How were the patterned curtains for the mishkan woven?

Question

Shemot 26:1 describes the mishkan's curtains thus:

וְאֶת הַמִּשְׁכָּן תַּעֲשֶׂה עֶשֶׂר יְרִיעֹת שֵׁשׁ מָשְׁזָר וּתְכֵלֶת וְאַרְגָּמָן וְתֹלַעַת שָׁנִי כְּרֻבִים מַעֲשֵׂה חשֵׁב תַּעֲשֶׂה אֹתָם:

And the Mishkan you shall make out of ten curtains [consisting] of twisted fine linen, and blue, purple, and crimson wool. A cherubim design of the work of a master weaver you shall make them.

I had always assumed that, given that we're using four different colors here, the cheruvim were somehow woven in using contrasting colors (like maybe the background was one color and the other three were used to make the pattern). But Rashi says the four colors were all plied together into one 24-strand yarn containing all four colors:

of twisted fine linen, and blue, purple, and crimson wool: Thus there are four kinds [of material] together in each thread, one of linen and three of wool, and each thread was doubled six times. Thus, four kinds [of material], when they are twisted together, yield twenty-four strands to a thread. — [from Baraitha Melecheth HaMishkan, ch. 2, Yoma 71b]

(I didn't find more info in Yoma, and I don't know Baraitha Melecheth HaMishkan.)

One might think the cheruvim are then embroidered on top of this blended background, but Rashi points out that the text says "work of a master weaver", not embroiderer:

A cherubim design of the work of a master weaver: Cherubim were drawn on them [the curtains] in their weave; not with embroidery, which is needlework, but with weaving on both sides, one face from here [one side] and one face from there [the other side]: a lion from this side and an eagle from that side, as silk girdles, called feysses in Old French, are woven. — [from Yoma 72b]

Now I don't know a lot about the construction of the tabernacle, but I do know a little bit about weaving. If you weave a pattern into the cloth, you get the pattern on one side and its inverse on the other. So we can't be talking about that, according to Rashi. Another technique is brocading, where you use a second weft thread to weave in the pattern on top of the cloth as you go (on one side). This second weft doesn't contribute to the structural integrity of the cloth (that's what the first, main weft is for), but you can use it to make patterns like those shown in this article and this one. This technique was used in Rashi's time for silk girdles.

It would be quite difficult to brocade on both sides of the cloth using the looms and techniques available in Rashi's time, let alone those in the time of the midbar, but the torah also tells us that the mishkan craftsmen were embued with exceptional spirit and skill, so maybe they could do that.

If they did use brocading to make the images, though, then what did they use for the brocade thread? The torah is very detailed in describing the materials for the mishkan, and it doesn't say anything about gold thread, for example. You can brocade with the same yarn that you weave with, in which case you get a raised textured effect, but it'd be pretty hard to see because it's of uniform coloring.

And all of this is being done with only a single blended thread color, according to Rashi. (As opposed to having red, blue, purple, and linen threads separately available to make contrasting patterns.) If you ply threads of different colors together as Rashi says, you get a thread with color that varies along its length as each component thread rises to the top alternately.

It's possible, of course, that Rashi is mistaken, either about the colors all being twisted together into one thread or about the patterning. None of the chumashim I checked, including Stone, brought an opposing opinion.

So what did these curtains look like, both coloring and patterning? How did they use linen, blue, purple, and crimson to make curtains with cheruvim (and maybe other things) woven into them? Did they use some additional material beyond the four threads/colors the torah describes? (Like gold brocading on top of this base?)

Answer 1

Not everybody agrees with Rashi.

Abarbanel (as cited in the JPS Miqra'ot Gedolot English translation) says that the linen threads formed the warp and the colored yarns were used in the weft to make the designs. Gersonides adds that the design work involves calculating the number of threads that will produce the image when it's woven. That's not very clear; I'll get back to that in a moment. Finally (from JPS MG), Ibn Ezra says that it was a "worked design", meaning neither woven nor embroidered but "like silk today": two threads are held at a distance and each point on the drawing is pulled into a pucker.

Ibn Ezra lived in the 12th century CE in Spain, which is to say al-Andalus. I went looking for information about silk weaving in that time and place; I found this high-level description (with photos) but it's short on details of production. The Production of Silk, Andalusia by David Jacoby (registration at Academia.edu required for download) describes silk from this time period in more detail but stops short of production details. From that paper, it sounds like some of the finest silks are tapestries, so I then consulted this overview of medieval and renaissance tapestry production from the Met, which describes weaving a design through two sheds:

The warps are arranged so there is a small space between the even and odd warps, called the shed, through which the weaver passes the colored weft threads that are wrapped around a handheld shuttle. Alternate warps are attached to drawstrings with which the weaver can pull them forward (on the high-warp loom) or backward (on the low-warp loom) to create a second shed, through which the weft is then passed back again. By passing the weft back and forth through the two sheds, the weaver inserts the weft over one warp and under the next in one direction and then back in the opposite direction over and under the alternate warps. Periodically, the weaver beats down the developing web so that the warps are completely covered by the weft. [...]

In European medieval and Renaissance practice, the design was invariably copied from a full-scale colored pattern, known as the cartoon, a practice that continues to this day. Before starting work, the weaver traces the pattern from the cartoon onto the bare warps.

This is a woven-in design, not brocade (and not embroidery). The planning of the design aligns with Gershonides. The weaving aligns somewhat with Ibn Ezra, except that I still don't know what he means about threads being pulled into a pucker. Patterned weaving as described in the Met article is also consistent with Abarbanel.

I cannot reconcile Rashi's description with results that would show a pattern. The weaving technique described by Abarbanel would have been known to Ibn Ezra and was used for fancy silks, so I tentatively conclude that it was a woven pattern on a linen "base" and that Ibn Ezra was probably not a weaver who could provide a detailed, accurate description of silk production in his time.