But now that I have given this much deserved tribute, let me begin to quibble. Only books with great general credibility are likely to be taken by others as authoritative, and consequently only in such cases is it worth the reviewer's time to express reservations that might otherwise seem too minor to mention. Perhaps most of my dissatisfactions center around a reluctance of the author to address some of the broader issues of structure and its relationship to provenance. What I am saying is that I wish the author had delved more deeply into an exploration of who made these rugs and where.
One case in point is the large number of rugs in apparent Turkish designs he attributes to various parts of the Balkans. At least 12 of the 37 carpets of the Turkish section are thus assigned tentatively to Wallachia. A look at the technical descriptions, however, seems to reveal that these rugs do not fit together into a discrete structural category. Those in the double-arch design with cartouche borders, of a type traditionally labeled "Transylvania" (Nos. 31-33), are uniformly knotted with single strand pile yarn. Virtually all of the rugs found in Europe in this design show this structural peculiarity. Rugs in other designs attributed by Ellis to the same area, however, for the most part show two-plied pile yarns (for example, the opposed arch rugs, Nos. 27-9). Now this might appear to be a minor point, but Ellis's evidence of a European source, based upon weaker colors in some of th pieces under question, is even more vague. At the very least the whole issue deserves a more substantive discussion.
I have the same objection to the caption for rug No. 53, with a dragon rug design but apparently woven in Persia. There is no effort made to reconcile the Kurdistan label provided here with the asymmetrical knotting, a feature that has been found on no Kurdistan rugs within living memory. But my point is that 1 believe it would be more valuable to discuss the possible provenance, speculate about the reasons for such copying and explain what Z7S cotton warps might mean to the dating of such a rug rather than to tell us that "tall panels along the side axes provide opportunities for graceful tree palmettes on long, chevroned stems."
Discussion of the several truly magnificent classic Persian rugs in the Philadelphia collection is, for the most part cautiously navigated, although I believe Ellis greatly overstates the case when he says the the half-carpet with a medallion design (No. 50) "can be attributed to Tabriz with reasonable confidence." While this attribution is traditional it comes into use through the same old rug historians who have grossly misattributed virtually every other groupof classic Persian carpets. He provides no evidence for attribution.
During Ellis's discussion of the Persian carpets, I found myself regretting that in preparing this lavish catalog the additional step of dye testing was not arranged. The author's descriptions of field color often tell us little, and rugs with insect dyes are usually described as showing one variety or other of wine red, without the author's venturing a speculation as to what he thinks these dyes are. At one point however, he suggests that in the medallion and animal fragment (No. 49) "the shade of the field's red...(seems) to point to Khurassan." Now the carpet almost certainly has a lac or cochineal red, but so do all the carpets of Kerman.
Dye testing is not nearly so rare or expensive as it once was. There are commercial operations now that will do spectrophotometric dye identification for as little as $25 when a large number of determinations are ordered, and surely could have provided Ellis with a wealth of information that might have aided his attributions. He deftly alludes to the possible relationship No. 49 has with the Salting group, one possible way to make a clear distinction is dye analysis. The l9th century sultan's workshops allegedly producing the Salting group would almost certainly have used cochineal which we find throughout Anatolia at that time. Earlier classic Persian and India carpets show an almost identical red dyed with lac.
One element of the text I appreciate, which might bother some others, is Ellis's occasional use of a value judgment as to the quality of particular rugs. In describing No. 30, an opposed arch rug, he notes that "the museum's rug constitutes a pleasant variation, for, by and large, the opposed-arch rugs are a rather monotonous class." Later in describing a Kerman "shrub rug" (No. 57), he notes that the carpet, "represents a situation in which rarity must compensate for lack of inspiration in design and of good fortune in the preservation of even the limited array of charms with which this piece originally faced the eye...." Not only do I agree with these comments and others of the same sort, but I believe that in general people writing about rugs need not be so reluctant to express artistic judgments. After all, not all rugs, even 17th century ones, are great masterpieces.
