i 
 
 i 
 
 I 
 
HAND-WOVEN CARPETS 
 
 VOLUME I 
 
Digitized by the Internet Archive 
 in 2016 with funding from 
 Getty Research Institute 
 
 https://archive.org/details/handwovencarpets01kend 
 
Persian carpet, I5th-i6th century. 
 
 (See pages 19, 103.) 
 
 Victoria and Albert Museum, 
 
.'i'ifrtns? rUoi-rijji .Jsq-ss 
 
 (.joi ,Qi c'jgsq 9fi>8) 
 jji-xobrT 
 
HAND-WOVEN CARPETS 
 ORIENTAL & EUROPEAN 
 BY A. F. KENDRICK 
 
 KEEPER OF THE DEPARTMENT OF TEXTILES 
 AT THE VICTORIA AND ALBERT MUSEUM 
 
 AND C. E. C. TATTERSALL 
 
 WITH 205 PLATES, OF WHICH 19 ARE IN COLOUR 
 
 VOLUME I TEXT 
 
 NEW YORK: 
 
 CHARLES SCRIBNER'S SONS 
 
 1922 
 
 kIK 
 
 Kr-: 
 
 V. i 
 
 text 
 
PRINTED AND MADE IN GREAT BRITAIN 
 
THIS EDITION IS LIMITED TO 1000 NUMBERED 
 SETS, OF WHICH THIS IS SET.^fk5 
 
V 
 
 PREFACE 
 
 A brief explanation of the nature of the link between this volume 
 and Neugebauer and Orendi's excellent Handbuch der Orientalischen 
 Teppichkunde, published in 1909^ is due to the reader* By arrangement 
 with Messrs* Hiersemann of Leipzig many of the illustrations in that volume 
 have been made use of^ and these have been supplemented by a large 
 number of new ones in colour and half-tone* The scope of that handbook 
 did not render the text easily adaptable as a basis for the text of the present 
 volume* The necessary amplification and revision would have entirely 
 destroyed its character, and to undertake the task would have hindered 
 rather than furthered the end in view* These considerations explain 
 why the text of the present volume is entirely new* Responsibility 
 must in no case be laid, therefore, at the door of the earlier writers, except 
 in regard to a few statements, of no vital import, which could have only 
 been verified by reference to the original carpets illustrated but not now 
 accessible* The great value of Neugebauer and Orendi^s work lay in 
 the varied and admirable series of illustrations of carpets woven within 
 the last hundred years and in the descriptive notes appended* When 
 it first appeared, there was no other book of so modest compass and price 
 which aimed at classifying the bewildering variety of types of Oriental 
 carpet obtainable by those who had the money to buy, and no other book 
 has since taken its place* 
 
 The literature on the subject of the older carpets was already fairly 
 voluminous* Of all such works, that published under the auspices of 
 the Austrian Government after the great exhibition of carpets at Vienna 
 in 1891 remains the chief* The truly magnificent series of coloured 
 illustrations to that work places it beyond all chance of rivalry for many 
 years to come* But they also render it unobtainable, except occasionally 
 at a price which would represent more than a year’s income of many a 
 serious student* The text was naturally subject to the limitations of the 
 time* The day had not then come when a sound historical survey, 
 covering as wide a range as the exhibition itself, could be written* 
 
 A supplementary volume, with descriptive text by Dr* Friedrich 
 Sarre, was published in 1907* 
 
 Next followed Dr* F* R* Martin’s book. Oriental Carpets Made Before 
 
 V 
 
PREFACE 
 
 1800. This eminent Swedish traveller and writer took a fearless and 
 original line, and although the lapse of time has not since tended to 
 confirm some of his theories, a great deal was added to the common stock 
 of knowledge. Martin's work was published in 1908. 
 
 Two years later the exhibition of Muhammadan art was held at 
 Munich. A remarkable collection of carpets was brought together on 
 that occasion, and they received due attention in the fine publication, 
 Meisterwerke Muhammedanischer Kunstf produced at the close of the 
 exhibition as a record of the principal works of art shown. 
 
 All the books so far mentioned are costly and scarce. The frequent 
 references made to them in the following pages are called for by the 
 impossibility of adequately illustrating, especially in colours, a subject 
 of so wide a range in any volume of moderate si2;e and cost. The books 
 are accessible in our national libraries, and the student of the subject 
 cannot afford to miss any available assistance in consulting them. 
 
 The number of smaller and less ambitious volumes on the subject 
 of carpets is gradually growing. Few of them claim to embody much 
 original research, although in their discursive way they are not without 
 value, and they do a useful service in contributing to the available stock 
 of illustrations. They have a place in the list of useful works at the 
 end of this volume. 
 
 Dr. W. von Bode's book, Vorderasiatische Kniipfteppiche, published 
 in 1902, should be particularly mentioned, as being, within the modest 
 limits set by its author, a scholarly and indispensable book. 
 
 Work still remains to be done in clearing up obscure points, 
 especially, though not entirely, in regard to the early history of the 
 subject. 
 
 A number of the carpets illustrated in this volume are already well 
 known to students, but they are such as cannot be ignored without 
 disadvantage in a volume dealing, however briefly, with the subject as 
 a whole. 
 
 The Council of the Royal Society of Arts have kindly accorded 
 permission to adapt a paper on English carpets read before them and 
 published in the Journal of the Society (Vol. LXVII, 1919^ p» 136) for 
 the purposes of this work. 
 
 My collaborator and colleague, Mr. Tattersall, is responsible for the 
 classification of the modern carpets and the technical notes throughout. 
 Such value as the book may have is largely due to him. 
 
 My thanks are due to the following for their courtesy in allowing 
 
 vi 
 
PREFACE 
 
 the reproduction of carpets in their possession : — The Duke of Buccleuch, 
 The Duke of Northumberland, Lady Cunliffe, The Hon, Lady Hulse, 
 Lord Verulam, The Hon, H, McLaren, M,P,, The Earl of Ilchester, 
 The Mobilier National (Paris), The Louvre, The Gobelins Museum 
 (Paris), The Museum of Decorative Arts (Paris), The Vienna Museums, 
 The Berlin Museums, The Leip2;ig Museum, The Victoria and Albert 
 Museum (London), Mr, George Mounsey, Mr, Lionel Harris, The 
 Girdlers^ Company, Dr, W, von Bode, Dr, Friedrich Sarre ; and to 
 M, Albert Levy, of the Librairie Centrale des Beaux-Arts, Paris, for 
 permission to reproduce Plate 26 from his Exposition des Arts Musulmans, 
 It is an agreeable task to record my indebtedness to Mr, Victor 
 Gollancz for dispositions which have lightened the labour and added 
 to the pleasure of writing the following pages, 
 
 A, F, Kendrick, 
 
 September t 1922* 
 
 Vll 
 
CONTENTS 
 
 VOLUME L—TEXT 
 
 Page 
 
 PREFACE .............. V 
 
 LIST OF ILLUSTRATIONS IN VOLUME I . . . . . . . . xi 
 
 PART I.— HISTORICAL 
 
 I INTRODUCTORY ............. 3 
 
 II PERSIA ............... 8 
 
 III INDIA 37 
 
 IV TURKEY ............... 43 
 
 V THE CAUCASUS ............. 58 
 
 VI EASTERN ASIA— 
 
 (a) TURCOMAN TRIBES 61 
 
 (&) CHINA AND CHINESE TURKESTAN ......... 62 
 
 VII EUROPE— 
 
 (а) SPAIN 68 
 
 (б) POLAND .............. 72 
 
 (c) FINLAND 72 
 
 (d) FRANCE 73 
 
 VIII ENGLAND 75 
 
 PART II.— TECHNICAL 
 
 I THE TECHNIQUE OF HAND-WOVEN CARPETS— 
 
 (a) KNOTTING AND WEAVING 87 
 
 (b) GROUP CHARACTERISTICS OF EARLY CARPETS ...... 99 
 
 (c) MATERIALS ............. 114 
 
 (d) DIMENSIONS AND SHAPE . . . . . . . . . .116 
 
 (e) COLOUR AND DYEING ........... 118 
 
 II THE DESIGN OF CARPETS ........... 123 
 
 III PRACTICAL CONSIDERATIONS— 
 
 (a) PURCHASE 133 
 
 (b) PRICES .............. 142 
 
 (c) TREATMENT 145 
 
 IX 
 
CONTENTS 
 
 IV CARPETS IN THE MARKET— GROUPING AND IDENTIFICATION— Page 
 
 (а) TURKISH 149 
 
 (б) CAUCASIAN i6i 
 
 (c) PERSIAN .............. 172 
 
 (d) CENTRAL ASIATIC 186 
 
 BIBLIOGRAPHY 193 
 
 INDEX 1 95 
 
 VOLUME II —PLATES 
 
 LIST OF PLATES v 
 
 PLATES xiii 
 
 MAPS ............ At end of volume 
 
 X 
 
LIST OF ILLUSTRATIONS 
 
 VOLUME I 
 
 Plate I. PERSIAN CARPET, isth-i6th CENTURY Frontispiece 
 
 Plate 89. ENGLISH CARPET, DATED 1614 ..... Frontispiece to Part I 
 
 Plate 53. GHIORDES PRAYER-CARPET, LATE 17th CENTURY . Frontispiece to Part II 
 
 DIAGRAMS OF KNOTTING AND WEAVING, ETC.— 
 
 Fig. Page 
 
 A. DIAGRAM OF LOOM, ILLUSTRATING THE SEPARATION OF WARP-THREADS 
 
 FOR INSERTION OF WEFT-THREAD 90 
 
 B. DIAGRAM OF LOOM ILLUSTRATING SEPARATION OF WARP-THREADS FOR 
 
 PASSAGE OF WEFT-THREAD IN REVERSE DIRECTION . 91 
 
 C. DIAGRAM ILLUSTRATING KILIM WEAVING 92 
 
 D. DIAGRAM ILLUSTRATING SOUMAK WEAVING 93 
 
 E. DIAGRAM ILLUSTRATING THE GHIORDES KNOT 94 
 
 F. DIAGRAM ILLUSTRATING THE RIGHT-HAND SEHNA KNOT . 94 
 
 G. DIAGRAM ILLUSTRATING THE LEFT-HAND SEHNA KNOT . . . . 95 
 
 H. DIAGRAM ILLUSTRATING THE SEHNA KNOT WITH THE WARP ON TWO 
 
 LEVELS . 95 
 
 I. DIAGRAM OF KNOT FOUND ALMOST EXCLUSIVELY IN EARLY SPANISH 
 
 CARPETS . 96 
 
 J. PLAN ILLUSTRATING THE CONVENTIONAL ARRANGEMENT OF CARPETS IN 
 
 A PERSIAN ROOM 117 
 
 XI 
 
PART I 
 HISTORICAL 
 
T ',,1 ^r>v' 
 
 
 
plAte 89. 
 
