography, below, offers some suggestions for information on reversable double weave. Becker & Wagner from the Historical Bibliography is also useful for technical information; it is an English version of the Norwegian-language explanation by Signe Haugstoga first printed in Engelstad, with the same set of drawings.

The Överhogdal IV textile proves that warping over a cord is an (although likely not the only) appropriate method for this technique. The cord-loop method for warping is recorded in “Åklevev pa Oppstodgogn,” a documentary film shot by Per Gjaerder and Marta Hoffmann that was posted on the website of the Norsk Folksmuseum for a couple of years. This film of early twentieth-century weavers at the warp-weighted loom, and several others like it, was shot during the research that led to Marta Hoffmann’s publication of The Warp-Weighted Loom. It was thrilling to discover the museum had made them available for viewing (and devastating when they disappeared from the site), because they offered practical physical documentation of that rarest and most precious of information, how people go when they “go like this.”

Warping on a cord can be done very simply over two pegs or posts an appropriate distance apart. This type of textile requires a warp with pairs of threads in each system. Accordingly, a linen and a wool thread would be held together and wound simultaneously; every complete circuit of the two pegs would create one warp unit (two threads of each fiber type). When the right number of warp units have been wound, a cord is passed through the entire set of loops on the “far” peg, that is, the peg at which new threads are never started. The cord and loops are carefully lifted off and transferred to the loom.

Marta Hoffmann’s suggestion that pick-up double weave was probably woven on the warp-weighted loom using “not more than one heddle rod” is manifestly unsatisfactory. Even assuming that at most one free shed might be provided by either a shed stick (which she doesn’t mention) or by the main shed rod of the loom, it would still be necessary to pick out at least two of the required four separate tabby sheds, plus all the pick-up patterning, by hand. This method would be at least a level of magnitude more complex than the freehand rosepath method to which she compares it (Hoffmann, 186f). She also considers the meaning of the medieval textile designation “ferskeptr,” or four-shaft (Hoffmann, 209), mentioning that Engelstad suggests it might refer to a type of double-face weave (Hoffmann, 364, note 45). On a warp-weighted loom fitted with four separate heddle rods, each with its own set of brackets, it would be a laborious yet somehow familiar task for a modern weaver to weave reversable double weave. It would
be largely the same as working with a table or floor loom using a straight twill draw (1234...) and a skeleton tie-up. However, Viking Age weavers might have been able to exploit the basic shed separation provided by the shed rod of a warp-weighted loom to simplify the weaving method.

In my puzzling on the best way to weave a pick-up double cloth on a warp-weighted loom, I have begun to think that the simplest tool might be an upright loom with a weighted warp. It might even be possible for two people to work in tandem, one on each side of such a loom. Most considerations of the warp-weighted loom do not tackle the question of the angle at which such a loom may slant. Many such looms are (or were) leaned against a convenient wall or beam; the slant of the uprights combines with gravity to make the natural tabby shed large and easy to clear. It is only convenient to work from one side of such a loom, the side that slopes away from the weaver’s head. However, the specific angle at which any warp-weighted loom did slant is completely conjectural. It is likely to have varied from instance to instance. While it is likely that the angle of slant tended to vary within only a fairly small number of degrees, it is also possible that some looms did in fact not slant.

A full-size, or even oversize, loom would be required for wide pieces such as the Kyrkås coverlet. However, the small upright loom found in the Oseberg ship burial (dated to 834 CE) might be a perfect tool for weaving a narrow double weave band. I am pursuing this question in a practical way with my warp-weighted loom and my small Oseberg loom. If I am successful, watch this space for a report!

Historical Bibliography

Becker, John, and Wagner, Donald B. Pattern and Loom: A Practical Study of the Development of Weaving Techniques in China, Western Asia and Europe. Copenhagen: Rhodos International Publishers, 1987. A good technical discussion of the historic method, borrowing from the explanation in Engelstad.

Branting, Agnes, and Lindblom, Andreas. Medieval Embroideries and Textiles in Sweden, 2 vols. Uppsala and Stockholm: Almqvist & Wiksells Boktryckerei, 1932. Exceptional plates of some of these pieces, especially the Kyrkås piece, which was what sparked my interest in the topic. Authorship of Chapter II, “Swedish Double-Weavings and Double-Sided Weavings of Foreign Origin” is credited in a footnote to Viivi Sylwan.

