Speculation on the Earliest Surviving Chinese Rugs

by Murray Eiland, Jr.

Originally appeared in OCTS 7, based on a paper delivered at the 7th ICOC in Hamburg, Germany

The dating of Chinese rugs has long been controversial, and, over the years, there have often been extravagant claims for rugs that are surely less than a century old. Slowly however, grouping have emerged which allow the division of Chinese rugs into those with handspun foundation yarns, which there is good reason to assign to the first two thirds of the 19th century or earlier, and rugs with machine-spun cotton foundations, which may be assigned to a later period. As one moves backwards toward the early 19th century, one finds progressively more austere designs until the carpets begin to show an increasingly prominent use of various brown shadesEventually one may group together rugs that can be characterized by their substantial use of brown color that erodes much more quickly than the rest of the carpet. Often these carpets show an eroded brown border on all four sides, an the designs – with relatively few exceptions – are more austere than on those rugs thought to date from th elate 19th century.

Figure 1. Fragment of a Chinese rug, 2'1" x 5'9". The particularly graceful drawing on this fragment and several other similar pieces has suggested that they may be among the earliest surviving type of Chinese rugs. (Rutherford Collection)

The question as to how far Chinese carpets can possibly be dated into the past. To place these pieces showing eroded brown in the early 19th century or even the late 18th century seems reasonable, but is ether any reason to extend this dating to the mid or early 18th century or earlier? Occasionally one find a piece labeled by a dealer as Ming work, or with more specific attributions to the 14th or 15th century. Could this possibly be justified?
In spite of this lack of documentation, however, a group of
carpets has attracted attention among Chinese rug enthusiasts during the last several ecades. The largest inividual cache of these rugs, including four large fragments, turned up at a southern California auction in 1974. The current owner of these rugs later determined that they had been de-accessioned by the DeYoung Museum in SF. Obviously someone with a good eye ha gathered them together an someone with less sharp sensibilities had approved their sale for what turned out to be a few hundred dollars.

Figure 2. Border fragment of a large Chinese rug, 3'5" x 7'. While there is nothing extraordinary in the drawing of this border, the colours and texture of this example area close match to Figure 1 and both pie es probably belong to the same small group of early Chinese rugs. (Rutherford Collection)

The only one of these pieces to show part of a field pattern is drawn with surprising skill, and this small fragment suggest that the original was an exceptionally powerful rug. A border fragment of the same structural type, but no the same rug, also shows exceptional power and in scale suggests an extremely large rug. Two other border fragments were all part of the same group and these were obviously from a single rug. No likely candidates for field design of these borer fragments have emerged, but there are other fragment with design similar to the field of Fig. 1. One of these fragments, in the Museum fur Kunsthandwerk, Frankfurt, and published by Lorentz (Fig. 4) appears to be of the same genre, perhaps of the same carpet although the colours are not really a good match. Another fragment of similar design is in Paris, but there are reasons for assuming it is from a different carpet. Several other fragments which have passed through the international art market in recent years are almost certainly from the same or similar carpets.

Figure 3. Border fragmetn of a large Chinese rug, clearly related in colour and texture to to Figs. 1 & 2. (Rutherford Collection)

The next important cap[ets from this group appeared in auction of the Frank Michelian Collection in New York (1982). Figs 5 & 6 were published in the sales catalogue, and they are obviously from the group to which we refer. In their colours, and extraordinary drawing they are suggestive of the four pieces from the 1974 southern California sale, and they also have other features in common.

Since the Michelian sale, well over a dozen pieces have come onto the international market showing essentially the same range of features. This includes Fig. 7 from an American collection. Indeed, it appears, in retrospect, as though one example of this type, apparently the largest surviving and one of
the most impressive, was published in 1911 in G. Griffin Lewis’ Practical Book of Oriental rugs Appearing in full colour, its dimensions are listed as 23’ x 24’, and it is described as included by courtesy of Costikyan an Co. of New York. It is further stated that, as one of the “most celebrated and costly rugs in America” it was “presented by the late J. Pierpont Morgan to the Metropolitan Museum.” The rug appears from a comparison of photographs, to be the same piece later published in the catalogue of rugs in the Metropolitan Museum as from the Collection of Fred Mueller. Obviously it was recognized early in the century as a special carpet, and it was then attribute to the 18th century.

