Recognising a hand-knotted rug

Start by turning it over

The back of a rug says what the front hides. On a hand-knotted rug the design appears on the back with the same crispness as on the face: each knot forms a point of colour there, and you find the medallion, the spandrels, the border, row by row. Nothing is added, nothing is glued. What you see on the back is the structure of the rug itself, and that is why a professional always looks at the back first.

On a tufted rug the back is a backing cloth. An unbleached cloth, often covered with a glued textile lining, showing no design at all. The pile is not knotted into it: it has been punched through a canvas with a gun, then locked in by a layer of latex. That layer is what ages worst; it hardens, it cracks, and the pile comes loose in patches. Turning up one corner is enough: a uniform cloth and a strip of fabric sewn all the way round, and the question is settled.

On a machine weave — Wilton or Axminster type looms and their descendants — the back does show a design, but a design that is too clean. The weft lines are perfectly straight, the motifs repeat without the slightest drift, and you often see coloured threads running continuously across the back where they play no part in the design. No weaver works with that regularity.

The design is crisp, the knots are not

That is the paradox of hand knotting. Look at a border along its whole length: it undulates slightly. A motif shifts by half a knot, a palmette is a little wider at one end than at the other, the width of the rug varies by a centimetre or two from top to bottom. These are not faults, they are the marks of a hand. On a loom, the tension of the warp threads changes with the humidity of the workshop and the tiredness of the person knotting.

Then look at the rows of knots on the back. They are not aligned to a string: their thickness varies, they sometimes run slightly on the diagonal, and on village or nomadic rugs they can undulate markedly. A machine-made rug, by contrast, shows a grid in the strict sense. If everything is perfectly regular over three metres, that is not virtue, it is a machine.

A variation of tone within a single colour, in horizontal bands crossing the field, is frequently met with on hand-knotted rugs: this is abrash, which is discussed in the valuation guide. A machine-made rug produces no abrash.

The fringes: extended warp or sewn-on strip

On a hand-knotted rug the fringes are not an ornament. They are the warp threads — the threads stretched on the loom before weaving — coming out at both ends once the work is finished. They are therefore part of the structure, in direct continuation of the pile. Pull gently on a fringe: it is one with the rug, you can feel that it runs back inside, and there is no stitching to be seen at its base.

On a tufted or a machine-made rug the fringes are most often a decorative strip, sewn or glued onto the edge. Part the pile where they are attached: you will see the stitching, sometimes a reinforcing tape, and the fringe comes away from the rug in one clean piece. Some machine-made pieces imitate the extended warp very well seen from the front; here again it is the back that settles it.

One ambiguous case is worth flagging. On a genuine antique rug the fringes may have been remade, replaced or consolidated after wear: an added fringe is then not a sign of industrial manufacture, it is a repair. You recognise it by the fact that the body of the rug is still knotted, and that the join sits right against the last row of knots. That is the work described on the fringes page.

The fold test

Fold the rug onto itself, face to face, and part the pile along the fold, the way you part hair to see the roots. On a hand-knotted rug the base of the pile appears: small loops of wool tight around the warp threads, two strands per knot, and the weft threads passing between the rows. The ground is visible, and it is textile.

The same action on a tufted rug shows something quite different: the strands plunge into a backing cloth and stop there, held by glue, with no individual point of attachment. On a machine-made rug you see a continuous weave in which the pile is caught in the weft. In neither case is there a knot to see, because there is none.

This test also tells you something useful for what follows: how much pile height is left. If the base of the knots is flush as soon as you part the pile, the rug is worn to the thread. That weighs more heavily than people imagine, and it can only be made good by repiling, knot by knot.

Counting the knots over ten centimetres

Density is measured on the back, never on the face. Take a ruler. Count, along a horizontal line, the number of knots — that is, of points of colour — over ten centimetres. Then count the number of rows over ten centimetres vertically. Multiply the two: you get the number of knots per square decimetre, then multiply by a hundred to move to the square metre.

An example. Thirty knots over ten centimetres in width and forty rows over ten centimetres in height give one thousand two hundred knots per square decimetre, that is one hundred and twenty thousand per square metre. Count over three different areas, one of them near a border, and take the average: the density is not the same everywhere. On a dirty back, damp it very slightly with a fingertip and the points stand out better.

The orders of magnitude, very broadly: a nomadic rug with a thick pile is often counted in tens of thousands of knots per square metre; a village rug frequently sits between fifty thousand and two hundred and fifty thousand; a fine workshop rug goes beyond four hundred thousand and can go well past that, particularly in silk. These ranges overlap from one region to another. Density alone classifies nothing, it locates.

What this examination does not tell you

It does not tell you the origin: a rug can be knotted by hand in a dozen countries, and designs have been copied from one region to another for a century. It does not tell you the age, since a hand-knotted rug may have been made last year. It does not tell you the material with certainty: mercerised wool or a shiny viscose passes for silk to the naked eye, and burn tests on a plucked thread have no place on a rug you intend to keep.

Nor does it tell you the state of the foundation, which is read by transparency, holding the rug up to a window: the areas where the weft has been eaten show up light before the pile even gives way. In other words, these six checks settle one question, and one only: hand-knotted or not. That is already the most important one, because it decides how the rug is cleaned, restored and valued. For the rest, the rug has to be seen. The origin entries give the markers region by region.