 English carpet, dated 1614. (See pages 
 79. II4-) 
 
 The lion. ImUu llulse. 
 
Chapter I 
 
 INTRODUCTORY 
 
 The pile-carpet, though long ago naturali2;ed in Europe, is an alien 
 in the Western world* Its origin, in the East, is the outcome of conditions 
 very different from those which control its use in Europe, and if pile- 
 carpets are now made to serve similar purposes in their native home, 
 the reason must be sought in the spread of western habits of life* With 
 us carpets serve as an agreeable background for chairs and tables ; they 
 help to cure draughts and give the room a cheerful tone ; they soften 
 the impact of the shoe on the floor, and deaden noise* Their use is 
 warranted by these services, which the pile-carpet renders better than 
 any other floor-covering ; but they are not indispensable* The mind 
 must be disencumbered of such associations before the genesis of this 
 toilsome and ingenious type of weaving can be made clear* To produce 
 a pattern by interlacing continuous threads of different colours is natural 
 enough, and there is general agreement that the simple process of weaving 
 is to be reckoned among the earliest artistic efforts of prehistoric mankind* 
 But to set the threads in a vast number of short lengths up on end, and 
 to pack them so tight together that they keep that position, entails so 
 much toil, and uses up so much material, that the contrivance can only 
 be regarded as a response to conditions not common to all men* There 
 can be little doubt that a wandering shepherd -life, which involved the 
 need and at the same time provided the means of supplying it, gives 
 the right explanation* We think of some nomad tribe of shepherds, 
 following their flocks from one pasture to another, with no roof or floor 
 but such as they could carry with them from place to place* They needed 
 a tent over their heads, and at times they needed even more a dry and 
 warm covering for the ground of their temporary abode* 
 
 The wool of the flocks — ^whether sheep or goats — or even of the 
 camel and yak — offered an ideal material for weaving into a compact and 
 durable texture* The invention did not merely provide an agreeable 
 addition to the household effects ; it met an elementary need and became 
 indispensable* 
 
 From these nomad surroundings the carpet found its way into the 
 
 3 
 
HAND-WOVEN CARPETS 
 
 dwellings of agricultural neighbours^ and its adoption by all classes of 
 people came about in due course. 
 
 In the West, it is now completely domesticated, and consequently 
 we find it hard to realize that the pile-carpet was practically unknown 
 among us before the fourteenth century, very rarely seen before the 
 fifteenth, and not in general use before the eighteenth. The date of 
 its advent may be approximately traced. Its strange patterns and bright 
 colours excited wonder and interest, and painters everywhere made use 
 of its decorative qualities.^ Centuries were yet to pass before it 
 superseded the layer of rushes or the plaited rush matting in general 
 use, and still common in great houses in England as late as the days of 
 the Stuarts. 
 
 As the Oriental carpet became known and used over a wide area, the 
 obvious simplicity of the knotting process caused attempts to be made 
 here and there to meet demands by local production, and at last in this 
 way a craft evolved by nomad shepherds in the heart of Asia spread 
 thence over the civilized world. The machine-made ** Axminster of 
 modern times, with a daily output amounting to thousands of square 
 yards in Great Britain alone, is a direct successor in one line of descent. 
 
 It is only in general terms that the origin of carpet-knotting can 
 be discussed. The clue to the actual locality, and even to the approximate 
 date, of the invention is lost beyond prospect of recovery. It has been 
 suggested that pile-carpets were made as early as five thousand years 
 before Christ.^ 
 
 Carpets of some sort must have been used in very primitive times. 
 Homer refers to them by a name which they still retain over a large part 
 of Europe ; but the unchanged name must not be taken to signify that 
 the nature of the fabric to which it is applied has not changed. While 
 it is just possible that pile-carpets were known to the Asiatic Greeks 
 of Homeric days, it would be rash to base any argument on such an 
 assumption. 
 
 The designs most favoured by the ancient Greeks for the textile 
 ornamentation of their houses were hunting scenes and subjects from 
 mythology and history. The carpet-kiiotting method would be tedious 
 and cumbersome for such representations. It is true that such subjects 
 
 ^ Dante achieves an effect no less vivid by similar means. In describing the bright 
 markings on the back of the monster Geryon he says, “ Never did Tartars or Turks make 
 cloth with more colours in ground or pattern ” {Inf. xvii). 
 
 2 The late Sir G. Birdwood, in Oriental Carpets, Vienna, 1892. 
 
 4 
 
INTRODUCTORY 
 
 are to be found in the days of the highest development of the industry 
 in Persia, but they tax the utmost skill of the carpet-knotter, and they 
 would not have been deliberately chosen under the primitive conditions of 
 Homeric life. As records of the Mesopotamian civili2;ations we have the 
 marble slabs from Nineveh. Those showing floor-coverings have patterns 
 of lotus-flowers, rosettes and diapers. Nothing is left to show how they 
 were made, but the tapestry method is more likely to have been used 
 than any other.^ 
 
 The skins of beasts, or some kind of pileless woven stuff, would 
 serve in a primitive community as a floor-covering for special occasions, 
 and those who could not come by such luxuries, or whose habits of life 
 did not admit of their use, found a sprinkling of rushes or leaves a passable 
 substitute. 
 
 It is indubitable, though surprising, that somehow the Oriental 
 method of pile-knotting in wool became known in Europe by the end 
 of the twelfth century of our era. A happy chance has preserved to the 
 present day some portions of a panel made then at Quedlinburg, in the 
 Har^-Mountain region, by this method, and now preserved in the 
 Schlosskirche there. This work, which measured about 24 ft. high by 
 20 ft. wide, represents rows of scenes illustrating Martianus Cappella's 
 ** Marriage of Mercury and Philology.'' It is obvious that it was not 
 intended to be spread on the floor, although it was found in use on the 
 floor of the Abbess' Stall in the church early in the last century. The 
 nuns by whom the work was done must have had access to a piece of 
 carpet-knotting, probably without realizing that it was made for a floor- 
 covering. The method of knotting is that used in Turkey carpets, but 
 the need of fine lines for the figures was met by devising a single-warp 
 knot, unknown in the East, for those parts. ^ 
 
 The possibility that the nuns of Quedlinburg made an independent 
 discovery of this knotting process is so very slight as to be negligible. 
 The hanging is known to have been made in the time of the Abbess Agnes 
 (1186-1203) and it affords a unique piece of evidence of the actual 
 importation of pile-carpets into Europe before the end of the twelfth 
 
 ^ See A. H. Layard, Second Series of The Monuments of Nineveh, London, 1853, 
 Plate 56. 
 
 2 J. Lessing and M. Creutz, W andteppiche und Decicen des Mittelalters in Deutschland, 
 Berlin, Pis. 1-7. The process continued to be used for wall-hangings in Germany until 
 the sixteenth century, although for such a purpose it is wasteful both of time and material. 
 Its use for small details to relieve tapestry-work, as seen sometimes in mediaeval German 
 hangings, is effective. 
 
 5 
 
HAND-WOVEN CARPETS 
 
 century* There is nothing else to show that they were brought in earlier 
 than the fifteenth century, except representations in paintings ; and 
 even those are not conclusive* Carpets seen in pictures of the fourteenth 
 century might equally well have been woven by the more primitive 
 tapestry process* A strange obscurity hangs over the origins of carpet- 
 knotting everywhere* 
 
 In considering the traces left in the East itself we are much in the 
 dark* There are no pile-carpets existing which are demonstrably earlier 
 than the Quedlinburg hanging, except the fragments lately found in the 
 desert sites of Central Asia* Those regions must be very much nearer 
 the original home of carpet-knotting than Quedlinburg, and anything 
 of the kind discovered there has a peculiar interest* Nothing but small 
 fragments have been found up to the present* Von le Coq's expedition 
 brought to light at Qyzil a red piece with portion of a pattern in yellow 
 outlined in brownish black, but there is not enough to reveal the nature 
 of the pattern*^ Sir Aurel Stein's excavations afford no better clue*^ 
 These fragments are obviously worthy of the closest examination, as 
 presumably they were all made well within the limits of the first 
 millennium ; their significance cannot properly be measured until further 
 investigation and research are made possible* Leaving them out of 
 account, the oldest existing Oriental carpets appear to be those in the 
 mosque of Ala-ed-Din at Konia, first brought to notice by Dr* F* R* 
 Martin,^ and attributed by that eminent authority to the early years of 
 the thirteenth century* Towards the end of that century, the Venetian 
 traveller, Marco Polo, wrote that the finest and handsomest carpets 
 in the world " were made in Turkomania (i.e* Asia Minor ^)* 
 
 The conditions under which pile-carpets may have been first devised 
 have already been touched upon* Where the invention took place is still 
 an open question* Sir George Birdwood ^ is inclined to favour Egypt as 
 the home of a very early civilisation* But it is a real question whether 
 civilisation had much to do with the discovery* Moreover, there are 
 serious objections to the claim of Egypt* The soil and climate of the 
 country render carpets a luxury rather than a necessity ; wool, the natural 
 
 ^ F. Sarre and T. Falkenberg in Berliner Museen, Berichte, XLII, 1921, p. no. 
 
 2 Sir M. A. Stein, Desert Cathay, London, 1912, Fig. 116, 4 ; Serindia, 1922, Vol. IV, 
 
 PI. 37. 
 
 ^ Oriental Carpets, 1908, PI. 30. See post, p. 45. 
 
 ^ The Booh of Ser Marco Polo, translated and edited by Colonel Sir Henry Yule, 3rd 
 ed., London, 1903, Vol. I, p. 43. 
 
 ^ Vienna, 0 .( 7 ., 1892. 
 
 6 
 
INTRODUCTORY 
 
 material for making them, was not used by the ancient Egyptians ; and 
 the wealth of textile material hitherto turned up by excavators in the 
 burying-grounds, varied as it is, includes no example of carpet-knotting. 
 Arguments of a similar nature apply to India, China has no better case. 
 The native textile materials of these three countries are, respectively, 
 linen, cotton, and silk. We shall do better to look for the ideal home 
 of woollen pile-carpets in uplands where the cold is sharp at times, where 
 a nomad people require a portable floor and keep flocks for the provision 
 of the wool. On these grounds, the great plateau of Persia, the highlands 
 of Anatolia or the plains of Central Asia have the best claim. Carpet- 
 weaving reached its highest limits of attainment, and is known to have 
 been practised from early times, both in Persia and Anatolia, The task 
 of balancing the claims of these two regions to priority might be endlessly 
 pursued without getting any nearer to a solution. There is more hope 
 of further developments in regard to Central Asia, where so much research 
 is going on, and the near future may throw a flood of light on this question, 
 as well as on others equally obscure. 
 