Figure 4. Fragment of a Chinese rug with field design suggestive of Fig. 1. (Museum fur Kunsthandwork, Frankfurt, Germany

Perhaps the most dramatic difference between these carpets and th etypical pieces with eroded brown borders is in the colour scheme. First, some of this group show relatively lttle brown and ,k when brown does occur, it is not usually of the type that erodes more rapidly than the rest of the carpet. I have found no pieces I consider to be amon this group that show the eroded brown borders, so common on early Chinese rugs. Most of these rugs again show fewer distinct shades of blue than the typical late 19th century Chinese rug. Some show only one rather dark blue, often contrasted with a blue-green, a colour that appears on most of these rugs. At times there is a dark an light blue, while more than two blues is unusual but not unknown for this type. The main colours are shades of yellow and gold, with smaller traces of faded mauve. From the appearance of early Chinese rugs in ancestor portraits, it seems likely that the rugs were originally much more brightly coloured with substantially more re than we see now. The faded mauve and even some of the of the golden shades could originally have appeared as reds, an the blue-green likely appeared as a brighter green before fading of the yellow component.

Figure 5. Chinese rug with stylized dragons, 20'6" x 9'9" Sale of rugs from the Frank Michaelian Collection, Edelman Galleries, NY, 10/29/82

The wool of these pieces is quite coarse, much harsher in feel than the usual wool of later Ninghsia carpets, and the weave, with typical asymmetrical knots open to the left may be eve coarser than 10 knots per square inch. The remarkable degree to which a curvilinear effect is achieved with knotting this coarse is the result of the several technical innovations. Remarkably gracelful lines can be achieved by tying knots over 1 or 3 warps, at times with the yarn moving from one shed to the next, leaving what appears to be a short diagonalb line of colour on the back of the rug. While this technique is used sparingly in later Chinese lrugs, it is nowhere so common as on this group, where it helps achieve a curvilinear effect with surprisingly coarse knotting.

The foundation yarns of many of these rugs are hand-spun cotton, but the warp material usuaed in other has become controversial. In HALI #72, p. 72, the matter was seeinmingly
settled with the observation that most actually have silk warps but an unusual silk, very thickly cabled an of course flossy appearance which could easily be mistaken for cotton. We watched one leading scholar analyzing an example an refusing to believe the warp was silk tuntil, in frustration, the owner clipped off a small bit and burned it under his nose.

In the next issue (HALI #73, p138), however, the description of a similar piece brought the comment, “the composition of its foundation (coarse silk or jute) provided a source of continuing discussion both before an after the sale. That there should still be some doubt suggest the need for competent microscopic analysis, as burning provides a subjective, unreliable result. While the foundation material in the rugs I have examined appears to be t cotton, an answer must be deferred until the rugs are more intensively studied.

Figure 6. Chinese rug with a single large dragon, 17'6" x 10'. Sale of rugs from the Frank Michaelian Collection, Edelman Galleries, NY, 10/29/82

The othe feature of these rugs, suggesting that they are more than standard workshop precuts is the elegance of the design themselves. Aesthetic quality is surely the most difficult feature of any art form around which to communicate, but there are few dissenters to the observation that there is something special in the drawing of these rugs. While not all are great masterpieces (Lot 111 in Sept. 22, 1993 auction, Sothebys NY, seems to this observer, too repetitive to be interesting), some seem recognizable as superior work even from a black an white photograph.