Engelstad, Helen. Dobbeltvev i Norge. Fortids Kunst i Norges Bygder, Serie II  Publikasjon VI. Oslo: Gyldendal Norsk Forlag, 1958. Excellent historic survey of large numbers of historic Norwegian double weaves; includes a catalogue and an English summary.

Franzén, Anne Marie, and Nockert, Margareta. Bonaderna från Skog och Överhogdal och andra medeltida vägbeklädnader. Stockholm: Kungl. Vitterhets Historie och  Antikvitets Akademien, 1992. Best single source for technical as well as art historical information on this topic, although some of the datings they report have since been reassessed.

Hoffmann, Marta. The Warp-Weighted Loom: Studies in the History and Technology of an Ancient Implement. Oslo: The Norwegian Research Council for Science and the Humanities, 1974 [Robin and Russ Handweavers reprint; original printing 1966, Studia Norvegica 16].

Nockert, Margareta, and Possnert, Göran. Att Datera Textilier. Södertälje, Sweden: Gidlungs Förlag, 2002. The most recent word on dating. Some wonderful photos and a brief English summary, but not very technical.

http://www.norskfolke.museum.no/prosjekt/WebStar/katalog.html?kategori=10224&side=1&video=0 . Norsk Folkesmuseum, Oslo. This URL leads to what used to be the listing of the museum’s archival film footage relating to the practice of traditional textile arts in Norway. According to my notes, I was last able to view these films on 15 May 2003.

Sundstrom, Amica. “Tidigmedeltida Textilier från ett Hus i Sigtuna: en tekstilarkeologisk analys och diskussion.” Laborativ Arkeologi 2003-2004, University of Stockholm. http://www.archaeology.su.se/pdf/asundstrom.pdf, last accessed 26 February 2005. Brief consideration of these textiles as part of her survey of early medieval domestic textiles; two photos. 

Technical Bibliography

Atwater, Mary M. Mary Meigs Atwater Recipe Book: Patterns for Handweavers. Salt Lake City: Wheelwright Press, Ltd., 1969.

Black, Mary E. The Key to Weaving: A Textbook of Hand Weaving for the Beginning Weaver, Second Revised Edition, pp. 219-225. New York/London: Macmillan Publishing Company/ Collier MacMillan Publishers, 1980 [1957]. Some terminology confusion (she calls both reversable and non-reversable “Finnweave”), but still useful for the orderly way in which the double weave options are laid out.

Cyrus-Zetterström, Ulla. Manual of Swedish Handweaving, 2nd U.S. edition, trans. Alice Blomquist. Newton Centre, Mass.: Charles T. Branford Co., 1977.

Irwin, Allison. “Doubleweave Pick-up.” Handwoven, vol. XX, no. 1 (January/February 1999), pp. 36-39.

Janson, John. “Celtic Knot Scarf." Handwoven, vol. XXIII, no. 1 (January/February 2002), pp. 32-33.

Moore, Jennifer. “Doubleweave: a workshop in your studio.” Handwoven, vol. XXIII, no. 1 (January/February 2002), pp. 26-31.

Van der Hoogt, Madelyn. “A Pick-up Handbook for Handweavers.” Weaver’s, Issue 24 (Summer 1994), pp. 8-13, 48-52. See page 50 for the section on reversable double weave. Includes a lengthy bibliography.

Scorgie, Jean. “Patterned Double Weave.” Handwoven, vol. VII, no. 2 (March/Apri.l 1986), pp. 56-57.

*  *  *  *  *  *  *  *  *  *

This article was originally published in Issue 43 (March 2005) of Medieval Textiles, the newsletter of the medieval textiles study group of Complex Weavers.  Some of the information may be a bit dated now, and my thinking has evolved a bit on some aspects of the problem.  Also, I have yet to reconsider the section on loom types based on what I have learned since about Norwegian folk tapestry looms.  Nevertheless, I'm putting the article up here in the interest of keeping my work together and accessible in one place.