 7 
 
Chapter II 
 
 PERSIA 
 
 In any systematic record of pile-carpet weaving it is advisable to 
 begin with Persia. To do so is convenient^ and almost inevitable. It 
 is true that the Persian carpets were not the first to become known in 
 Europe. All available evidence seems to show that they were not known 
 at all for some centuries after pile- carpets were first brought in from 
 the East. Of all the carpets represented in European pictures of the 
 fourteenth, fifteenth and sixteenth centuries, not a single one is Persian. 
 The configuration of Persia and its place on the map were obstacles to 
 the transport of bulky wares which long continued to be almost completely 
 effective. A few carpets from Persia found their way to Europe in past 
 centuries. Here and there a Persian carpet in a church or palace in 
 Italy may have been imported straight from the loom, just like the Anatolian. 
 The carpet shown in Plate 5A belonged formerly to Signor Stefano 
 Bardini of Florence. Others also from the same collection had probably 
 been long in Italy ^ ; but had there been many of them in that 
 country before the end of the sixteenth century they would assuredly 
 have been copied by the painters. A woollen carpet with inscrip- 
 tions and arabesques in silver was in a Spanish cathedral until a 
 few years ago.^ A very remarkable Persian carpet from Mantes Church 
 was acquired for the Louvre in 1912, where it is exhibited in the 
 large Oriental gallery.® It has a pattern of Chinese dragons, phoenixes 
 and unicorns ** amid cypresses and fruit trees, on a blue middle ground 
 and a red border. This carpet was probably made in the last years of 
 the sixteenth century. A fragment of the border of another fine Persian 
 carpet of about the same date, with a pattern of interlaced arabesques, 
 formerly in the Cathedral of Troyes, is now in the Victoria and Albert 
 Museum.^ A carpet in the Cathedral of Cracow will be referred to later. 
 The famous hunting-carpet in the Austrian imperial collections (Plate 10) 
 
 ^ See {e.g.) Martin, 0 .( 7 ., Figs. 6i, 109, 127. 
 
 2 Illustrated in colours, Ancient Oriental Carpets (supplement to the great book published 
 at Vienna), PI. II ; see also Martin, 0 .( 7 ., Fig. 131. 
 
 ® Illustrated in colours in G. Migeon, L’Orient Musvlman, Paris, 1922, PI. 36. 
 
 Illustrated in colours, Martin, O.O., Fig. 108. 
 
 8 
 
PERSIA 
 
 is supposed to have belonged at one time to the Czars of Russia. The 
 neighbourhood of Russia to the north-western provinces of Persia accounts 
 for the number of fine Persian textiles in that country. Probably it was 
 in North-West Persia that these carpets which have long been in Europe 
 were made. They would have been conveyed to the nearest port on the 
 Black Sea^ and thence by water to Constantinople (whither multitudes 
 of Persian carpets have since been carried) ; or perhaps in some instance 
 by direct passage in trading ships of the countries concerned. The 
 carpet in the Salting Collection (Plate 14) with an ode by the Persian poet 
 Hafiz worked into the border in silver thread is supposed to have been 
 sent from the loom to the Turkish sultan's palace at Constantinople ; 
 from thence it passed into Mr. Salting's possession not many years ago. 
 The celebrated silk and gold carpets were exceptional, as they were borne 
 by special embassies from the Shah to the European Courts (see p. 31). 
 
 The British Muscovy Company of Merchant Adventurers found 
 their way through Russia to Persia before the end of Queen Elizabeth's 
 reign, but it was a troublesome and interminable land journey from Persia 
 to the White Sea or the Baltic, and in the meanwhile the harbours of 
 Asia Minor afforded easy access to the carpet-producing districts of 
 Anatolia. 
 
 The consequence was that most of the carpets imported were from 
 Anatolia, and the term Turkey-carpet " bid fair to make good its claim 
 to be applied to all pile- carpets.^ The subsequent fame of Persian carpets 
 brought about a reaction, and to-day the appellation as often as not is 
 inexcusable, unless it be conceded that the name of the country whence 
 the best pile-carpets have come may be legitimately applied to any carpets 
 of the kind. How far back the making of pile-carpets goes in Persia we 
 do not know, and probably we never shall. The oldest existing carpets 
 to which dates can be assigned with confidence are not Persian ; but that 
 is probably due to accidental circumstances. It is safe to assume that 
 carpet-knotting was either invented in Persia or came in with nomad 
 tribes from the North at an early date in the history of the craft. Whether 
 that happened two thousand years ago, two thousand five hundred years 
 ago, or even fifteen hundred years ago we cannot say. 
 
 A description of a sumptuous carpet of very early date has long 
 
 ^ The French traveller Jean Chardin, who visited Persia in the second half of the 
 seventeenth century, explains that Persian carpets were still included in this general term 
 in his day, because they came by way of Turkey before the ocean route was explored 
 (J. Chardin, Voyage en Perse). 
 
 9 
 
HAND-WOVEN CARPETS 
 
 survived the carpet itself^ as well as all others of its time^ in the annals of 
 the Arabian writers who recorded the conquest of Persia. 
 
 When Ctesiphon^ the Sassanian capital near the site of the later city 
 of Baghdad^ fell into the hands of the Arabs in A.D. 637^ the carpet was 
 found in the famous White Palace. 
 
 The late Dr. v. Karabacek^ of the Vienna Royal Library^ extracted 
 from the Arabian manuscript the following account of the carpet.^ It 
 was originally made for Chosroes I (a.d. 531-579) and his successors 
 used it ever after until the last Sassanian king Jazdegerd, but only during 
 the stormy and rough winter seasons when it was impossible to stay in 
 the garden. On these occasions the drinking feasts which usually took 
 place outside were transferred to the carpet^ for its pattern represented 
 a garden in the bloom of spring. The carpet was called The Winter 
 Carpet or '' The Spring of Chosroes.'' The materials of this carpet 
 were notable and rich : silk^ gold, silver, and semi-precious and precious 
 stones. The middle space of the carpet represented a pleasure-garden 
 planted with trees and spring flowers, and intersected by brooks and 
 pathways. The broad surrounding border represented magnificent 
 flower-beds, and the blossoms were shown by blue, red, yellow, white 
 and green stones. The yellow colour of the earth was represented in 
 the background of the carpet with gold ; the banks of the streams were 
 rendered by stripes, between which stones clear as crystal made the 
 representation of the water so perfect as to deceive the eye. The pathways 
 were indicated by stones as large as pearls, stems and branches were 
 made out of gold and silver, the leaves of the trees and flowers as well 
 as all the plants were made out of silk, and the fruit out of coloured 
 stones. 
 
 Though it would be expecting too much of us to ask that we should 
 accept this description as a bare statement of facts, especially in view 
 of the lively fancies of the Oriental imagination, it would nevertheless 
 be a mistake to treat the description of this carpet as being a pure fable. 
 To judge from the nature of the materials and the size (it was said to 
 be 60 ells square), it cannot have been of knotted pile. Rather must it 
 be classed with the elaborate embroidered carpets such as have been 
 made elsewhere in the East down to modern times. For all that, account 
 should be taken of it, since the description of the pattern is one that might 
 be used with little modification for carpets made in Persia more than a 
 thousand years later. No such carpet is shown in any of the Sassanian 
 
 ^ Pers. Nadelmalerei Susandschird, Leipzig, 1881. 
 
 10 
 
PERSIA 
 
 reliefs or silver-ware now existing, but the veracity of the story is attested 
 by several existing carpets of far later date*^ Some years ago a carpet 
 made a thousand years after the one described by Karabacek was dis- 
 covered in a Styrian hostelry ; it has since passed into the possession of 
 Dr* A. Figdor in Vienna. The design obviously represents the flower- 
 beds of a garden through which runs a maze of water-channels opening 
 into small lakes, much in the manner of Chosroes' carpet. The water 
 is represented in silver thread, and fish are seen, with water-fowl swallowing 
 them. There are flower-beds, and the streams are bordered by trees 
 with birds in the branches. The carpet is knotted in wool, with gold 
 and silver threads for many details. 
 
 The variations, which bear the character of a much later epoch, 
 cannot efface the impression that here is a striking confirmation of the 
 old description ; in fact, one may almost suppose these later carpets to 
 be conscious imitations. The design is peculiarly suitable for a carpet, 
 and it is one which would naturally be evolved in a country where flower- 
 gardens have always been so intimately associated with daily life. Conse- 
 quently there is no occasion to assume that the theme of Chosroes" carpet 
 was singular in Persia at the time. A few more such carpets are known, 
 but there is nothing to bridge the gap of more than a thousand years between 
 the time of the Sassanian ruler Chosroes, when the first was made, and 
 that of Shah Abbas the Great, in whose reign Dr. Figdor's carpet was 
 most probably woven. 
 
 Two garden- carpets are in the possession of the Hon. H. D. McLaren. 
 One of them, made perhaps a little later than the carpet just described, 
 is the largest existing of its kind. It measures 31 ft. by 12 ft. 6 in., and 
 it seems to have been even larger originally. The water is rendered by 
 a clever arrangement of zigzag lines producing a shimmering effect 
 when seen foreshortened. This is the usual convention where metal 
 threads are not employed. There is a square pond in the middle with 
 a central island and four swimming birds. A broad water- channel runs 
 off from this in four directions as far as the borders of the carpet. Minor 
 channels, interrupted at intervals by circular mounds, divide up the 
 rest of the space. The water is everywhere edged with a strip of soil 
 represented in dark blue, upon which trees and flowering plants grow, 
 with birds in the branches. Square and star-shaped beds with blossoming 
 
 ^ A Sassanian silver dish in the Stroganov Collection shows a king seated on a mat of 
 some kind with a free floral pattern and a wavy stem border (A. Riegl, Ein Or. Teppich, 
 p. 16). 
 
 II 
 
HAND-WOVEN CARPETS 
 
 trees and plants fill the intervening spaces » The narrow floral border 
 all round is on a white ground.^ Mr, McLaren's smaller carpet is of 
 the same type, but less complete (Plate 2). Another carpet of the kind, 
 but more diversified, with ducks and fish in the water, and hunting-animals 
 and birds on the land, is here illustrated (Plate 3). This fine carpet passed 
 some years ago from Constantinople to America by way of Berlin, but it 
 has since been brought back to Europe, Mr, Carl Robert Lamm of Naesby 
 House, near Stockholm, the possessor of some fine carpets (several of 
 which will be mentioned later), has two woollen carpets of the garden ** 
 type. Both are relatively late examples — perhaps not made before the 
 eighteenth century. The finer of the two, measuring about 10 ft, by 
 6 ft,, has a large pond in the middle from which branch four broad water- 
 channels, all showing fishes and water-plants* At the sides are four 
 square ponds. The zigz^ig lines for the water are in red, green, blue 
 and white, giving a rippling effect of colour. Broad deep blue borders, 
 varied with flowering plants, edge the water. More formal arrangements 
 of flowers, on a red ground, fill the remaining space, Mr, Lamm's other 
 carpet is not complete ; it shows a stream running throughout the length, 
 while the ground on either side is divided into squares with formal floral 
 patterns in fairly light colours,^ It draws near in design to the carpet 
 reproduced in Plate 4, but the latter shows a further modification of 
 the earlier motives. The large rectangular space in the middle with the 
 flowering plant replaces the pond, unless indeed the plain dark background 
 is meant for water. The small squares covering the rest of the space, 
 each with its plant, shows that a garden is depicted, though without 
 the almost indispensable water-channels. 
 
