On this page you will find comparisons of various types of knotting and / or weaving styles (and, eventually, other information relating to carpets from Tibet, etc, as seen on this web site). One thing that certainly needs to be addressed in the wider carpet world is the unfortunate practice – that seems to becoming more prevalent – of private sellers, dealers and even auction houses describing Chinese woven carpets as ‘Tibetan’, and occasionally vice versa; especially so when incorrectly describing the origin in carpet selling circumstances. And yes, although Tibet may now be part of China, the knotting style of a genuine Tibetan knotted carpet is completely different from the knotting style used in all the other regions of China, and for that matter basically every other carpet producing area in the world!
That is, the Tibetan knot is more of a loop than what we would normally think of as an actual knot and is described by Philip Denwood, the pioneering author on Tibetan carpet weaving, as “The ‘Senna Loop’ [not to be confused with the ‘Senneh / Persian knot’ used in other regions carpets] is the technique of the Tibetan knotted carpet. Essentially it involves looping sections of the pile yarn from a length held in front of warp threads, behind individual or groups of warp threads, round them and back round a gauge rod.” (Denwood, 1974. The Tibetan Carpet, page 91) and then continuing with same across the carpet. – See photos, etc, here https://warpandweft.club/tibetan-carpet-weaving-photos/ – The loops are then cut along the gauge rod leaving only a single node for each loop at the back of the carpet, as opposed to there being two nodes to a single knot in all other regions woven carpets. However, the ‘loop’ style of Tibetan carpet weaving is still referred to as a ‘knotting’ or ‘a knot’. With that in mind, we often find the back of a carpet tells just as interesting and exciting story as the front – if not more so at times – as sometimes the very intricate weaving styles, even at times with large knots, leaves one in awe and admiration at the artistry and skill of the weaver – and the dyer for that matter – all those years ago. Anyway, we are not going to go into detail here regarding all the ins and outs (no pun intended) of the different region’s knotting styles, as I am a big believer in a picture being worth a thousand words, so enjoy the following images, and maybe if an uninformed person happens to stumble onto this page they will hopefully learn something re the different knotting styles, and how the genuine Tibetan knot stands apart from all others.
NOTE: The photos of individual ‘squares’ with a grey metal frame shown below are all one square inch, i.e. the inside of the metal ‘frame’ is one inch square on the backs of various carpets (and the knot count therein is expressed as ‘knots per square inch’ or ‘kpsi’). And for all intents and purposes there are 15.5 square inches in one square decimetre, should one care to make further calculations themselves (the knot count then is expressed as ‘knots per square decimetre’ or ‘kpsdm’).
NOTE WELL: What is referred to on this page as a ‘Chinese knot’ is the type of knot used in many rug weaving regions of the world (except Tibet) and is more factually and commonly known as the Persian or Senneh knot; but for simplicity’s sake in this page will be referred to as the Chinese knot when differentiating between Chinese and Tibetan carpets. And all major carpet weaving regions of China (except Tibet), e.g. Ningxia, Baotou-Suiyuan / Inner Mongolia, Gansu and Xinjiang / East Turkestan (i.e. Khotan, Yarkand, etc) used the same knot, that is the asymmetric knot open to the left, with a slight variation (i.e. generally using varying degrees of warp depression) in those from the Xinjiang region.
The back, or underside, of a Tibetan carpet with 49 knots per square inch (7 wide x 7 high) showing the shape of the knots used. In this photo the weft can barely be seen at all. Note the distinct difference in the shape of the back of the Tibetan knot when compared to the shape of the (back of the) Chinese knot. NOTE WELL; This is not simply a Chinese knot with severe warp depression hiding one node of the knot. There is only one ‘node’ to the Tibetan knot, or loop.
The back, or underside, of a Chinese carpet also with 49 knots per square inch (7 wide x 7 high) showing the shape of the knots used. In this photo the weft can clearly be seen. Note the distinct difference in the shape of the back of the Chinese knot – which shows both nodes of the one knot – when compared to the shape of the (back of the) Tibetan knot, which has only one node.