Here are links to images of the textiles mentioned in the text.  None of these were available on the web when the article was originally published.
Even more remarkably, since I wrote this article the textile research footage from which Marta Hoffmann worked has re-emerged.  The Norsk Folkesmuseum channel at YouTube has reposted many of these very valuable films which show mid-century Sami women working with basic warp-weighted looms.  You can find the specific film mentioned above here at YouTube.  The section on warping over a cord begins at around 6:30.

I haven't been able to find a live current link to the Amica Sundstrom article.

Here are some additional bibliographical references that are relevant.

Keasbey, Doramay.  "Pick-Up Pattern: Five Techniques."  Handwoven, January-February 2011, pp. 46–49.  I haven't read this article on pick-up doubleweave yet, but Keasbey's technical articles are always useful.

Oscarsson, Ulla.  De gåtfulla Överhogdals-bonaderna [The enigmatic Överhogdal tapestries], trans. Anita Lahiri.  Östersund, Sweden:  Jamtli Förlag, 2010.  Although this only discusses the Överhogdal textiles I-III which are soumak and not double weave, this book does provide a list of early furnishing textiles that includes all the ones named in my article.  It also provides a post-Viking Age date for the Rennebu double weave (page 80).

10 April 2015

Woad, Weld, and Madder Dyes

© Carolyn Priest-Dorman, 2004

The samples in my set are all commercial yarns.  The Icelandic wool single is "eingirni," imported from Iceland by Louise Heite in the late 1990s; it makes good hats and pouches.  The Persian wool is from Robin & Russ, where it is available in one-pound hanks of six-strand yarn; I unplied the last stage of plying to make a two-ply wool yarn for coarse Bayeux Tapestry-style embroidery.  The 12/2 wool was a closeout from Webs several years ago that I use for experimenting with new dyestuffs or recipes.  The silk organzine (a plied, reeled silk) is from HaBu Textles and is destined for tablet weaving.

The woad is from Bleus de Lectoure in France; nowadays it is carried by Carol Leigh's Hillcreek Fiber Studio.  The madder came from Earth Guild; I buy it in root form and process it myself.  The brazilwood came from Aurora Silks.  The weld is from my back yard.  The walnut hulls were harvested from the playground at my daughter's school. 

I always cold-mordant my silks, based on the recommendations in the Plictho and several other early dye compendia.  My wools are mordanted in a simmer-and-soak procedure adapted from Jim Liles.

None of the woad was pre-mordanted.  For the woad vat I use a recipe adapted from Gayle Bingham's article in Issue 29. 

detail:  woad

All the weld is mordanted with alum; my current weld recipes are refinements of the one presented in my article in Issue 29.

detail:  weld

detail:  weld/woad
detail:  weld/woad; weld/madder

The madders were mordanted with alum, and the brazilwoods were mordanted with alum plus tartar.  For the madder I use a slightly adapted version of Jim Liles' "Madder Red:  Wool" recipe.  For the experiments with brazilwood I followed Liles's general recommendations for making a brazilwood bath but did not expressly follow any of his recipes.  The brazilwood-madder yarn was inspired by several recipes in the Plictho for overdyeing madder with brazilwood.

detail:  madder (1-3); madder/woad (4)

detail:  brazilwood; brazilwood/madder

The walnut hulls were harvested green.  I let them stand in a stainless container about three weeks, until they were good and decayed, then removed them from the nuts and made a very dark brown decoction from them.  (The local squirrels were thrilled to eat all the leftover nuts, even in their fermented state.)  The resulting dye vat is very strong.  I let some of it sit until it had completely dried, which gave me nice dry chips of walnut concentrate-easy to store and to reconstitute.  Next season I'll experiment more with this technique.

Detail:  walnut; woad/walnut


Bingham, Gayle.  "Woad Dyeing (Isatis tinctoria)," Complex Weavers' Medeval Textiles, Issue 29 (September 2001), pp. 1-3.

Liles, J.N.  The Art and Craft of Natural Dyeing:  Traditional Recipes for Modern Use.  Knoxville:  The University of Tennessee Press, 1990.

Priest-Dorman, Carolyn.  "'A Grass that Grows in Bologna':  Dyeing with Weld," Complex Weavers' Medeval Textiles, Issue 29 (September 2001), pp. 1, 6-8.