 This carpet brings us to the point where the traditional garden " 
 design dies away into the general stock of decorative motives. But in 
 order to grasp the basic idea of Persian carpet designs it must be borne 
 in mind that most of them have some relation, more or less direct, to 
 a garden or park. Inevitable modifications, brought about by centuries 
 of repetition, may have reduced the scheme to a formula, but even under 
 its disguises the origin is often unmistakable. 
 
 The designer of Chosroes' carpet aimed at bringing indoors for 
 the winter-time the outdoor setting of a Persian garden in spring. Let 
 this conception be widened to bring within its compass the sports in 
 
 ^ Reproduced in colours in W. A. Hawley, Oriental Rugs, New York, 1913, Colour- 
 plate VI. 
 
 2 These two carpets are reproduced by Martin, O.C., PI. 24 (in colours) ; Fig. 104. 
 
 12 
 
PERSIA 
 
 the open country, with the huntsmen and animals, the trees, birds and 
 flowers, and let allowance be made for the conventional perspective proper 
 to Eastern art, and the majority of Persian carpet patterns are made clear* 
 Brief reference to the varied guises these designs assume will be made 
 later. Some other features foreign to the soil of Persia, but appearing 
 there at an early time, must first be accounted for. In the art of Western 
 Asia, and even of Europe, during the last thousand years there is abundant 
 evidence of the subtle and penetrating influence of Chinese design. The 
 outward forms in which the singular art of the ancient and exclusive 
 civilization of China found expression — the dragons, phoenixes, con- 
 ventional clouds and waves, symbolic objects, and the peculiar types 
 of architecture and costume — seem to have acted like a spell on the crafts- 
 men who met with them, although there is nothing to show that a moment's 
 thought was given to the complex philosophy which called them into 
 being. It is natural that Manchurians, Mongolians, Coreans, Tibetans 
 and people of more or less kindred stock to the Chinese, though of ruder 
 civilization, should make use of Chinese wares and copy their ornamentation 
 so far as they were able. But the effect of Chinese art upon a people 
 of totally different race, like the Persians, with a great artistic tradition of 
 their own, is not so easy to understand. Yet it is there. No more potent 
 external influence is to be found in the carpet designs of Persia and Western 
 Asia than the Chinese. The question when it first made its appearance 
 is involved in another, that of the date of the earliest existing Oriental 
 carpets. Two examples reproduced in this volume may be instanced. 
 The first (Plate 37A), a carpet of Asia Minor, has a design obviously 
 representing the Chinese dragon and phoenix. The carpet will be dis- 
 cussed in a later chapter (p. 46) ; it need only be pointed out at the moment 
 that, whatever the actual date of this carpet may be, the motives repre- 
 sented on it had certainly found their way to Asia Minor before the fifteenth 
 century, the period to which the carpet has been assigned. The other 
 carpet (Plate 5A) is Persian. It belongs to a group to which an even 
 earlier origin has been ascribed. The intention of the craftsman to 
 represent the Chinese dragon is again beyond dispute. The archaistic 
 rendering has led an eminent authority. Dr. F. R. Martin, to attribute 
 the carpets belonging to this group to the thirteenth century, a period 
 when Chinese motives are already quite frequently seen on the lustred 
 pottery of Persia. Even this is not the first appearance of such motives 
 in Western Asia ; it is rather due to the fresh impetus given by the irruption 
 of Hulagu the Mongol, who caused Chinese craftsmen to follow the 
 
 13 
 
HAND-WOVEN CARPETS 
 
 track of his conquering armies. He is said to have transferred a hundred 
 families of Chinese craftsmen to Persia about the year 1256. It is per- 
 fectly safe to assume that Chinese motives were to be found on the Persian 
 carpets at that time^ but it may be questioned whether any existing carpet 
 is so old. The complexity of the subject requires a cautious student 
 to be on his guard, for such motives persisted for at least four centuries 
 after, and the relatively coarse texture of such a carpet as that reproduced 
 in Plate 5A might account for the angular rendering, and in some degree 
 for the archaic appearance. A very similar carpet with the same motives 
 on a brilliant red ground, belonging to Mr. Lamm at Naesby House 
 in Sweden, is attributed by Dr. Martin to the middle of the thirteenth 
 century. Another carpet, with leopards attacking various animals among 
 trees, rendered in the same primitive angular manner, he assigns to the 
 end of the century. A third, with large angular palmettes and floral 
 patterns, is ascribed to about the year 1320 ; and a carpet in the Kaiser- 
 Friedrich Museum at Berlin, with floral ornament within formal com- 
 partments, to the middle of the fourteenth century.^ Thus several 
 distinct Persian types are traced back to a remote period, the evidence 
 adduced being that of lustred pottery and dated Persian miniature paint- 
 ings. The eminent Swedish authority has done a great service to students 
 by investigating and recording these parallels, which are perfectly 
 just. But the example of the garden-carpet of Chosroes is a warning 
 of the length of time that pattern motives linger in the East with little 
 essential change. The risk of error is particularly to be guarded against 
 in the case of pile- carpets. Wherever the fineness of the texture is reduced, 
 and this may take place at any period, and at any place, an angular treat- 
 ment is inevitable, bringing about a deceptive appearance of archaism. 
 
 Similar problems have faced those who have attempted to assign 
 dates to the carpets of Asia Minor by comparison with representations 
 in European pictures, and the consequences have been the same, although 
 each line of inquiry is in the right direction, and most valuable in its 
 results.^ 
 
 As an illustration of the difficulties met with, the carpets reproduced 
 in Plates 5 and 6 may be compared. The first belongs to the group 
 attributed to the thirteenth century. It came from an Italian church. 
 
 Oriental Carpets, PI. 28, Figs. 58, 59 and 64. 
 
 ^ See Dr. von Bode’s important essay on Decorative Animal Figures in old Oriental 
 Carpets in the great publication Oriental Carpets, Vienna, 1892-4, p. i. Also J. Lessing, 
 Alt Orientalische Teppichmuster, Berlin, 1877. 
 
 14 
 
PERSIA 
 
 whither it probably went as a new carpet ; but that circumstance affords 
 little help in dating it* The other (Plate 6) is similar to a carpet exhibited 
 in 1910 at the Munich Exhibition^ and attributed there, no doubt rightly, 
 to the sixteenth or seventeenth century*^ In each, the field is broken 
 up into irregular lozenge shaped spaces by means of broad multi-coloured 
 bands, intended at least in part, and perhaps entirely, for long leaves 
 overlapping one another, and relieved with slender floral stems super- 
 imposed.2 The enclosed spaces are filled with dragons or palmettes, 
 and other palmettes are placed at some of the points where the bands 
 join. There is a narrow border of angular floral forms on a white ground. 
 This description serves equally well for both carpets, and it is almost 
 a question which is the older of the two. A greater multiplicity of detail, 
 with a certain waywardness of treatment in Plate 5A, singles that out as 
 the earlier, although there is little likelihood that it was made much before 
 the sixteenth century.^ 
 
 These carpets were probably made in N.W. Persia or Armenia. 
 The Armenians, of kindred stock to the Persians, were skilful carpet- 
 weavers. One argument in favour of this attribution is the existence 
 of such examples in Italy The Armenian region was conveniently 
 situated for the transit, having access to the Black Sea, while the difficulties 
 of transport from the heart of Persia excluded bulky goods of that country 
 almost entirely from the Mediterranean trade. Another point to be ob- 
 served is that the prominence of Chinese motives excludes the south 
 of Persia, where such motives hardly penetrated at all. Greater help 
 still in placing these carpets is afforded by a remarkable example which 
 found its way to London in 1899 (Plate 7). This carpet resembles in 
 texture and colour, and to some extent in pattern, that illustrated in 
 Plate 6. The chief difference is that there are no dragons. The spaces are 
 larger, and the middle paimette is replaced by a radiating floral device. 
 Its special significance lies in the Armenian inscription knotted in at the 
 upper end. It reads thus : ** I, Gohar, full of sin and feeble in soul, 
 have knotted this with my own hands. May he who reads pray for my 
 soul.’' Then follows the Armenian date 1129, corresponding to a.d. 1679. 
 
 ^ Munich, Meisterwerke, PI, 65. 
 
 2 The use of leaves or leaf-shaped panels as a background to hold floral forms is a 
 common feature of Persian textile design in the seventeenth century. 
 
 ^ Dr. F. Sarre [Ancient Oriental Carpets, Leipzig, 1906-8, p. i) considers the fifteenth 
 century the earliest possible date. 
 
 ^ Besides the example already mentioned Dr. von Bode obtained one from a church 
 near Venice (W. Bode, V orderasiatische Kniipfteppiche, Leipzig, ist ed., p. no). 
 
 15 
 
HAND-WOVEN CARPETS 
 
 The route this carpet took to the West late in the nineteenth century- 
 may be that by which such carpets in Italy came long before. It was 
 taken by Turks from an Armenian church and was carried through the 
 province of Batum to the Black Sea, whence it found its way to London.^ 
 
 The first carpet of this class to attract general attention was shown 
 at the great exhibition of Oriental carpets held at Vienna in 1891.^ It 
 had been obtained from a mosque in Damascus, and was then the property 
 of Herr Theodor Graf, but it has since been acquired by the Kaiser- Fried- 
 rich Museum at Berlin. Both Chinese dragons and unicorns (cfii-lin) 
 are represented ; an effect of considerable age is given to this carpet 
 by the unusually subdued colouring. A few years ago the Victoria and 
 Albert Museum acquired the example (already mentioned), with dragons 
 and ** unicorns,'" remarkable for its black ground which gives full salience 
 to the archaic forms. The dividing bands are in pale blue and yellow, 
 and the narrow formal border is white (Plate 8). Another example in 
 the same collection, less archaic in design, has dragons on a red 
 ground (Plate 5B). This carpet is rather later in date, and obviously 
 not earlier than the seventeenth century. Agreement is more universal 
 in regard to the early origin of another type of design. There is generally 
 a large central panel extending practically across the whole width of 
 the carpet, and broken up by different-coloured grounds ; smaller panels 
 extend to the right and left. The ornament throughout consists of slender 
 floral stems, running into spirals, interlaced with arabesques. The 
 restraint and severity of the design point to a date before that of the Ardabil 
 carpet, and probably within the limits of the fifteenth century, though 
 features of the type are retained in later carpets. There is an example 
 in the Victoria and Albert Museum (Plate 9).^! 
 