LEFT. Close up of the back – or underside – of one Tibetan knot (surrounded by eight other knots) as used in Tibetan carpets. (Because of the unique cut loop technique – see image / details below – used for Tibetan knotting, the Tibetan knot, or more precisely the Tibetan loop, only has one ‘node’; that is there are not two ‘sides’ to the Tibetan knot as there is in other region’s carpet knots.) RIGHT. Close up of the back – or underside – of one knot (surrounded by eight other knots) as used in Chinese carpets. Note the two ‘nodes’, or sides, of the knot.
NOTE THEN DISTINCT DIFFERENCE WHEN COMPARING THE ‘SHAPE’ OF THE TWO KNOTS.
That is, on the back of a Tibetan carpet only the single node of the ‘knot’ (i.e. loop) can be seen, while on the back of the Chinese carpet both nodes of a knot can always be seen (except for when / if one node is severely depressed).
As pointed out further above, in the one square inch on left of this Tibetan carpet there are 7 knots across (and 7 high), and in the one square inch on right of this Chinese carpet there are also 7 knots across (and 7 high), with the two ‘nodes’ – or ‘sides’ – of each of the seven knots clearly visible in the Chinese carpet. However, although there are seven knots across in both carpets, and seven ‘nodes’ across in the Tibetan carpet, in the Chinese carpet there are fourteen nodes across! (This then requires the number of knots to be counted differently in either type of carpet, as the next example below explains.) That is because in a Tibetan knotted carpet there is only one node for every knot / loop, whereas in a Chinese knotted carpet there are two visible nodes to every knot (unless on the very odd occasion where one of the nodes is severely depressed and then only one can be seen. But this is not a common occurrence). A Tibetan carpet on the other hand NEVER has two nodes (visible) to the same knot, or loop, as Tibetan knots are constructed completely differently to almost ALL other knots. NOTE; Both Tibetan and Chinese carpets can have more or less knots than seen in the two examples above, with both here having 49 knots per square inch.
In the one square inch examples above the Tibetan carpet on left has 10 knots across while the Chinese carpet on right has 5 knots across, with the two ‘nodes’ – or ‘sides’ – of each knot in the Chinese carpet clearly visible (however importantly, both carpets have ten nodes across the one square inch). The knot count on Tibetan carpets is thus ascertained by counting every knot / loop (or ‘node’), whereas on Chinese carpets only every second node is counted, as each knot in a Chinese carpet has two visible nodes (unless on the very odd occasion where one of the nodes is severely depressed and then only one can be seen). A Tibetan carpet NEVER has two visible nodes to the same knot, or loop, as Tibetan knots are constructed completely differently to almost ALL other knots; so every ‘node’ is (counted as) one knot in a Tibetan carpet. The Tibetan carpet in this example then has, counting the row of half-knots at the top, 45 knots per square inch (10 wide x 4.5 high), while the Chinese carpet here has 35 knots per square inch (5 wide x 7 high). To reiterate then, although there are ten ‘nodes’ visible across both carpets, and ten knots across the Tibetan carpet there are only five knots across in the Chinese carpet. NOTE; Both Tibetan and Chinese carpets can have more or less knots than seen in the two examples above.
CLOSE-UP PHOTOS OF A TIBETAN WOVEN CARPET
LEFT. Pile / front side of an early 20th century Tibetan carpet; note in what would be the center horizontal row of the ‘flower’ the colours would go red, yellow, dark blue (center), yellow and finally red again. CENTER. Underside / back of same area as photo at left; note in the center horizontal row that there are six red Tibetan knots, then two yellow Tibetan knots, then one dark blue Tibetan knot (center), then two yellow Tibetan knots and then six red Tibetan knots. RIGHT. Close-up of the middle portion of the center photo clearly showing the central single dark blue Tibetan knot with the two yellow Tibetan knots on either side of it.