Rosetti, Gioanventura.  The Plictho of Gioanventura Rosetti: Instructions in the Art of the Dyers which Teaches the Dyeing of Woolen Cloths, Linens, Cottons, and Silk by the Great Art as Well as by the Common, trans. Sidney M. Edelstein and Hector C. Borghetty.  Cambridge, Mass.: The M.I.T. Press, 1969.

*  *  *  *  *  *  *  *  *  *

This article was originally published in Issue 40 (June 2004) of Medieval Textiles, the newsletter of the medieval textiles study group of Complex Weavers.  It was part of a theme issue on medieval dyestuffs that included several sample cards.  I have photographed my sample contributions and added the images to the document for posting here. 

Gayle Bingham's recipe for dyeing blue using the Bleus de Lectoure woad can be found in this reprint.

09 April 2015

"A Grass that Grows in Bologna": Dyeing with Weld

© Carolyn Priest-Dorman, 2001

Written dyeing records from various parts of medieval Europe cite the use of several plants to achieve a yellow color: young fustic (Cotinus coggygria) and saffron (Crocus sativus) for warm shades; and weld (Reseda luteola), broom (Genista tinctoria), sawwort (Serratula tinctoria), trintanel (Daphne gnidium), and buckthorn (Rhamnus catharticus), also known as "Persian berries,"  for colder shades.  This article focuses on weld, perhaps the most accessible medieval yellow to the modern dyer.  It is cheap, quick and easy to grow, and safe and easy to dye with; the astonishingly vivid yellows it produces can also be overdyed in an array of interesting colors.

Weld gives a rubfast, fairly lightfast bright lemon-yellow dye that is slightly soluble in hot water (Colour Index, 3586), and it softens wools as it dyes them.  In Europe it was one of the most common commercially utilized dyestuffs from at least the early Middle Ages onward, and William Partridge notes its popularity in England in the early 19th century (Partridge, 106).  The earliest Western written recipes for weld come from the Mappae Clavicula, a Carolingian manuscript much copied in the early Middle Ages.  Therein are found several recipes for dyeing skins, leather, horn, and bone green using weld, usually cooked in urine and sometimes in tandem with other colorants.  There is a modern conceit that the term "Lincoln green," found in later medieval fictional literature, refers to a green dyed with woad over weld.  So far I have not traced any reference that leads to actual historic proof of this identification; most significantly, there seem to be no surviving later medieval recipes for dyeing cloth green from the area of Lincoln.  It is true that weld makes excellent greens.  Medieval recipes often call for it as the yellow component of greens, but they also call for other yellows to be used as substrates for greens.  Additionally, weld was used either by itself, in combination with a warm yellow such as fustic, or as a component in oranges, tans, and "quince" shades.

The two major colorants in weld are a pair of flavonoids, luteolin and luteolin 7-glucoside (Andary et al., 34).  In the leaves, more luteolin 7-glucoside than plain luteolin is found; in the seed capsules more plain luteolin than luteolin 7-glucoside is found.  Neither is very concentrated, however:  reports put the concentration at anywhere from 1% to 6.4%.  The seed capsules carry the most color, followed by the leaf; the flower and stem have much less colorant (Andary et al., 35). 

Growing and Using Weld

So far I have not been able to find a commercial source for weld as a dyestuff in North America.  The seeds, however, are readily available; I got my start with a packet from Richter's.  Since weld is easy to grow but hard to purchase, it makes a good choice for dye gardens.  It's a biennial weed, a low-growing rosette with a taproot that can tolerate a range of soil qualities and weather conditions.  Partridge actually warns that it produces less colorant when grown in good soil (Partridge, 111).  In the first season it only produces leaves; in the second season it bolts, producing at least one long shoot and several subsidiary shoots covered with tiny yellowish green flowers that become seed capsules.  This stalk is likely the source of one of weld's nicknames, "dyer's rocket," as it really does look kind of like a slender yellow-tipped rocket.  The huge numbers of tiny black seeds it produces guarantee that weld can readily take over any plot of ground upon which they fall.