Unfortunately, although the drawing usually shows a grace that many consider superior to the bulk of the eroded brown border carpets, there is no clear way to date these pieces, as none of them have inscribed dates. Recently there have been attempts to match up designs of these carpets with those found on various ancestor portraits, which could presumably be dated with some certainty. While several carpets of the eroded brown border type have been matched with ancestor portraits, efforts to do the same with the group under discussion here have been not been so convincing. So far as I am aware, no carpets from this group have been successfully matched with a portrait.

Some of these paintings, which extend back into Ming times, mat be dated within a few years, and some show enough carpet
for all border stripes as well as parts of the field to be seen. They are not without problems when used as a basis for dating, however, as some are apparently not actual portraits, but were produced years after the death of the honored ancestors as a means of establishing noble lineage. Another problem is that even ancestor portraits whose Ming attribution seems secure often show carpets quite unlike anything that has survived to the present time.

Among the comments made after the delivery of this paper at the Hamburg ICOC were wildly conflicting opinions as to date. While on carpet dealer expressed his opinion – unsupported by anything that could be construed as evidence – that these carpets were clearly Ming, Charles Ellis indicate that he had examine the four southern California pieces during the late 1960s, when they were still within the DeYoung Museum, and had concluded, along with May Beattie, that they were modern pieces intended to look old. Shortly thereafter, a London carpet dealer announced that he had dealt with seventeen of these rugs, nine of which had been traced back to the Imperial Palace in Beijing. HALI’s anonymous commentator on auctions is now routinely referring to them as Ming work (No.70, p.143); No.72, pp.129, 130), while there is mention that one of the carpets is “documented as having come from the Imperial Palace in Beijing, as do so many other examples of its type.” (HALI, No.72, p.130)

Figure 7. Chinese rug with two dragons. (Wendel Swan Collection)

This latter material is difficult to critique, and, until details of such claims appear in print and can be examined by others, one is left with the choice of accepting them on faith or filling them among the vast accumulation of heresay so common in the rug field. When I visited the Imperial Palace in 1978, I consulted with custodians there, who were completely unable to locate any archival material related to the carpets that were in or had been in the Palace, although I was able to examine a small number of pieces then to be found there. In three of the main reception rooms, there were carpets on the large squarish platforms under the thrones and other furnishings. These carpets were large enough to overhang on all four sides, and they provided an explanation for the function of similar pieces that appeared in the West. All three of these carpets were at the time covered by transparent plastic, and I was forbidden to touch or examine and therefore could not ascertain whether they belonged to the group under discussion here. As for their having been original to their location, I could only hypothesize that they had probably not been in place long. The emptiness of most of the rooms in the Imperial City is not difficult to understand when one considers that it has been looted twice in the last century. The first capture and looting of Beijing, by French and British armies in 1860, followed the Second Opium War. While the Imperial City was apparently left relatively intact at the time, the Summer Palace was thoroughly looted, and then, at the order of Lord Elgin (son of the Lord Elgin who brought to England the Parthenon statues) it was burned with the intent of punishing the Chinese Emperor for the mistreatment of European prisoners. Parts of the loot were auctioned in the courtyard of the Lama Temple, also plundered, on October 11, 1860.

Figure 8. Carpet from the Gion Matsuri Treasury, Kyoto. At least one carpet in an almost identical design has been found outside of Japan.

The Forbidden City was not similarly sacked until 1900, after the Boxer Rebellion was crushed, and armies from six foreign powers – Britain, France, Russian, Japan, Germany, and the United States – arrived in Beijing on August 14. These soldiers proceeded to pillage the city until they carried away everything perceived to be of value. As O’Connor describes the scene, “Looting quickly became the sole preoccupation of the allied garrisons and palaces, homes, and government buildings of the conquered city.” This included also the Forbidden City, whose “sacred imperial precincts were looted as thoroughly as the Mongol Market.” (O’Connor, p.287) August Simpson, in a letter providing an eyewitness account described, lumbering military trains going back to Tientsin, laden with countless chests of loot. Every place of importance, indeed, had been picked clean as a bone. Now that the road is well open, dozens of amateurs, too, from the ends of the earth have been pouring in to buy up everything they can.

