Tarim Basin Caftan – Prototype

The Unfinished Product


Level of Authenticity: Prototype for Kingdom A&S entry.

Who/When/Where, TL;DR version: prehistoric (Bronze Age and Early Iron Age) peoples living in the Tarim Basin in Central Asia between about 1000 and 500 BCE.  For more detail, including a bibliography, see my post on my alluring mummy beanie.

What: This is the second prototype for a Kingdom level A&S entry I’m working on.  Normally when I make a garment, I like to be able to refer to it using the term that the people wearing it would have used, or at least the English language equivalent (shift/stays/kirtle/gown).  Here that’s a bit difficult because it’s not even clear what language the people buried at Zaghunluq spoke, so I’m going to use a term that is used for similar garments worn by other peoples originating in Central Asia, caftan.

Why: At its simplest, we can assume that the members of the family buried at Zaghunluq are wearing the same clothes they wore in life.  Since they were all buried wearing clothes of the same colour, it might also be possible that these particular clothes were made specifically as grave clothes, but there isn’t enough information to be sure.


Materials: The Zaghunluq family were buried in garments made of naturally brown wool overdyed with madder.  The woman’s garment may have had some cashmere content as well as sheep wool.  If I recall correctly, one of the garments was woven in a plain tabby weave while the other was woven in a highly idiosyncratic “long hop” (3/2) twill.  I’m working on this project with Brigida Vadesbana and we are still considering various alternatives for fabric to use for the final Kingdom level entry.  When making a prototype garment, fabric selection is less critical although generally you want to aim for fabric with a similar weave, weight, and drape to that of your intended finished project.

Here, we wanted to make a wearable prototype, so we considered the range of fabric types found generally in the Tarim Basin and dating to the first millennium BCE.  The fiber content is almost always 100% wool and the weaves can be tabby, various types of twill, and some other decorative weave patterns.  The most common colours are reds (from madder), blues (from indigo), yellows (from an unidentified source), and oranges (a mix of red and yellow dyes).  As well as plain colours, fragments found at certain sites, especially Hami, demonstrate two-colour checks and stripes, and plaids in up to six colours.  Although located in the same general geographic region and dating to about the same time frame, the sources I’ve read have all been in agreement that Zaghunluq and Hami represent two separate archaeological cultures.  Further, there are no plaids in the Zaghunluq burial.  For these reasons, I wouldn’t want to use a plaid fabric for the final Kingdom entry, but at the time of beginning this prototype, the green wool plaid was the best available selection from among my and Brigida’s fabric stashes.

The Zaghunluq garments have a type of decorative seam finish that is described as “piping” stitched down with wool of a contrasting colour.  It’s unclear what the authors mean by piping but it’s unlikely that they mean it in the modern sense of a cord encased in bias strips.  For my first prototype, I used modern worsted weight knitting yarn for both stitching and piping.  This choice was difficult to stitch with and I occasionally needed to resort to the use of pliers.  This time, Brigida was able to provide me with wool thread that is much closer to the weight of modern sewing thread.  We haven’t yet reached the stage of choosing a material for the piping, but we will most likely consider a material much larger than worsted weight yarn, either super bulky weight yarn or lightly spun pencil roving.

Tools:  None of the Tarim Basin sites described in my research materials contained evidence of looms, and Elizabeth Barber concludes the Tarim peoples were likely using simple ground looms.  She noted that none of the textiles she and Irene Good examined exceeded a loom width of about two feet.  As indicated under “Materials” above, Brigida and I are still considering what approach we will take to producing or obtaining a textile for the competition piece.  The plaid fabric used here was presumably produced on a modern commercial loom.

When I made my first prototype, I used a modern steel darning needle.  This size was required because I was sewing with worsted weight wool.  My research revealed that several of the mummies were buried with bronze and even iron needles.  I was able to obtain bronze needles made by a modern artisan in a traditional style and I used them for this project.  Unlike modern needles, the bronze needles rapidly developed a noticeable curve after being used, but despite this, I did not notice any appreciable differences in working characteristics or techniques needed when working with the bronze needles as opposed to the modern steel needles.