 Before reverting to the carpets with the simpler patterns of trees 
 and flowers, some other more complex types may be mentioned. One 
 is the hunting " carpet, which first appears in the sixteenth century. 
 The best example, and one of the most celebrated of all carpets, is in the 
 Imperial Austrian collections at Schonbrunn (Plate 10). It is woven in 
 silk and gold and silver thread. The huntsmen, mounted and armed 
 with spears, swords and arrows, attack lions, leopards, wolves, boars, 
 deer, antelopes, wild asses, jackals and hares. The ground of the carpet 
 
 ^ The Victoria and Albert Museum was not able to do more than take a photograph 
 of this carpet before it left the country again. 
 
 2 Vienna, O.C., 1892-4, PI. 36. 
 
 ^ Others are reproduced, in colours in Martin, 0 .( 7 ., PI. 2, and in Ancient Oriental 
 Carpets, Leipzig, 1906-8, PI. 4. 
 
 16 
 
PERSIA 
 
 is salmon-pink. A large eight-lobed green compartment in the middle 
 has great gold dragons and phoenixes^ following Chinese models more closely 
 than those on the carpets hitherto described. The deep crimson border 
 has a succession of winged figures repeating a scene in which a kneeling 
 figure offers a bowl of fruit to another^ who is seated and wears a crown. 
 Such winged personages, often engaged in the ordinary occupations of 
 humanity, will be met with again. Each of the corners of the carpet 
 shows a fourth part of the central design. The heads of men in the 
 palmettes of the outer green border, and of lions in those of the inner 
 white border, will be noticed. The miscellaneous nature of the quarry 
 and the confused haste of the riders galloping pell-mell in all directions 
 give a pantomimic air to the scene, but the picture is true to the traditional 
 way of hunting in Persia, as well as in India and Central Asia,^ still followed, 
 at the time this carpet was woven. All the animals in the neighbourhood 
 were driven into a prepared enclosure, which the hunting-party entered 
 afterwards, to engage in an indiscriminate massacre. 
 
 The elaborate dress of the figures on the Schonbrunn carpet suggests 
 a royal hunting-party. The carpet was probably woven in the time of 
 Shah Tahmasp (i 524-1576) There is an old tradition that it was received 
 by the Austrian Emperor as a present from Peter the Great of Russia. 
 Another carpet no less celebrated, though of a different type, is dated 
 within the period of Tahmasp's reign. This is the great carpet from 
 the mosque at Ardabil, a small town of the province of Azerbaijan, in 
 the north-west of Persia (Plate ii). The reason why such a work of 
 art should have come from a town of so little note is made clear by the 
 circumstances of the time. Tahmasp was the second king of the Sefavi 
 or Safidian dynasty — the first line of native rulers since the overthow of 
 the Sassanian dynasty by the Arabs more than eight centuries before. 
 The eponymous ancestor of the new dynasty was Sheikh Safi-ed-Din 
 who, dying in 1334, was buried at Ardabil. There also Shah Ismail, 
 father of Tahmasp, was buried. The veneration in which the tombs 
 were held, and the association with the ruling house, raised Ardabil to 
 
 ^ The cheetah or hunting leopard is not seen in this example, though it is found in 
 others. The animal was employed in Persia from an early date, and was still used when 
 these hunting-carpets were woven. 
 
 2 Dr. F. R. Martin suggests that Tahmasp ’s Court painter, Sultan Muhammad, may 
 have designed the carpet (F. R. Martin, Miniature Painting, p. 117). It is illustrated in 
 colours in Vienna, Oriental Carpets, Pis. 81, 86-89. See also Munich, Meisterwerke, Pis. 
 42, 43. A carpet of somewhat similar design, in the collection of Baron Maurice Rothschild, 
 is illustrated by Bode [V orderasiatische Kniipfteppiche, 2nd ed., Fig. 3). It was obtained 
 from the Marchese Torrigiani in Italy. 
 
 17 
 
 0 
 
HAND-WOVEN CARPETS 
 
 the dignity of the religious and national capital of Persia ; its prestige lasted 
 until Shah Abbas rebuilt Ispahan. A carpet of this size and importance 
 can only have been made to the order of Tahmasp himself for the mosque 
 where his ancestors were buried. The date upon it (A.H.946 = a.d. 1540) 
 is sixteen years after the death of his father, and most of that time would 
 have been taken up in making it and its fellow. A cartouche at one end 
 of the carpet contains a woven inscription which has been translated 
 as follows : 
 
 “ I have no refuge in the world other than thy threshold. 
 
 There is no place of protection for my head other than this door. 
 
 The work of the slave of the threshold, Maqsud of Kashan, in the year 
 946.” 
 
 The first two lines are the beginning of an ode by the Persian poet 
 Hafiz (d. 1389). It is most likely that the carpet was made at Ardabil; 
 the Court would naturally wish to watch its progress, and it would have 
 been as easy for the weaver to go there to make it, as for a carpet of this size 
 (34 ft. 6 in. by 17 ft. 6 in.) to be transported from Kashan when made. 
 It remained in the mosque until shortly before its acquisition by the 
 Victoria and Albert Museum in 1893. Fifty years earlier it was seen by 
 an English traveller, W. R. Holmes, who visited Ardabil in 1843. He 
 wrote of it as follows : On the floor (of the long lofty ante-chamber to the 
 
 principal tombs) were the faded remains of what was once a very splendid 
 carpet, the manufacture of which very much surpassed that of the present 
 day. At one extremity was woven the date of its make, some three hun- 
 dred years ago.'' ^ The disparaging note of its condition, if ever warranted, 
 is no longer applicable. The carpet has been skilfully repaired, parts 
 of another carpet of similar design, removed from the mosque at the 
 same time, having been sacrificed for the purpose.^ 
 
 Some idea of the design of the carpet will be gained from the illus- 
 tration. The pulsating blue ground is covered with thousands of flowers. 
 The great circular panel of lobed outline in the middle is yellow, with 
 sixteen pointed compartments radiating from it in yellow, green or red. 
 All are filled with cloud bands, arabesques and floral stems in colours. 
 Two hanging lamps are represented, to the right and left of this central 
 device, A fourth part of the same device is repeated at each corner. The 
 panels in the border are red and yellow, on a ground of deep purple tone. 
 
 ^ W. R. Holmes, Sketches on the Shores of the Caspian, 1845. 
 
 2 What remained of the second carpet passed into the Yerkes Collection ; it has since 
 been sold again, but it remains in America. 
 
 18 
 
PERSIA 
 
 It will be observed that human and animal forms^ both prevalent in Persian 
 carpets of the time, are not included* Such representations would not 
 have been suitable for a mosque* The design is of a double character* 
 The central ornament, with its sixteen radiating ogival panels, belongs, 
 not to the ground like the floral stems, but to the dome overhead* Such 
 embellishments may be found on Persian domed ceilings* There is 
 an example in the vestibule of the college of the Shah Hussein at Ispahan, 
 where the ceiling has in the middle a circular panel with sixteen pointed 
 projections and sixteen ogival panels radiating from them, very similar 
 to the device on the carpet*^ The two mosque lamps are appropriately 
 introduced as hanging from the ceiling on either side* The pile is 
 entirely of wool, on silk warps* There are about three hundred hand- 
 tied knots to the square inch, making a total of about 30 million knots for 
 the whole carpet* This represents approximately twenty-four years^ 
 work of a single skilled weaver* It is probable, however, that several 
 assistants worked with the master-weaver* Assuming that so many as 
 eight people were able to work together at the loom, even then the carpet 
 would take three years^ continuous labour to make* 
 
 Another carpet in the Victoria and Albert Museum almost challenges 
 the claim of the Ardabil carpet to the first place in the collection* Both 
 are the most famous examples of the respective classes to which they 
 belong — for they are conspicuously different in design, as well as in colour* 
 Every detail of the elaborate design is planned with the utmost care, and 
 copied by the weaver with consummate skill* The colouring is harmonious, 
 with a simplicity and at the same time a subdued splendour which never 
 wearies the eye* The knotting is very fine, as required by the design* 
 The warps are of silk, and the pile is of wool throughout, with about 
 450 knots to the square inch* The first plate of this volume reproduces 
 the carpet in colours* It will be seen that the ground is a purplish crimson 
 of a carmine tone, although no coloured plate can do justice to the colouring 
 of the original* Upon this ground is a network of panels in very deep 
 blue, almost black* There is a large circle of lobed outline to the right 
 and left of the centre of the carpet* In the middle of each side is a half- 
 section of this panel, and a quarter-section in each corner* Eight pointed 
 oval compartments unite the central panels with the sections round the 
 edge* This disposition recalls the scheme of the Ardabil carpet, but it 
 is less elaborate and more closely linked up* The panels are filled with 
 a pattern of arabesques and palmettes, and each pointed oval has two 
 ^ P. Coste, Monuments modernes de la Perse, Paris, 1867, PI. XXI. 
 
 19 
 
HAND -WOVEN CARPETS 
 
 flying ducks in addition* The centre of the carpet^ often made the most 
 conspicuous feature by the Persian designer, is here occupied by a small 
 round pond with fish. On either side of the pond stands a large two- 
 handled vase supported by two dragons on a lion-pedestal, and containing 
 flowers. One vase and half of the pond is repeated in the middle of 
 each of the short sides of the carpet. The rest of the field is covered 
 with blossoming and fruit-bearing trees and floral stems, amid which 
 are lions (some preying on antelopes and black oxen), gazelles, and falcons 
 pouncing on herons. The wide border is divided in counterchange 
 fashion by slender arabesque stems, with a dark blue and crimson ground 
 matching the middle colours. It has a pattern of slender floral stems, 
 and cloud forms, with animals, birds and dragons. The heads of lions 
 and other animals among the interlaced stems of the narrow outer border 
 should be noticed. There is a restraint and regular sequence both in 
 design and colouring which have induced writers to attribute it to an early 
 period. Dr. von Bode ascribes it to early Safidian times. Martin assigns 
 it to Eastern Persia at an earlier date still — about the middle of the fifteenth 
 century.^ The place of the design in order of sequence is undoubtedly 
 where these experts put it. We must, however, admit the greater 
 probability of a sixteenth-century origin until more light is thrown on 
 the problem of early Persian carpets. The use of silk warps, and the 
 character of the vases and other details of the pattern, favour that view. 
 This carpet, with that from Ardabil, has an indisputable place in the 
 foremost ranks of the great carpets of the world. Several of the classes 
 of motives used in it are retained as the main theme of Persian carpets 
 in the sixteenth and seventeenth centuries. A remarkably fine carpet 
 belonging to the Earl of Ilchester is of that time (Plate 12). The hunting 
 animals are set in a pattern of floral stems, on a red ground. 
 