Around some attributions, I am not in such agreement, however, as I question Ellis's tentative attribution of the museum's five Indo-Persian rugs to Agra. While that city almost certainly did produce some carpets during the Mughal period, there are carpets of the same type still in the Jaipur collections that have early labels identifying their source as Lahore.
Ellis's comment about silk in Chinese carpets seems similarly unsupported:
Among these Chinese rugs, there must undoubtedly have been constant imitation of the silk carpets produced in the capitals for the court, which we see reflected in ever greater diversity and bold, bright coloring in the scroll portrait paintings. Cultured Chinese obviously did not disdain carpets -- merely woolen ones.
If such were the case, where are the silk carpets used by the Chinese court? Surely Ellis cannot be referring to the early 2Oth century series of silk fakes with elaborate inscriptions, and yet he does not seem to be referring even to a scrap of surviving carpet. There is nothing about the carpets in the scroll paintings suggesting to me that the rugs must be silk. I am uneasy with such a bold, yet unsupported statement.
I am similarly uneasy with Ellis's occasional use of the condition of a rug as evidence for its age. One of the Lotto rugs (No. 6), for example, is in every respect a plausible l6th century example, but he adds, "the fresh condition of the back, the completeness of the original selvage...the survival of kilims, and the good condition of the pile...suggest that the rug is not nearly so old -- possibly instead a revival product of the l9th century...." Again in his discussion of one of the Chinese rugs, he adds, "...the excellent preservation of the side finish and of the warp loops at the lower end militate against centuries of age...."
Considering the varieties of use to which a rug might be put, I cannot set much store in such evidence. During the last 20 years, I have had many new rugs in steady use within my own home, and I can attest that even during this period, examples of about the same weave quality as a Lotto have become quite worn indeed, including the reinforced edges. Such a rug could easily be a tattered rag in 40 years, and it is safe to assume that any piece that has survived from the l6th century or even the mid-19th century has been given special attention. Of course, the rug has not been in use even a fraction of that time, but its condition is more a function of the care it has received rather than its age.
The edges of certain Chinese rugs are even less of a test of age. Since the types of rugs to which Ellis refers were made to cover kangs or platforms, the overhanging edges may have been the parts least exposed to any kind of abrasion. Indeed, in his discussion of a type of Chinese carpet in the caption to No. 72, he makes the assumption that all rugs of this type, including the squarish example he cites in Fig. 72a, are kang covers. This is not the case, as I personally saw three of these large squarish rugs in use on the large central platforms of reception halls in the Imperial City in Peking. I informed Ellis of this in a letter soon after my return from China in I977, but perhaps he did not believe me. Possibly he had already finished his part of the catalog by then and did not wish to delay its publication with further changes and amendments.
Indeed, this work has been rumored to be in preparation since I entered the rug field over 20 years ago. A number of stories have circulated as to the reasons its appearance was delayed, and several of these assert that the manuscript itself was finished and submitted for publication many years ago. I was told by one reliable source that the museum had brought in a young woman to "edit" the work, and that this had led to the kind of disagreements that can drag on for years. Another source indicated to me that the editor "made the text readable," suggesting that otherwise it was written in a style too convoluted for modern audiences. If this is truly the illusion of the powers at the Philadelphia Museum, I can only react in astonishment. The non-edited work that Charles Ellis has produced for other publications all these years has always had about it a kind of clarity and precision that I would have imagined required no such "help." I have the impression that this misguided notion delayed publication for at least five years.
But now that we have it, we must all be thankful. As I mentioned above, I have devoted substantial space to criticisms of the work and small space to praising it. But the excellent qualities of this sumptuously produced volume outweigh any trivial defects. No one else who has written about rugs has had at his disposal all of the information Ellis calls upon in his comparison of these rugs with others of similar types, nor can we expect from others the same depth of perception of the cultural milieu that produced these rugs. Altogether this is a magnificent achievement and one deserving of our deepest respect and appreciation.