The Tibetan looping (or ‘knotting’) technique. The weaver is weaving in the direction of the arrows, i.e. from left to right. To make the loops the yarn is brought forward and pulled down tight over the gauge rod (1), then woven back into the carpet by going under the gauge rod and passing behind the next two warp threads (2) before being brought back out to pass in front of the next two warps threads (3), then feed back in and back behind the same two warp threads (4) before being brought forward and over the gauge rod again (1) with the same procedure being repeated across the carpet. The pile is then formed by cutting the loops all along the gauge rod before it has been removed to be used again for the next row of knots. For photos depicting the various stages of Tibetan carpet making and the Tibetan knotting process please see this page https://warpandweft.club/tibetan-carpet-weaving-photos/
CLOSE-UP PHOTOS OF A CHINESE WOVEN CARPET
LEFT. Pile / front side of a late 19th century Chinese carpet; note how the center horizontal row of the flower shows, in the following order, first white, then red, then yellow, then red and then white coloured wool. CENTER. Underside / back of same area as photo at left; note in the center row that there are three white *Chinese knots, then two red Chinese knots, then one yellow Chinese knot (at center of flower), then two more red Chinese knots, then three more white Chinese knots. RIGHT. Close-up of the middle portion of the center photo clearly showing the central single yellow Chinese knot with the two red Chinese knots on either side of it. *NOTE WELL: As stated previously, what is referred to on this page as a ‘Chinese knot’ is the type of knot (asymmetric open left; see below) used in many rug weaving regions of the world except Tibet and is more factually and commonly known as the Persian knot; but for simplicity’s sake in this page will be referred to as the Chinese knot when differentiating between Chinese and Tibetan carpets.
The asymmetric knot open to the left used in almost all Chinese carpets is shown above as viewed from the position of the weaver weaving the carpet. To the far right is the shape of a single asymmetric open left knot; in the center is a closeup photo of the knotting in a fragment of an actual Inner Mongolian Chinese carpet from the early 1900’s showing three asymmetric open left knots along with associated warp threads as viewed looking down into the weave from what would be the top (of the carpet) with the pile side facing the viewer (i.e. the same view the weaver sees). At left is a drawing depicting three open left knots with the warp threads passing on through them and the weft threads that would be between each row of knots.
Outside of Tibet there are three kind of knots used in most carpet weaving areas; the asymmetric knot open to the left (top left), the asymmetric knot open to the right (top right), and the symmetric knot (bottom). Sometimes these knots are woven with what is called ‘warp depression’, as also shown above. The type of knot used in Chinese carpets is the asymmetric knot open to the left, usually without warp depression, but sometimes with in the East Turkestan / Xinjiang region (and some ‘modern’ Chinese commercial carpets).
DISCERNING THE KNOT TYPE. Shown above left is the asymmetric knot open to the left and to the right the asymmetric knot open to the right (while the outer knot on either side of the schematic at top is the symmetric knot, which is often used as the knot closest to the selvedge even on rugs woven with the asymmetric knot). To find the knot type, it is first necessary to find what is the bottom of the rug (i.e. where the weaver started weaving from). This is found by running your hand along the pile in the same direction as the warps lay. One direction will feel smooth, the other rougher and offer more resistance. The bottom of the rug is in the opposite direction to that which offers the most resistance, i.e. the direction you hand comes from when feeling this resistance as it were. With that end at the bottom, ideally find a spot where two colours meet – as this makes discerning which pile tuft comes from which knot much easier – and fold the rug back horizontally along the weft that goes through this spot. This exposes the base of the knots (bottom photos), and which tuft belongs to which knot – and hence which side of the knot is ‘open’ and which warp is enclosed, or ‘collared’. IMPORTANT NOTE; when using this method – or looking at a photo – the carpet must be held or positioned in the exact same way as the weaver wove it, that is the top of the carpet towards the top, and the bottom towards the bottom, otherwise your ‘type of knot’ result will be reversed / wrong! (If however it were to be a carpet woven with the symmetric knot then both warps are enclosed, or ‘collared’, and the pile threads come up through the center between the warps – as opposed to the knot being ‘open’ left or right of the collared warp.)