 One of the best known of the carpets representing roaming animals 
 is in the Museum of the Gobelins, Paris (Plate 13).^ The lions, tigers 
 and leopards, seen preying on oxen, stags and antelopes, are on a red 
 background, covered with conventional stems in place of the naturalistic 
 trees usually shown. A lobed circular compartment in the middle is in 
 pale blue and red, with animals, floral stems and cloud-bands. On the 
 right and left of this compartment there is a device, half-finial, half-vase, 
 evidently a modified rendering of the vase with two peacocks shown on 
 the carpet just described (Plate i). The spandrel in each corner is 
 
 ^ Bode, V orderasiatische Knupfteppiche, 2nd ed., p. 33 ; Martin, O.C., p. 33. 
 
 2 Reproduced in colours in Vienna, O.C., PI. 74. 
 
 20 
 
PERSIA 
 
 brownish yellow with floral stems and arabesques interlaced. The longer 
 panels in the border are inscribed with a poem, describing the carpet as 
 pressed by the foot of a world-ruling monarch, and beautiful as Eden, 
 so that it is an object of envy to Chinese art. Its flowery bed shows an 
 ever-lovely spring, that no gales or storms can touch. 
 
 Verses which have given the inspiration for the design, or perhaps in 
 some cases have been suggested by it, are inscribed on numbers of Persian 
 carpets. They are usually to be seen, as in this instance, in bold letters on a 
 row of panels in the border. Sometimes they run continuously along narrow 
 bands, or fill smaller spaces. They speak of the springtime, of the stars, 
 the clouds, the winds, the dew and the raindrops ; of gardens, flowers and 
 birds ; of love, and the wine-bowl ; they give expression to good wishes 
 for freedom from care, and for springtime all the year round. Sometimes 
 these verses are borrowed from their national poets, whom the Persians 
 never weary of quoting. A poem in praise of the spring flowers — the 
 rose, narcissus, tulip and violet — is woven in silver thread on a carpet 
 with a similar border to the last, in the Musee des Arts Decoratifs in 
 Paris.^ There is a large circle in the middle of this carpet with blossoming 
 trees, flowers and peacocks, on panels of different colours. The ground 
 beyond is black, and covered with stems and cloud forms in colours and 
 silver thread. An ode to the nightingale among the rose-bushes is on the 
 inner border of a carpet in wool, gold and silver lent by Baron Nathaniel 
 Rothschild to the Vienna Exhibition of 1891.2 The bird and the flowers 
 are represented in the middle of the carpet, round a small pond. The 
 rest of the design consists of a formal arrangement of arabesques and 
 palmettes. Verses describing the delights of spring, and others praising 
 the nightingale are inserted on a carpet with arabesque patterns in a private 
 collection in Paris.® The most celebrated of all Persian poets, Hafiz, 
 is quoted at some length on a remarkably fine carpet in the Salting Collection 
 at South Kensington (Plate 14). The materials are wool and gold and 
 silver thread on silk warps, and the texture is very fine. The verses are 
 in silver thread on elongated dark red panels in the border. They have 
 been translated as follows : — 
 
 Call for wine and scatter roses : what dost thou seek from Time ? — thus spake 
 the rose at dawn, O nightingale, what sayest thou ? 
 
 ^ Reproduced in colours, Vienna, O.G., PI. 71. 
 
 2 Reproduced in colours, Vienna, O.C., PI, 97. 
 
 ® Illustrated in colours in Ancient Oriental Carpets, Leipzig, 1906-8, PI. 2. 
 
 21 
 
HAND-WOVEN CARPETS 
 
 Take the cushion to the garden, that thou mayest hold the lip and kiss the cheek 
 of the beloved and the cup-bearer, and drink wine and smell the rose. 
 
 Proudly move thy graceful form and to the garden go, that the cypress may 
 learn from thy stature how to win hearts. 
 
 To-day while thy market is full of the tumult of buyers, gain and put by a 
 store out of the capital of goodness. 
 
 Every bird brings a melody to the garden of the King, — the nightingale, songs 
 of love, and Hafiz, prayers for blessing.^ 
 
 The pattern of floral sterns^ cloud-bands and birds, and tigers fighting 
 dragons, betrays Chinese influence. The device of a lion's mask within 
 a palmette, frequently seen on Persian carpets, is repeated several times ; 
 the masks are sometimes merely attached to the stems like flowers. In 
 spite of the wonderful freshness of colour and the fine state of preservation 
 of this carpet, it cannot very well be attributed to a later date than the 
 seventeenth century. It was obtained in Constantinople rather more than 
 thirty years ago, when it was stated to have been in the collections of 
 the Sultan. 
 
 Another carpet reputed to have come from the Turkish sultan's 
 palace is of the same class, in the same materials and equally well preserved. 
 It was obtained by Prince Lobanoff-Rostowsky when Russian ambassador 
 in Constantinople in the late 'seventies of last century, and afterwards 
 ceded by him to the Stieglitz Museum in Petrograd.^ The border has 
 an ode to a conqueror, greater than Darius, Alexander and Feridun, 
 before whose victorious progress throne-carpets are spread. The central 
 panel is similar to that on the carpet in the Salting Collection. The 
 white ground around is covered with delicate interlacing stems of flowers 
 and cloud-bands, amid which tigers, leopards stalking stags, jackals and 
 Chinese dragons are seen. Among the stems in the green spandrels 
 are parrots, pheasants, and a falcon seizing a heron. The alternation of 
 inscribed elongated panels with circular panels to form a border, as 
 exemplified in the carpets just described, is quite often found in other 
 Persian works of art. A segment of a damascened brass dish, showing 
 such a border, is illustrated on Plate 15B. A Persian begging-bowl, of 
 engraved copper, reproduced on the same Plate (15A), shows the Chinese 
 cloud-band as seen on the carpets. These animated forest or park-land 
 
 ^ Hafiz, Divan (ed. H. Brockhaus, Vol. Ill, p. 175-6). 
 
 2 Illustrated in colours in Vienna, 0 .( 7 ., PI. XI. A description of this carpet by M. A. 
 Polovtsoff, Director of the Museum, has been published in the Burlington Magazine (Vol. 
 XXXV, 1919, p. 16). 
 
 22 
 
PERSIA 
 
 scenes have not always the accompaniment of descriptive verses in gold 
 and silver, 
 
 A woollen carpet in the possession of Prince Schwarzenberg ^ has 
 a pond with ducks in the middle. The deep blue ground beyond is closely 
 covered with cypresses and blossoming trees and plants naturalistically 
 drawn^ with a great phoenix sei 2 ;ing a small bird in each corner, and more 
 birds and animals below. 
 
 Half of a carpet, in wool on silk warps, with a similar design in different 
 colours, is in the Cathedral of Cracow ; the other half found its way to 
 Paris, where it is now in the Musee des Arts Decoratifs,^ 
 
 In other examples, which need not be specified, the trees and flowers 
 cover the whole ground, not balanced in opposing directions from the 
 middle, as is usual in carpets Vv^ith central panels, but continued in an 
 unbroken stretch from one end to the other. The borders of such carpets 
 vary considerably ; they are generally wide, with a formal design as a 
 contrast to the free rendering of the middle pattern,® 
 
 One example, in which the landscape idea is very clearly marked, 
 should not be allowed to pass unnoticed. It is in the Museum of Art 
 and Industry at Vienna, The trees, shrubs and plants are represented 
 with careful fidelity to nature. There are peacocks, pheasants and storks 
 on the ground, and various smaller birds in the foliage,”* 
 
 Much help may be derived from Persian miniature paintings, 
 particularly from those of ascertained date, in the study of carpets. Their 
 rendering of landscape often explains a carpet design, and sometimes 
 an actual carpet is represented, A painting of the seventeenth century, 
 here illustrated, will suffice to show the lines on which comparisons may 
 be made (Plate i6), A Persian prince is seen in a garden with its central 
 pond and ducks, its straight water- channels and vases. Carpets with 
 figures remain yet to be described ; the other motives have already been 
 met with. Beyond the enclosure are the cypresses and blossoming trees 
 so often seen on the carpets. Suspended over the princess head, ignoring 
 in its delineation all rules of perspective, is a floral carpet serving as a 
 canopy. Courtiers are seated on others spread on the ground. Another 
 
 ^ Illustrated in colours in Vienna, O.C., PI. XXXI. See also Meisterwerke, Munich, 
 PI. 45. 
 
 2 Vienna, O.G., PI. 31 ; G. Migeon, Exp. des Arts Musulmans, Paris, 1903, PI. 73. 
 
 2 There is a specimen in the Victoria and Albert Museum, No. 3 — -1887. 
 
 ^ Reproduced in colours, Vienna, O.G., PI. I. Dr. Martin [Oriental Garpets, Fig. 236) 
 claims an Indian origin for this carpet. It may have been made in India, but the motives 
 are Persian. 
 
 23 
 
HAND-WOVEN CARPETS 
 
 miniature of about the same period shows three pile-carpets — one spread 
 on a low platform for a prince to sit on, another forming a canopy over 
 his head, and a third thrown over the back of an elephant (Plate ly)* 
 One of the simplest and most pleasing kinds of Persian carpet-design 
 is based on plant forms throughout. An incomplete carpet of this type 
 in the Austrian Museum of Art and Industry, Vienna, is illustrated in 
 Plate 1 8. The white ground is covered with an arrangement of trailing 
 floral stems, while another system of slender stems superimposed and 
 carrying large palmettes suggests a partition into panels of regular shape. 
 In the middle of one of these is a vase ; there would have been several 
 others in the entire carpet.^ In many of these carpets the stems issue 
 from vases, generally disproportionately small, placed at regular intervals. 
 They may seem a little tame after such carpets as have been described 
 hitherto, in which much use has been made of the enlivening effect of 
 animals and birds, but the variety of plant forms and the usual quiet 
 harmony of the colouring make amends. Two examples in the Victoria 
 and Albert Museum show a variation in the design. Instead of the 
 trailing stems, flowering plants are shown entire to the roots, as they 
 would be seen in a garden-bed. They are enclosed by the stems bearing 
 palmettes, as in the carpet just described. The larger one was formerly 
 a valued possession of William Morris. The other was obtained in 
 Constantinople. Each has a deep red ground and a blue border.^ 
 
 The naturalistic floral treatment is somewhat modified, in a fine 
 carpet in the same Museum, by a succession of large lozenge -forms of 
 lobed outline in dark and light blue, purple-brown, salmon pink and 
 white (Plate 19). The carpet is large, and the panels are introduced to give 
 relief to the wide stretch of deep crimson colour. The stems pass through 
 them without interruption. The vases are shown sometimes in the 
 panels and sometimes beyond. There is a deep blue floral border. 
 
 When the vases disappear from these carpets, there is sometimes 
 a more assertive arrangement of the floral motives in panels. A carpet 
 belonging to Mr. Lamm of Naesby has a series of panels, each nearly 
 4 ft. long, in various colours, containing formal floral ornament ; the 
 intervening spaces each contain a blossoming tree on a dark blue ground.® 
 
 ^ Reproduced in colours in Ancient Oriental Carpets, Leipzig, 1906-8, PI. ii. 
 