I always start a spring crop of weld indoors using peat pellets.  After the danger of frost is past, they can be transplanted.  If there is a cold spell after the plants are outside, the plants might get confused and think they're in their second year of life.  If that happens, they'll bolt in the first year, which is a bonus!  If you want to force bolting in the first year, you could try setting seedlings in the refrigerator for a day or so before planting them outside.  After second-year plants have bolted (around midsummer), you can pull them up and re-plant the bed for a fall crop of leaves; the plants will bolt the following year.

Weld leaves can be cut and used at any time; I've even used frostbitten ones successfully.  However, in order to take advantage of the highest levels of colorants, it's a good idea to harvest the flowering shoots in the second year, as soon as the seed capsules have formed.  As the capsules ripen, the leaves turn yellow; ideally you should harvest while the leaves are still green. If you harvest after the seeds have fully ripened, be sure and save some for the next planting.

Medieval sources do not specify whether weld was used dry or fresh, although given the proportions involved in some of the recipes I believe it must have been dried.  Partridge also mentions that it was habitually pulled up and dried (Partridge, 110).   Weld leaves and seed capsules, which are all I save when I dry it, dry down to about one-quarter of their fresh weight.  I haven't had any luck using dried weld yet, but that's probably because I have only tried to do so in extremely small quantities.  Instead, I prefer to use weld fresh from my garden. 

Mordanting and Dyeing

Medieval recipes for the use of weld usually call for alum as a mordanting agent.  A few mention lye (e.g., Titus D.XXIV) or urine (e.g., Mappae Clavicula), either by themselves or as an additional mordant; lye or lime is said to warm up the color (Brunello, 28).  However, I haven't been able to find any medieval recipes for weld over an alum and cream of tartar mordant.  All the yellow recipes I found that called for cream of tartar, even if they called for weld, also involved the use of fustic or sawwort.  Interestingly, I have found that my own experiments with weld over alum and tartar never come out as bright and clear as those using only weld over alum.  This fact also seems to hold true for all the other yellow-dyeing weeds I've used.  One of the apparent functions of cream of tartar as a mordant is to soften the hand of an alum-mordanted wool, which by itself can feel harsh.  Since weld naturally softens wools dyed with it (Partridge, 111f), cream of tartar may be safely omitted at least in weld dyeing.  Accordingly, I recommend that weld be mordanted with alum only rather than with alum plus tartar.

Spectrochemical analysis may have a contribution to make to this issue.  Two-dimensional thin-layer chromatography examinations of weld extract differ from those of weld-dyed wools that have been mordanted with alum and tartar.  Specifically, the luteolin 7-glucoside spot is much more faint in the profile of tartar-mordanted wool than in that of the weld extract (Andary et al., Figs. 5 and 6, p. 35), while the luteolin spots in the two samples are more similar in size and brightness.  I suspect that the presence of tartar may inhibit take-up of luteolin 7-glucoside.

The basic process for weld dyeing is simple: make weld soup, strain, and gently cook the yarn in the broth.  Although weld is not as notoriously sensitive to prolonged heating as many of the weed yellow dyes, nevertheless low heat always gets better results when working with weld.  The higher the heat, or the longer it lasts, the duller the result.  A potful of fresh weld (a pound or so of leaves) will extract very nicely in about an hour of simmering time, and it never takes me more than about 15 minutes to exhaust the bath.  Often as little as five minutes of dyeing time suffices.

Here's a tested recipe utilizing the quantity of weld you can easily grow in a smallish plot, say, 8-12 plants.  With this recipe you'll get very saturated yellows, like those at the bottom of the sample card.  Using the proportions of three parts weld to one part wool, or by including the stalks in your pound of weld, you'll get lighter yellows, more like those at the top of the sample card.

  • 4 oz scoured wool (fine yarns give the best results)
  • 1 lb fresh weld leaves and/or seed capsules
  • 2 level tablespoons granulated alum (potassium aluminum sulfate)

Dissolve the alum in a potful of blood-warm water.  Add the thoroughly wetted yarn; slowly raise heat and maintain it below a boil until the yarn has been in the pot for two hours.  Let cool in the pot five hours.  Remove, wring, and dry without rinsing.

Cover weld with warm water.  Heat and simmer gently until the weld has been in the pot for one hour.  Strain out the weld and put the dyebath back into the pot.  It will be fairly light in color, with a greenish cast.