Lest it be imaged that the imperial household had been able to hide away the most treasured items before the sack of Beijing, one must recall that the Empress herself had not recognized the threat until a few days before the fall of the city and had been compelled herself to leave at the last minute disguised as a peasant. Eyewitness accounts make it clear that the vast palace was systematically pillaged, with great train loads of booty being carried away to find their ultimate destination in the antiques markets of Europe and America. Under the chaotic circumstances described, it would seem virtually impossible to determine, after the event, just which artifacts came from the Imperial City at that time an which from the thousands of private homes also looted.
The Empress returned nearly a year later and re-established an imperial household with what could be salvaged from other sources, but even this reconstituted treasury was again removed from the Forbidden City in February of 1933, and almost all of it came to reside in the National Palace Museum in Taipei. In all, 242,542 items were crated and taken from one location to another ahead of the Japanese advance, and this odyssey is recounted by the Nationalists’ retired Minister of Education, who bore much responsibility for the removal. It too six trains and five months to carry all of these treasures from Beijing, an, eventually, thousands of crates were carried to Taiwan on three cargo ships.

While the size of the Imperial art collection of the Qing emperors must have been astonishing, it is difficult to imagine – with the looking of 1860, the vastly more extensive looting of 1900, an the efficient removal of every item before the Japanese advance in 1933 – that there could have been left in Beijing of the imperial treasures that filled the Forbidden City in 1850. No wonder almost all of the rooms of the palace, which I methodically surveyed in 1978, were completely empty of artifacts.

An yet on continues to hear of objects on the art market that can confidently be trace to the Forbidden City, where, I was told by the custodians in 1978, there were no inventories of pre-1900 materials. The problem thus becomes on of the determining the provenance of looted objects, a difficult job at best. I eagerly await publication of the documentation that demonstrates these rugs are from the Imperial Palace.

Figure 9. Carpet from the Gion Matsuri Treasury, Kyoto. The border, as in Fig. 8, strongly resembles borders of several early rugs probably woven in Anatolia. It appears to be a Kufic variant and thus suggests an origin from the Islamic cultural sphere.

Another Group of Early Chinese (?) Rugs
Another group of early carpets that may have some claim to being the earliest surviving rugs of China has recently become known and, for the most part, they are to be found in the treasury of the Gion Matsuri Foundation in Kyoto, Japan. Figs. 8, 9, and 10 are representative examples of this group, which numbers over twenty in Japan, in addition to one, and perhaps a second, that are alleged to have been found recently in Tibet.

As to where these rugs were woven, speculations are varies. Charles Ellis believes they are of Japanese manufacture and relatively recent. He reports that at least on of them bears a tag from a Japanese carpet manufacturer, although it appears possible that this carpet was copied from an earlier original that have become too worn for continued use. That the entire group originate in Japan seems unlikely to this observer in that wool
rugs have no tradition among Japanese handicrafts. There are cotton-pile Japanese rugs an, rarely, those with silk pile, but it appears unlikely that the Gion Matsuri rugs should be the only surviving wool-pile Japanese rugs.

There have also been suggestion that the rugs are Mongolian or from Eastern Turkestan. This carries with it a certain plausibility, as the borders of Nos. 8 and 9 are so similar to the Kufic or Kufesque borders of some Islamic rugs that one can scarcely ignore the possibility of a relationship. The animals of Figs. 9 and 10, however, seem beyond the realm of Islamic are and are more suggestive of a Chinese origin. The carpets are coarsely knotted, and, unlike the rugs previously discussed, the drawing in crude. It would be ironic, but perfectly possible, if the earliest surviving Chinese rugs were aesthetically so undistinguished.