Both photos on left both come from the same asymmetric open left knot carpet, and both photos on right both come from the same asymmetric open right knot carpet. Note how the shape of the knot not only differs on the front, i.e. open left or right, but also on the back; that is the left node of the asymmetric open left knot appears to be offset to the left ‘underneath’ – or slightly lower than – the right node; and the right node of the asymmetric open right knot appears to be offset to the right ‘underneath’ – or slightly lower than – the left node.
The asymmetric knot open to the left as seen on the front side (photo left) of a carpet with very worn low pile showing the exposed knotting structure, and the back side of another carpet with the same open left asymmetric knot (photo right). LEFT. Note the horizontal strands of the exposed pile knot as they wrap over (over-weave) a warp, and the ‘tufts’ of (what is remaining of) the pile that ‘stand up’ to the left of those horizontal threads (after having passed behind the second warp in this instance). Hence a knot that is open to the left; or put another way, the leftmost of the two – or more in some instances – warps used by the knot is exposed or ‘open’ to the left as seen from the weavers perspective. (In this photo the exposed warp/s cannot be readily seen because of the relative tightness of the weave ‘hiding’ them so to speak.) RIGHT. The backside of a carpet with an open left asymmetric knot. See photo below for a close up of this same knot. For carpets that don’t have such low worn pile that show the exposed knotting structure as the example left, the knot type can be found using the ‘pile fold back’ method as described above headed ‘Discerning The Knot Type’.
LEFT. A close up of a single open left asymmetric knot as seen from the weavers perspective. Note how the pile thread wraps over – or ‘encloses’ – the right hand warp and then goes behind and back up beside the left warp leaving it exposed or ‘open’. RIGHT. A close up of the very same open left asymmetric knot as seen from behind in the photo above right. Note how the left hand node open of the knot appears to be offset to the left ‘underneath’ – or slightly lower than – the right node. (If it were an open right asymmetrical knot it would be / look the opposite, that is on the back side the right node of the asymmetric open right knot would appear to be offset to the right ‘underneath’ – or slightly lower than – the left node, as pictured several images above.)
Two views of an open left asymmetric knot structure as seen from the same perspective as the weaver who wove the carpet (note; although both photos show and open left asymmetric knot they are not of the same carpet). The white arrow points from the horizontal section of the knotting thread as it travels over the warp to the ‘same’ horizontal feature of the knotting thread on an over-woven warp in the photo at right. The black arrow points from the (faded full) pile on left to (what’s remaining of a) pile tuft in photo at right (and which can be seen all across that photo). Note in both photos how the pile tufts stand up to the left of the over-woven warp thread and which – as can be clearly seen in the photo at left – comes from behind the exposed warp thread (on left) leaving the knot ‘open to the left.
This ‘asymmetric open to the left‘ knot is the type of knot / knot shape that is used in almost all Chinese hand woven carpets; however it is NEVER used on a true Tibetan woven carpet. If you see this knot on a purported Tibetan carpet, then it is NOT a Tibetan carpet!
A close-up of the knotting on the back of a carpet made in the Ningxia region of China in the early 1900’s, and typical of what the knotting looks like on the back of carpets woven in most regions of China (except the Xinjiang region; and of course Tibet). Note in the photo how both nodes / half of each knot can clearly be seen, which is the ‘normal’ look for carpets woven with an asymmetric knot when viewed from the back of the carpet. [While the knot in the drawing appears (is) open to the right, if the carpet was viewed from the front as the weaver wove it it would of course be open left.] The red arrows point to the corresponding nodes of just one knot relative to the same positions in the drawing. Asymmetric type knotting is fairly easy to both see and feel from the back of a carpet, that is if one was to run their fingers across the back of the carpet it would feel relatively flat / even (except when the knot has been ‘depressed’. See description opposite right).