 2 The first is illustrated in colours in Ancient Oriental Carpets, Leipzig, 1906-8, PL 19. 
 Others similar are reproduced in Pis. 5, 9, and 10. Sir Isidore Spielmann has a portion of 
 a very fine carpet of this type (Victoria and Albert Museum, Catalogue of the Franco- British 
 Exhibition, 1921, PI. 32). 
 
 ® Illustrated in colours in Martin, 0 .( 7 ., PI. i ; also in Meisterwerke, Munich, PI. 66. 
 
 24 
 
PERSIA 
 
 A very effective design of the type was exemplified by a carpet lent by 
 Count Clam-Gallas to the Munich Exhibition of igiod The panels 
 are alternately quatrefoil and fan-shaped^ with floral and arabesque patterns^ 
 cloud-bands and peacocks on grounds of several colours. The background 
 is dark red^ with flowering trees. At the same exhibition was to be seen 
 a carpet belonging to the Imperial Ottoman Museum at Constantinople 
 in which the ground is entirely divided up into large lozenge-spaces of 
 various colours containing floral forms.^ A portion of a very similar 
 carpet is in the Victoria and Albert Museum (Plate 20 b) and another is 
 in the Kunstgewerbe Museum^ Berlin,® These floral carpets are generally 
 attributed to the weavers of the province of Kirman in the south-east. 
 It is not without significance that a favourite type of carpet made there 
 in more recent times shows close rows of vases of flowers^ generally on 
 a white ground.^ The almost entire absence of Chinese motives in these 
 carpets is a strong argument in favour of a South Persian origin. 
 Occasionally carpets of this type have a long looped nap at the back^ 
 like a bath towel, for warmth. This would be a very suitable provision 
 at Kirman, where the cold is sometimes severe, owing to its altitude 
 of 6,000 ft. A modern loom from Kirman, with a typical carpet of the 
 district partly woven upon it, is illustrated in Plate 92. 
 
 One of the largest Oriental pile-carpets ever made, and one of singular 
 design, was formerly in the Hall of the Forty Columns (Chihal Sutun) 
 at Ispahan, It measured more than 70 ft, long by 30 ft. wide,® So wide 
 a loom for a finely knotted carpet must have been a prodigious affair to 
 work at, and the difficulty of keeping the carpet even and straight during 
 its manufacture must have been great. It had a red ground, with a 
 formal arrangement of large white flowers and green stems behind a trellis 
 of yellow stems, each opening being 9 ft, in length. The border, which was 
 6 ft, wide, was filled with the same ornament on dark blue. This carpet was 
 placed in the great hall behind the throne-room.® It has been gradually 
 dismembered by travellers within the last forty years. Three fragments 
 ultimately found their way to the Victoria and Albert Museum, where also 
 
 ^ Illustrated in MeisUrwerhe, PI. 50 ; in colours in Vienna, O.C., PI. 32 ; also in 
 Martin, O.C., Fig. 102, and Bode, Vorderas. Kniipfteppiche, 2nd ed., Fig. 19. 
 
 2 Meisterwerke, PI. 52 ; Martin, 0 .( 7 ., Fig. 86. 
 
 3 Reproduced in colours in Ancient Oriental Carpets, Leipzig, 1906-8, Pis. 9, 21. 
 
 ^ A fine example is in the Victoria and Albert Museum, No. 100. It is inscribed with 
 the name of a Kirman weaver. 
 
 ® See p. 106. 
 
 ® There is a plan of the Pavilion, showing the room in which the carpet was placed, 
 in P. Coste, Monuments modernes de la Perse, Paris, 1867, PI. 41. 
 
 25 
 
HAND-WOVEN CARPETS 
 
 may be seen a drawing reconstructing the design (Plate 21 )* The carpet 
 was of the period of Shah Abbas the Great (1586-1628), who built the 
 hall in which it was placed ; it was probably made during the latter part 
 of this monarch's reign* Dr. F. R. Martin, who emphasises the 
 approximation of the design to those on Indian carpets at Jaipur, claims 
 it for the Indian weavers. The arguments against such an attribution 
 are formidable. Why should Abbas, under whom carpet- weaving in 
 Persia flourished greatly, have the carpet for the great open pavilion of 
 his own palace woven in India s' And assuming that he did, how was 
 the wellnigh insurmountable task of transporting such a bulky object 
 over the difficult country between the Panjab and Ispahan overcome s' 
 Dr. Martin claims that all such carpet designs were woven in India. It 
 appears easier, on the whole, to adapt the theory to fit the carpet 
 than the carpet to fit the theory. What was done at Lahore in factories 
 inaugurated by Persian weavers may have been done in Persia as 
 well. 
 
 The Persians have not shackled their artistic gifts by notions about 
 the precedence of one kind of art over another. They have ignored all 
 distinctions of the sort, and much of the fine quality of their art 
 is due to this. The range of design seen in their carpets, for example, 
 is no less extensive than that in their paintings, and it includes 
 every class of subject attempted by carpet-weavers elsewhere. Hunting 
 scenes, first shown on the carpets of Persia, are only found outside that 
 country in Indian copies. Figure-subjects form another special group, 
 shared only by India (again as a copyist) and China, if we may pass over 
 the diminutive and grotesque human forms which occasionally fill odd 
 corners in Caucasian and Turcoman work. It has been reserved for the 
 modern Kirman weaver to turn his whole carpet into a picture in semi- 
 European style.i In older Persian carpets figures are freely introduced, 
 though with a better sense of restraint. In hunting scenes the theme 
 requires the figures to be generally distributed. In most other cases they 
 appear singly or in small groups — in the border or spandrels, or in 
 compartments upon the central field. In the silk brocades of Persia, 
 where a better opportunity for showing detail renders comparison with 
 
 1 A carpet of extraordinarily fine technique, in the possession of Sir Charles Marling, 
 shows a group of figures taken from Watteau’s Fetes venitiennes. An inwoven inscription 
 states that it is from the Atelier of ‘ All Kirmani, and the dates a.h. 1324 and 1327 (a.d.= I9o6, 
 1909) are both woven. It must have taken several years to make. Other Kirman carpets 
 of similar type showing groups of ancient and modern worthies, or single portraits of the 
 Shah, are not rare. 
 
 26 
 
PERSIA 
 
 dated miniature paintings an easier task^ the weaving of figure-subjects 
 appears to have been in vogue during the reign of Shah Tahmasp 
 (1524-1576)^ lasting well into the seventeenth century. In the carpets 
 these subjects are found about the same time^ and they remain popular 
 throughout the following century. Figures are represented among the 
 trees of a garden^ eating fruit, drinking wine, or listening to music. By 
 a strange convention the figures are sometimes winged, but their 
 occupations are still the same, or they are seated alone. By the time 
 these carpets were woven the wings had lost all significance they may 
 ever have had, 
 
 A remarkable carpet in the possession of the Duke of Buccleuch 
 (Plate 22) is divided up into pointed oval and lobed compartments. The 
 former contain repetitions of three scenes — a falconer on horseback ; 
 two standing figures in a garden, one giving drink to the other ; and 
 three seated figures in a garden, one with a tambourine, another with a 
 bottle, and the third with a cup. In the oblong compartments are birds 
 and flowers. The red ground beyond these compartments is covered 
 with beasts of prey — lions, tigers, leopards and wolves — hunting stags, 
 antelopes and goats. The broad border has a succession of elongated 
 yellow panels enclosing dragons and phoenixes, alternating with pairs of 
 dragons intertwining in circles. The dark blue ground is covered with 
 running animals amid floral stems. The narrow outer border has masks 
 of lions and heads of oxen connected by slender stems, 
 
 A group of figures drinking and listening to music, seated round a 
 pond with water-fowl, is represented on a carpet formerly in the Bardini 
 Collection at Florence,^ The seated winged figures in the border 
 of the great hunting-carpet at Schonbrunn (Plate 10) are found 
 again in several carpets. One in the Lyons Museum has them in the 
 border, and scenes similar to those on the Duke of Buccleuch's carpet in 
 the middle,^ Single-winged figures are to be seen on the spandrels of a 
 remarkable carpet (of which barely one-half is preserved) of the sixteenth 
 century, given to the Victoria and Albert Museum by the late Mr. C, T. 
 Garland, It is in wool and gold and silver thread, with a fine design of 
 hunting animals,® A later hunting-carpet, entirely in wool, made in the 
 first half of the seventeenth century, has on each spandrel a group of three 
 
 ^ Martin, O.C., Fig. 127. 
 
 2 Martin, O.C., Fig. 138 ; R. Cox, L’Art de decorer les Tissus, Paris, 1900, PI. 53. 
 
 ® V.A.M. Notes on Carpet-knotting and Weaving, 1920, frontispiece. See also Bur- 
 lington Magazine, XXV, 1919, p. 12. 
 
 27 
 
HAND-WOVEN CARPETS 
 
 winged figures, in this case appropriately in cloudsd There is one more 
 instance in which some ultra-human attribute is apparently assigned to 
 the winged figures, on the celebrated silk carpet in the Poldi-Pe^2;oli 
 Museum at Milan,^ Among the trees and plants two of them are seated 
 before a bowl of flowers overshadowed by a canopy of Chinese form. 
 Perhaps in earlier versions of the motive the bowl was an altar. 
 
 There were originally two figures on the spandrels of a carpet in the 
 Kaiser-Friedrich Museum at Berlin ; one is seated and the other, probably 
 an attendant, standing. Nothing more than the lower portion of these 
 figures is preserved. The carpet was formerly in the synagogue at Genoa, 
 and the Jewish objection to the representation of human figures is thought 
 to explain the mutilation,® 
 
 Two old love-stories of Persia have kept their popularity for many 
 centuries — those of Khusrau and Shirin, and Laila and Majnun, They 
 appear again and again in Persian miniatures and textiles. Both are 
 shown on a carpet in the Musee des Arts Decoratifs in Paris (Plate 23), 
 Each scene is rendered in the traditional manner. In the first, the beautiful 
 Shirin has dismounted from her horse to bathe, and she is seated dressing 
 her hair, with her garments hanging on the bough of a tree above, when 
 Prince Khusrau (Chosroes), who has been out hunting, rides past. The 
 hunting-party, from which the Prince has strayed, has the authority of 
 the old story. The appearance among the riders, of Laila encountering 
 the poor distraught poet Majnun in the desert, may perhaps be explained 
 on the supposition that it was thought appropriate to bring in this story 
 where the other was shown. It will be noticed that two of the huntsmen 
 throw their bows over the necks of antelopes. The ground is deep 
 blue, and that of the border red. The outer stripe of the border is dark 
 blue, and the inner white. The latter has a fantastic pattern of heads 
 of men, lions and horses linked together by blossoming sterns.^ 
 
 A strange nautical scene is occasionally found in the spandrels of 
 carpets of a relatively late class, probably made towards the end of the 
 seventeenth century. One example, in the Lyons Museum, shows two 
 
 ^ In the Victoria and Albert Museum. Reproduced in colours, Ancient Oriental Carpets, 
 Leipzig, 1906-8 (supplementary volume to the great work published in Vienna in 1892), 
 PI. 22. 
 
