The Techniques of Rug Weaving. And as Collingwood notes, it is "extremely secure": this is a knot that will not slip or migrate, once you have tightened it.
I was fascinated by this knot, which was one of several that I was testing as I played with Sample #10: Pile Techniques in the Handweavers Guild of America Certificate of Excellence handbook. This is when craft becomes amazing: that we are still using techniques that were invented in the Bronze Age, which is c. 2300-600 BC. And that some ingenious anthropologist or textile researcher examined knots made by Danish Bronze Age people and did something to preserve the heritage.
I've been noticing that several of the samples in the COE call for types of weaving that one rarely encounters. When, for instance, will I make a hand-knotted rug? But working with the yarn on the loom, looking back-and-forth from the drawings in Collingwood's book, I began to get an infinitesimal feel for what a rug weaver encounters.
My favorite knot of the four tried - Ghiordes knot with continuous pile, Ghiordes with cut lengths of pile yarn, Sehna knot, and the Danish Bronze Age - was the Sehna. You can make it lean to the right or the left, depending on the way that you wrap the yarn around the warp ends.
Collingwood remarks that this leaning quality is used to advantage in Mongol saddle rugs, which were made in two halves and joined so that the pile slants downward on either side of the rider. This allows rain and dust to shed toward the ground and the rug is more comfortable on the rider's legs.
I love Collingwood's tone in the introduction to his book. It reminds me of Marcella Hazan and her dictums about cooking ("Ready-ground pepper is one of those modern conveniences that keep giving progress a bad name. Why it exists I do not know" from Italian Cooking, p. 16, 1979 edition). Here's Collingwood:
Basically, a loom is nothing but a device to produce this necessary warp tension...All the complexities that have been added to the loom are only to save time or effort or to make intricate weave structures possible. They do not alter the fundamental function of a loom as a warp-stretcher.
(p.41 of the 1972 reprinting)
An author who knows his mind.