A close-up of the knotting on the back of a carpet made in the Xinjiang region of China in the early 1900’s. Note in the photo how only one node / half of each knot can be clearly seen, the other node / half of each knot is partially hidden, as it is off-set and ‘depressed’ because of the method generally used in the carpet weaving process in this region, which is known as using ‘warp depression’. This however is still an asymmetric knot open to the left but with a warp depression (of which there can be vary degrees of severity, here being about 45 degrees, as shown in the drawing) and viewed here as if looking from the back of the carpet. [While the knot in the drawing appears (is) open to the right, if the carpet was viewed from the front, i.e. the weavers viewpoint, it would of course be open left.] The red arrows point to the corresponding nodes of just one knot relative to the same positions in the drawing. This warp depression type of knotting is easy to both see and feel (unless it is a fully depressed warp), as if one were to run their fingers across the back of the carpet it would feel rather ‘corrugated’ / uneven.
The image above displays four separate one inch square areas of the backs (or the underside) of four carpets woven in four different regions of China. Clockwise from left; Tibet (because of the unique ‘cut loop’ technique used for Tibetan knotting, the Tibetan knot, or more precisely the Tibetan ‘loop’, only has one node, that is there are not two sides to the loop / knot as in other region’s carpet knots), Ningxia, Baotou-Suiyuan and Xinjiang. (Note: one node / half of each knot in the Xinjiang example is somewhat ‘hidden’, as it is partly ‘off-set’ and ‘sunken’ because of the ‘warp depression’ method generally used in the weaving process in that region). Counting the number of knots per square inch in the various carpets is discussed in more detail further down the page.
The image above displays one inch square areas of the backs of four carpets woven in four different regions of China, with the arrow pointing to just one single knot on the backside (or the underside) of each carpet. Clockwise from upper left; Tibet (Note: because of the unique ‘cut loop’ technique used for Tibetan knotting, the Tibetan knot, or more precisely the Tibetan ‘loop’, only has one node, that is there are not two sides to the loop / knot as in other region’s carpet knots), Ningxia, Baotou-Suiyuan and Xinjiang. (Note: one node / half of each knot in the Xinjiang example is somewhat ‘hidden’, as it is partly ‘off-set’ and ‘sunken’ because of the ‘warp depression’ method generally used in the weaving process in that region.) Counting the number of knots per square inch in the various carpets is discussed in more detail further down the page.
The image above displays a one inch square of the back of six different carpets woven in Tibet, from, top left, a relatively small Tibetan knot with 84kpsi / 1,302kpsdm (12 knots across x 7 rows – or lines – of knots high = 84 knots per square inch) to, bottom right, a relatively large Tibetan knot with 18kpsi / 279kpsdm (6 knots across x 3 rows – or lines – of knots high = 18 knots per square inch); and other size variations in between.
The image above displays three separate one inch square areas of the backs (or the underside) of three carpets woven in three other carpet weaving regions of China (besides Tibet), each with a different knot count, that is a different amount of knots per square inch. Left to right; from Xinjiang (with 64 knots per square inch, i.e. 8 knots across x 8 rows – or lines – of knots high), from Baotou-Suiyuan (with 49 knots per square inch, i.e. 7 knots across x 7 rows – or lines – of knots high) and from Ningxia (with 30 knots per square inch, i.e. 6 knots across x 5 rows – or lines – of knots high). Note also the different way the weft lays in each carpet, and how in the example from Xinjiang one node / half of each knot is quite ‘depressed’.
The image above displays a one inch square of the backs of two carpets, the left woven in the Ningxia region (of China), the right woven in Tibet. Note how in the Ningxia both ‘nodes’ of the back of a knot can be seen, whereas in the Tibetan carpet only the single ‘node’ can be seen. In the Ningxia carpet there are 36kpsi / 558kpsdm (6 knots wide X 6 rows – or lines – of knots high = 36 knots per square inch), while in the Tibetan there are 45kpsi / 698kpsdm (10 knots wide x 4.5 rows – or lines – of knots high = 45 knots per square inch.)
The image above displays a one inch square of the backs of a carpet woven in the Ningxia region (of China) on left, and on the right a carpet woven in Tibet. The Ningxia carpet has both warp and weft made of cotton, whereas the one from Tibet in this example has both warp and weft made of wool. [The Ningxia has 6 knots across – and in this example 5 rows high – and shows both the left and right node of the knot, whereas the Tibetan has 9 knots across – and in this example approximately 7 rows high – although only the single node (of each knot) can be seen. Hence when counting knots on Chinese carpets you count every second node across, whereas on a Tibetan carpet you count every node across.]