Rinse the mordanted yarn well and add to the dyebath.  Maintain a hot temperature in the dyebath but do not permit it to simmer hard.  Stir a few times and check the yarn in five minutes.  If there is still color in the dyepot, stir and leave for another five minutes, repeating if necessary.  After 15 minutes, the dyepot should be exhausted; if not, proceed until it is.

Weld can be overdyed to a number of orange and green shades using madder and woad or indigo; see the sample card for an assortment of shades.  Or you can simply stop at yellow.  You can level your dye job with Synthrapol, as desired.  However, heating weld-dyed yarn in water will remove some of the color, whether you're overdyeing with a hot bath or leveling.  In my opinion, it's worth the slight loss of saturation to level weld-dyed yarns.  Once your yarn is leveled, though, don't ever wash it again in hot water.

Detail:  Yellows and Greens

Detail:  Oranges


Society of Dyers and Colourists/American Association of Textile Chemists and
Colorists.  Colour Index, Second Edition, Volume 3.  Bradford, Yorkshire/Lowell, Mass.:  Society of Dyers and Colourists/American Association of Textile Chemists and Colorists, 1956 [1957].

Andary, Claude; Prunac, Stephanie; and Cardon, Dominique.  "Yellow Dyes of Historical Importance, II: Chemical Analysis of Weld and Saw-wort," Dyes in History and Archaeology, no. 14 (1995), pp. 33-38.

Brunello, Franco.  The Art of Dyeing in the History of Mankind, trans. Bernard Hickey.  Vicenza: Neri Pozza Editore, 1973.

Buchanan, Rita.  A Dyer's Garden: From Plant to Pot, Growing Dyes for Natural Fibers.  Loveland, Colo.:  Interweave Press, 1995.

Cardon, Dominique.  "Yellow Dyes of Historical Importance: Beginnings of a Long-Term Multi-Disciplinary Study; Part I, Yellow dye-plants in the technical and commercial literature from Southern Eruope: Italian, French and Spanish sources of the 13th-18th centuries."  Dyes in History and Archaeology, no. 13 (1994), pp. 59-73.

Liles, J.N.  The Art and Craft of Natural Dyeing: Traditional Recipes for Modern Use.  Knoxville: University of Tennessee Press, 1990.

Partridge, William.  A Practical Treatise on Dying of Woollen, Cotton, and Skein Silk with the Manufacture of Broadcloth and Cassimere Including the Most Improved Methods in the West of England, reprinted with an introduction by J. de L. Mann and technical notes by K.G. Ponting.  Edington, Wiltshire: Pasold Research Fund Ltd., 1973 [1823].

Rosetti, Gioanventura.  The Plictho of Gioanventura Rosetti: Instructions in the Art of the Dyers which Teaches the Dyeing of Woolen Cloths, Linens, Cottons, and Silk by the Great Art as Well as by the Common, trans. Sidney M. Edelstein and Hector C. Borghetty.  Cambridge, Mass.:  The M.I.T. Press, 1969 [1548].

Smith, Cyril Stanley, and Hawthorne, John G.  "Mappae Clavicula: A Little Key to the World of Medieval Techniques."  Transactions of the American Philosophical Society, New Series, Vol. 64, no. 4 (July 1974).

*  *  *  *  *  *  *  *  *  *

This article was originally published in Issue 29 (September 2001) of Medieval Textiles, the newsletter of the medieval textiles study group of Complex Weavers.  It was part of a theme issue on medieval dyestuffs that included several sample cards.  I have photographed my sample contributions and added the image to the document for posting here.

This early work with weld became the seed of a research project that I wrote up for delivery at the Colour Congress 2002:  The Art, History, and Use of Natural Dyes, an international conference on natural dyes and dyeing held at Iowa State University on 19-21 May 2002.  You can download that revised and greatly expanded paper here.

06 April 2015

A Medieval Kruiswerk Weave

© Carolyn Priest-Dorman, 2002

The ordinances of the London Guild of Weavers name several different linen products in their price list of 1456: linen cloth, plain towel, napery and towels of Paris work, "crosse werk," cross diamond, small knots, chains in work, "catrylettes," "damask knots with the chapelettes," and "all manner work made in draught work" (Consitt, pp. 205-207).  Many of these apparently were figured weaves.  While researching pavy weaves, I encountered a class of Renaissance linen patterning called in Dutch kruiswerk, or cross work.  Since cross work is listed as a 15th century weave in the London list, it drew my eye immediately.