Detail image of a fragment of a coarsely woven rug attributable to the Ming period (1368-1641)

Fortunately, some documentation survives in Japan among inventories of the groups taking part in the Gion Festival, an annual event in which various textiles an other objects are taken from secure storage, assemble onto floats or other displays, and paraded through old sections of Kyoto as part of a religious festival. While demonstrating exactly when and from where these carpets were acquired is somewhat problematic, surviving records apparently place some of them in the collection between 1525 and about 1705, and, as such they are the oldest possibly Chinese knotted pile rugs for which there is any specific information as to date. They are identified as Chinese in the inventories, and a factor that makes their origin in Japan unlikely is that a number of other rugs in the same treasuries are of obvious Indian or Near Eastern origin. In this context, foreign objects appear to have had more status than those made in Japan.

At the far en of the dating spectrum is the report from San Francisco Asian art dealer that one of these pieces - strongly resembling Fig. 8, but allegedly found in Tibet – has been carbon dated roughly to the fourteenth or fifteenth century. In this case, the question arises as to whether one finds the carbon date or the inventory entries more persuasive. While the art historian ordinarily has little business passing judgement on the physics involved in the carbon 14 test, even its advocates concede that it unreliable for materials less than about three hundred years old. In this case, the inventories provide some grounds for questioning the carbon date.

Even assuming the rugs are Chinese – and they are woven with asymmetrical knots on cotton foundations like Chinese rugs – this does not tell us whether they are from the same mainland crafts center we believe produced the surviving early Chinese carpets

Figure 10. Carpet from thhe Gion Matsuri treasury, Kyoto. As with the drawing of the animals in Fig. 9, the figure here also suggests a Chinese origin.

Unanswered Questions
It is interesting to speculate as to how both groups of carpets described in this paper may be related to each other an to the eroded brown border pieces often labeled as Ninghsias. The Gion festival carpets are of nearly equal coarseness to the first group, but the colors, while similar, do not suggest a relationship. Aside from a few vague similarities in the border stripes, t hey are vastly different aesthetically.
Whether the two types described here were made in the same place, or are even related, they are obviously carpets for different purposes. The surviving fragments of the first group suggest that many of the original carpets, represented by the examples in Kyoto, are almost certainly intended for a less formal setting. Perhaps they can even be classified as folk art.

While it may be tempting to try to separate these types by time, it may be more likely that they differ primarily in place of manufacture. Indeed, we have inadequate grounds for assuming that either group discussed in this paper was woven in the same place as the 19th century rigs usually assumed to be from Ninghsia. The mention of this town as a carpet center by the Jesuits who accompanied K’ang Hsi in 1697 may be given it undue prominence, as there are stray hints from other sources of a number of carpet weavings towns. As Jenyns (1981, p.36)
notes, “Pao Lu, a border town in Northern Shansi, Chahar district, was famous for its rugs, and also Kuei Hua in Suiyuan. Ninghsia was famous fir wooden saddle cloths. Wool rugs also came from Yu Ling-fu in Shensi and Hua Tu… north of the Ordos.” There are may have been other carpet weaving centers as well.

What this allows us to consider is that both types described in this paper, along with the eroded brown border pieces, could all have been roughly contemporaneous, but from different centers. In my opinion, it is not entirely plausible that rugs of such sophisticated design as Figs. 1, 4, and 5, for example, would have been woven in such provincial area as Ninghsia, but would more likely have originated closer to the court sphere of influence. Since there is not a single shred of real evidence linking then to the Ming court, which fell in 1644, one should be cautious in their dating. Comparisons with the designs and colors on ceramics, silk textiles, another Chinese arts are notoriously hazardous, as we know that designs associated with the Ming were used in Chinese art well into the 20th century. Little has so far been published about either group of carpets discussed here, and the major questions have certainly not been answered.

Reproduced with permission by Murray Eiland.
No text or photographs may be reproduced without permission from Dr. Murray Eiland or myself.

<---previous article