The image above displays a one inch square of the backs of two carpets from Tibet. The one on the left is from a carpet known in Tibet as a Tsuktruk (Tsukdruk) which is somewhat more ‘blanket-like’ with a soft handle, whereas the one on right is from a regular Tibetan carpet. Both have wool warps and wool wefts.
The image above displays a one inch square of the backs of two carpets from Tibet (well, the one on the left is actually an uncut girth strap hole in a saddle carpet). The one on the left has machine spun cotton warps and hand spun wool wefts, whereas the one on the right has hand spun cotton warps and hand spun wool wefts.
The image above displays both hand spun and machine spun cotton; in this case ‘warp’ threads both of which have been Z spun and then S plied. On the the left is the hand spun cotton; note first the irregularity in the thickness of each individual yarn, and hence the irregularity of thickness of each warp thread. This irregularity (in thickness) of each yarn / thread is caused by the variations in the amount of cotton used per yarn when cotton is first spun by hand, and then the individual yarns later plied into a thread (in this case a 7 ply hand spun warp thread). On the right is the machine spun cotton; note the evenness of the both the individual yarns and threads. This evenness / regularity in thickness of both the individual yarn and hence each thread (in this case an 8 ply machine spun warp) is only obtained with automated machine spinning of both the individual ply and then the overall finished thread.
During the 1950’s the Chinese increasingly exerted control over everyday life in Tibet. In March 1959 a massed uprising by Tibetans against this repression took place in the capital Lhasa, at which time the 14th Dalai Lama – Tenzin Gyatso – fearing for his safety and rightly so, fled to exile in India where he was granted political asylum, and remains to this day. His flight set off of a mass exodus of Tibetans, many never to return to their home country. Besides the many Tibetans that followed the Dalai Lama settling in India, many also settled in Nepal, especially in various parts of the Kathmandu valley. From this nucleus Tibetan carpet weaving ‘businesses’ slowly began to be established in Kathmandu, with the weavers making carpets using the Tibetan knotting method. ‘Tibetan’ carpets continue to this day to be woven in Kathmandu and its environs by 2nd and even 3rd generation ‘Nepali Tibetans’, and / or by Nepali people themselves – who prior to the arrival of the Tibetans (in the late 1950’s, and onward) did not have a history of carpet weaving. In this photo, taken in 1972, three Tibetan refugee women work on knotting a carpet in a workshop on the outskirts of Kathmandu.
The above very close-up photo is from the back of a (Turkoman) carpet that is woven using both wool and cotton for the pile. On the left are knots made from wool, on the right knots made from cotton. Note how the woolen thread has long and comparatively ‘wide’ fibers (as compared to the cotton), whereas the cotton threads fibers are very short and thin. Also note how the wool thread has a more or less regular ‘grain’ to it (in this instance running relatively horizontal), whereas the exterior of the cotton threads is much more ‘fuzzy’. Very few carpets though are found with cotton used in the pile, this image is just to show the noticeable difference in the structure of wool and cotton.
A weaver, circa 2010, sits at her vertical loom in a carpet workshop in Bodhanath, a ‘suburb’ that was once on the very outskirts of Kathmandu (Nepal), but is now just part of the capitals urban sprawl. Note she is using a ‘cartoon’ of the design image drawn to exacting specifications on graph paper – where every square on the graph is equivalent to just one knot – so as to get an exact replica of the design every time this design is reproduced. In days gone by, weavers either copied the design from the back of other carpets or from ‘cartoons’ drawn by hand on parchment; or for a master weaver, sometimes simply from memory. (This is also the reason that with old Tibetan carpets – and produced by any of these latter three ‘methods’ – it is unusual to find any two carpets exactly the same.) For instance, in the picture to the left, note the small oblong drawing just above the center weavers head that simply has the design of the flowers in the ‘medallion’ drawn on it.