J. Six's article mentions or shows photos of several kruiswerk examples, although none of them are dated to earlier than around 1600.  Kruiswerk is a subset of the gebrochene class of weaves, an elaboration of pavy in which certain design elements are stretched out by adding stairsteps to the typical Ms and Ws threading.  Unlike the pavy liseré weaves from the last article, they have twill floats of varying length in both systems on both faces of the textile.  The historic examples all require more than four shafts to weave.

None of the 17th and 18th century gebrochene variants in Zeigler or Lumscher look quite like kruiswerk; most of them are closer to pavy in style.  Marjie Thompson published the draft of an early 17th century kruiswerk, the "Earl of Mar canvas," in 1997.  However, late last spring I encountered an even earlier example, a textile dating to around the 15th century, Tx 60, documented among the holdings of the Abbey of St. Truiden in Belgium.  It is in the same pattern family as Six's kruiswerk examples and the Earl of Mar piece.

Here is a draft of Tx 60 from St. Truiden.  Daniël de Jonghe analyzed the weave and produced a draft which I have simply redrafted into a more standardized American notation.

This particular kruiswerk is woven on 12 shafts.  Unlike the Earl of Mar piece, its twill floats are balanced by an expanse of tabby ground.  The original textile remnant is a complete loomwidth of between 64 and 65.8 cm in width and 21.5 cm in length.  It is woven of linen, possibly bleached, at 23 warps and 17 wefts per centimeter.  Each selvedge is reinforced with a single S-plied thread (Stof uit de Kist, pp. 268-270).

It is not clear to me yet where Six got the term kruiswerk, but his article suggests there's an early manuscript with drafts of this type of weave structure.  I am looking into this more deeply, as it would be so nice to be able to identify kruiswerk as the "cross work" of the London regulations.


Consitt, Frances.  The London Weavers' Company from the Twelfth Century to the Close of the Sixteenth Century.  Oxford: Clarendon Press, 1933. 

Hilts, Patricia.  The Weavers Art Revealed.  Facsimile, Translation, and Study of the First Two Published Books on Weaving: Marx Zeigler's "Weber Kunst und Bild Buch" (1677) and Nathaniel Lumscher's "Neu eingerichtetes Weber Kunst und Bild Buch" (1708).  Ars Textrina, vols. 13-14 (December 1990).

Six, J. "Kruiswerk, Lavendel, Pavy en Pellen," Het Huis, Oud & Nieuw, vol. 10 (1912), pp. 105-122.

Stof uit de Kist: De middeleeuwse textielschat uit de abdij van Sint-Truiden.  Provinciaal Museum voor Religieuze Kunst, Begijnhofkerk, Sint-Truiden.  Leuven: Uitgeverij Peeters, 1991.  (Textile analysis and catalogue entry by Daniël de Jonghe.)

Thompson, Marjie.  "The Earl's Canvas," Weaver's, Issue 38 (Winter 1997), pp. 38-40.

*  *  *  *  *  *  *  *  *  *

This article was originally published in Issue 33 (September 2002) of Medieval Textiles, the newsletter of the medieval textiles study group of Complex Weavers.  The draft and drawdown I submitted were created using the freeware drafting program, WinWeave.  All my drafts were translated by the editor into better software for printing and publishing.  Due to the large size of the repeat, the draft was not printed in its entirety.  A separate drawdown was printed plus a draft of the complete threading plus the first half of the treadling.   

Using GIMP, I have re-edited my original draft into a graphic for inclusion here.  I have chosen to reintegrate the threading, tie-up, treadling, and drawdown back into a single graphic.  The tie-up is for rising shed/countermarche.  Having it all appear as a single image gives a better idea of how the threading, tie-up, and treadling work together to produce the patterning.

This post was delayed by several weeks due to my being so inexperienced at editing graphics.  I had to cut and paste several different pieces of the original draft together into one big image, then hand-retouch a great many of the seams literally pixel by pixel.  Fortunately, though, I learned a whole lot by doing so!  Hopefully the next work won't take so long to produce.