Other tools I used in this project were shears, snips, and pins.  In each case, I used modern tools.  My research did not reveal any equivalents to any of these tools in any of the Tarim Basin graves, although I will note that Barber and Good felt that some textile fragments and fragmentary garments showed evidence of the pieces having been woven in specific shapes and sizes or being shaped on the loom.  This would eliminate the need for shears.  For now, whether the Tarim peoples had such implements, or what they used instead, will need to remain an unanswered question, and one that could form the basis of its own Kingdom level project at some point.

Cut:  This is the part that is the most unclear from the sources that I have.  Barber clearly sets out a conjectured cutting pattern for the man’s caftan in “The Mummies of Urumchi”.  In that book, she indicates that the woman’s caftan is the same except that it is open under the arms.  Following this pattern, the body of the caftan should be straight and it should wrap closed in the front.  In “Secrets of the Silk Road”, Barber describes the same garment as though it is cut the same way as a child’s dress depicted in that book.  This dress is a pullover garment with triangle-shaped gores inserted at the side seams to widen the body.  The conflicting descriptions of the same garment are compounded by the lack of satisfactory photographs or sketches of the Zaghunluq woman or her garments.  In the one available picture, it appears to me that the garment is straight cut but may be closed at the centre front.  Because of the lack of clarity, I decided that it would be necessary to make one or more prototypes testing different cuts.

Since my fabric was produced on a modern loom and was considerably wider than any fabric that could have been produced by the Tarim people, I first had to determine what “loom width” to assign my fabric.  Evidence suggests that the Tarim people were reluctant to cut into material when they did not need to*, so my chosen loom width would determine the width of all pieces cut for my garment.   I consulted with various laurels and other knowledgeable weavers and two approaches emerged.  In approach 1, I would determine what width the extant garments had been woven at and cut my garments along those dimensions.  In approach 2, I would choose a loom width within the known parameters (no more than about 2 feet) that would give me garments that fit the people they were meant for.  I chose approach 2 because I wanted to make garments that would fit me or my models; and the lack of standardized loom widths among the extant textiles, plus occasional evidence of shaping garments on the loom, suggests to me that the weavers were choosing widths based on what would fit them and their families.  For my first prototype, I chose to proceed with an assumed loom width of 18 inches.  Using a straight cut pattern, this gave me a garment that, though theoretically big enough around to overlap at my bust and hips and stay closed, tends to gap open in the front when I move around.  At the same time, using a full 18″ loom width for each sleeve gave me sleeves that were much too big around and floppy.  For my second prototype, I again used an assumed loom width of approximately 18″ but added truncated triangle-shaped gores under the arms to provide extra width in the body.  I have not yet fully determined the shape or cut I will use for the sleeves.  Prototype 1 matches one of the descriptions for how the extant piece is constructed, but does not fit as well as I would like.  I believe the fit of Prototype 2 will be satisfactory, but it departs further than I would like from the descriptions of the extant piece.  This means that further prototypes will be necessary.

Construction: The research materials provide a high level of detail about the decorative scheme on the garments, but less information on the actual stitches used.  I therefore turned to Heather Rose Jones’ Archaeological Sewing page, as well as Elizabeth Barber’s “Prehistoric Textiles” for information on pre-medieval stitches.  Neither of these sources references the Tarim Basin textiles but both discuss the Hallstadt textiles.  These are of similar age and likely belong to a related archaeological culture**.  From the available options, I selected a stitch where the edge of each piece is turned down and hemmed, and the hemmed pieces are then attached together with an overcast or whip stitch.  I have used this construction method many times and I find it both decorative and sturdy.  For Prototype 1, I attached the seams together at the same time as I was attaching the piping over the seams.  The result was satisfactory but led to occasional gaps in the seams and places where the piping slid down between the hemmed pieces and became less visible from the surface.  For Prototype 2, I chose to begin whip stitching the seams together first, with the piping and decorative whip stitching to be added later.  Depending on how this approach works, I may use this two step method, or I may experiment further before deciding on an approach for the final piece.


* In one instance, a garment has shaped sleeves but instead of being cut away, the excess is folded to the inside of the seam.  In another case, a garment is shaped but examination of the fabric shows that the shaping was done on the loom rather than the finished cloth being cut to size.

**Elizabeth Barber believes that the Hami textiles were made by a people very closely culturally related to the Hallstadt-era Celts.  If not, the connection is likely to be more distant, but most of the scholars working on these materials believe at least that the Tarim Basin peoples were speakers of Indo-European languages, of which Celtic is another example.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s