 2 W. von Bode, V orderasiatische Kniipfteppiche , 2nd ed., Fig. 5; Martin, O.C., Figs. 
 118, 119. 
 
 ^ Bode, Vorderas. Kniipfteppiche, 2nd ed.. Fig. 12 ; Martin, 0 .( 7 ., Fig. 85 ; Vienna, 
 O.O., PI. 62. 
 
 ^ Illustrated in colours in Ancient Oriental Carpets, Leipzig, 1906-8, PI. 15. 
 
 28 
 
PERSIA 
 
 boats filled with figures^ and a man apparently drowning among the 
 fishes,^ The clue to this subject is not known ; the boats and figures 
 are European* Another carpet in the Reynolds room at Knole is closely 
 similar. A thirds belonging to Mr. Lamm, at Naesby, has the same 
 scene but there is only one boat.^ Two other carpets of the same class 
 are in the Metropolitan Museum, New York, and the Kunstgewerbe 
 Museum, Berlin. Each carpet shows a singular 2;ig2:ag arrangement of 
 the central space, with floral patterns on successive grounds of different 
 colours. This expanding 2;ig2:ag form of ornamentation is to be found 
 on the well-known velvets of Kashan in Central Persia. 
 
 This town has long been famous, and is still, for its carpets, and the 
 theory that the carpets with the nautical scenes were made there has much 
 in its favour. Ye^d also, the other great centre of velvet- weaving in Persia, 
 made carpets, but the industry there in modern times has declined.^ 
 
 In considering figure-subjects on carpets the remarkable cope of 
 Persian workmanship in the Victoria and Albert Museum should be 
 studied. It is very finely knotted in silk, like a carpet, with details in 
 gold and silver thread. Representations of the Annunciation and the 
 Crucifixion are seen, among the usual floral motives. 
 
 The description of the famous carpet of Chosroes (see p. lo), tells 
 of the use of silk, gold and silver, and even jewels, in a floor-covering. 
 It has already been pointed out that this sumptuous work cannot have 
 been a pile-carpet. For many centuries the knotting of Oriental carpets 
 was done in wool, and no material produces a more beautiful pile-surface. 
 When costlier threads began to be introduced, a type of carpet was made 
 which attracts the collector by its richness, though it is unsuitable for 
 placing on the floor. The first step was taken, so far as we can tell, about 
 the beginning of the sixteenth century, when the Persian state, after a 
 long period of alien rule, was becoming unified under the native Safidian 
 dynasty. The epoch was marked by an output of artistic wares on a 
 most generous scale, as the museums of Europe now bear witness. The 
 elaboration of carpet-patterns, involving the tracing of delicate curves 
 and minute details, called for a finer texture, and this necessitated the 
 use of a warp at once thin and strong. The luxury of the times permitted 
 the employment of silk for the purpose, although the thread formed 
 
 ^ R. Cox, L’Art de decorer les Tissus, Paris, 1900, PI. 52 ; Martin, O.C., Fig. 149. 
 
 2 Martin, O.C., PI. 6. 
 
 2 It will, of course, be understood that there is no technical affinity between velvets 
 and carpets. 
 
 29 
 
HAND-WOVEN CARPETS 
 
 only the foundation of the fabric^ and was hidden in the finished texture. 
 The pile was still entirely of woofi the warp threads on which the knots 
 were made and the weft shoots passed through to keep them in line being 
 alone of the richer material. The ** Ardabil carpet, of the year 1540, 
 was thus woven. The next step seems to have been the introduction 
 of gold and silver threads, sparingly at first, for details of the pattern. 
 As the use of metal threads was gradually extended silk was substituted 
 for the wool of the pile, until at last carpets were made entirely of silk 
 on a gold or silver ground. Wool thus disappeared altogether in these 
 sumptuous carpets. Sometimes the metal threads were dispensed with, 
 and the material used was silk throughout. The evolution was com- 
 pleted by the end of the sixteenth century. Meanwhile the making of 
 woollen carpets went on. 
 
 The seventeenth- century carpets woven entirely in silk, with metal 
 threads for the ground, have for long been the subject of conflicting 
 views. Attention was first drawn to the problem of their provenance 
 at the time of the Paris Exhibition of 1878. A very fine example, the 
 property of Prince Czartoryski, was then shown in the Salle polonaise 
 of the Palais du Trocadero.^ The ground of this carpet was of gold 
 and silver thread, not knotted but wrought into the warps by a method 
 analogous to tapestry. The design, of palmettes and floral stems, quite 
 Persian in character though arranged in a somewhat formal manner, 
 was knotted by the ordinary method in coloured silks. In the middle, 
 and again at each corner, was the coat-of-arms of the owner's family. 
 The European heraldry, associated with the unfamiliar materials of the 
 carpet, raised doubts in the minds of the experts of the day as to its Oriental 
 origin, and the theory was advanced that it was the work of a member 
 of a well-known family of weavers established in Poland, Madziarski 
 by name. The appellation of '' Polish," given to carpets of this class 
 in consequence, clung to them for many years. Doubts afterwards 
 arose, which were strengthened when it became clear that the weaving 
 establishment of the first of the Mad^arski was not in operation early 
 enough. But the theory served a good purpose, for when its foundation 
 was seen to be insecure an incentive was given to further investigation. 
 M. T. Krygowski, writing in the Orientalisches Archiv,^ quoted an 
 inventory of a member of the household of King Albert of Prussia, made 
 
 ^ E. Guichard and A. Darcel, Tapisseries decoratives du Garde-Meuble, 1877, PI. 94. 
 It is now in the Czartoryski Museum at Cracow. 
 
 2 II, 1911-12, pp. 70, 106. 
 
 30 
 
PERSIA 
 
 in 1578, in which three Polish carpets are mentioned, and another of 
 the Treasury of Prince Ostrogski in Dubno, of the year 1616, with an 
 entry of a Polish carpet inwrought with gold. He also showed that 
 there was a silk factory in Poland as early as the first half of the seven- 
 teenth century. All this is useful, but so far it does not involve the assump- 
 tion that carpets like Prince C2:artoryski's are Polish. 
 
 The Poles, like the Russians, always had a liking for rich Oriental 
 stuffs, and many must have been imported. So far as Mad2;iarski^s 
 weaving establishments are concerned, there is no evidence that they ever 
 made carpets. The first weaver of the family, John, was established 
 about the middle of the eighteenth century at Sluck. The discovery 
 that his signature to an agreement with Prince Rad2;iwill, for conducting 
 a ** Persian ** factory and teaching Persian art, was in the Armenian language, 
 lent colour to the theory that he was an immigrant weaver of carpets. 
 It appears, however, that the name Mad^iarski means Magyar,"" and 
 it is surmised that the weaver"s father may have lived in Hungary, whence 
 he would have been carried as a prisoner to Stambul by a raiding-party 
 of Turks. It was from Stambul that John Mad2;iarski smuggled the first 
 weaving appliances in pieces into Poland. He wove girdles and brocades 
 of silk and gold and silver after Persian models.^ 
 
 It is beyond dispute that carpets woven in gold, silver and silk were 
 made in Persia. These are the materials of the famous Schonbrunn 
 carpet already described (p. 16). The French traveller Jean Baptiste 
 Tavernier, as Dr. Sarre points out,® saw "" rich carpets in gold and silk "" 
 in the Royal Palace at Ispahan in the first half of the seventeenth century. 
 Rather earlier than the date of Tavernier "s journey, carpets of these 
 materials arrived in Venice direct from Persia. 
 
 Four fine silk and gold carpets are still preserved in the Treasury 
 of St. Mark"s at Venice. There is satisfactory evidence that three, at 
 least, were presented to the Doge by Shah Abbas the Great of Persia. 
 An embassy from the Shah, arriving at Venice in 1603, brought a carpet 
 in silk and gold and silver, with the suggestion that the church treasure 
 should be displayed upon it at the time of the annual exhibitions to the 
 public. In 1622 the Persian ambassador brought two more.® 
 
 ^ There is a collection of these in the Victoria and Albert Museum, one having Madziar- 
 ski’s name. Others were shown at the Munich Exhibition of Muhammadan art in 1912 
 {Meisterwerke Muham. Kunst, PI. 222). 
 
 2 Ancient Oriental Carpets, Leipzig, 1906-8, p. 3. 
 
 ^ G. Berchet, La repubblica di Venezia e la Persia (Turin, 1865). See A. Pasini, 
 Tesoro di 8 . Marco in Venezia, Venice, 1885-6, Pis. 77, 89-92. 
 
 31 
 
HAND-WOVEN CARPETS 
 
 A beautiful silk and gold carpet is in the Royal Palace at Stockholm. 
 Like the Austrian Emperor's, it has a ** hunting " subject, but much 
 simplified. The huntsmen are afoot — some struggling with lions, 
 and others carrying slain bears on their shoulders. There are also deer, 
 antelopes and other animals. The colours are full and deep. The 
 crimson ground is relieved by a large central panel and spandrels in gold 
 thread. There is a wide pale yellowish border, with an inner and outer 
 stripe, the former black and the other crimson.^ The famous Coron- 
 ation carpet " of the Danish royal house is kept at Rosenborg Castle, 
 Copenhagen. The ground is entirely of gold thread, producing an effect of 
 great richness. The pattern consists of floral stems, palmettes, arabesques 
 and cloud-bands in silk pile with touches of silver thread.^ Both these 
 carpets have been in Scandinavia since the seventeenth century. They 
 were probably brought by embassies from the East, but there is no precise 
 record. 
 
 During the seventeenth century large numbers of carpets in silk 
 pile, mostly with gold and silver thread as well, were made. The patterns 
 show interminable variations of floral stems, palmettes and arabesques, 
 with occasional additions of cloud-bands. A common form of border is 
 the counterchange cresting. There is often a liberal range of colour, with 
 pale, pure tones predominating. Pinks and pale greens are favourites, 
 with deep blue sometimes to give emphasis. Undoubtedly the silk 
 carpets have faded more than the woollen ones, but the contrasts aimed 
 at seem to be rather that between metal and colour than between one 
 colour and another. 
 
 An example of the best kind, the property of the royal house of 
 Saxony, is reproduced in Plate 24.^ The ground of the middle is gold, 
 and of the border silver. The colours are mostly pale in tone ; the 
 dark parts conspicuous in the reproduction are deep blue. The Duke 
 of Buccleuch possesses a very similar carpet. 
 
 ^ There is an excellent coloured illustration in Martin, O.C., PI. 5 (half-tone reproduc- 
 tion of the whole, PI. 4). The crimson is a little deeper than there shown, and the border 
 paler. The carpet is large, measuring rather more than 18 feet long by 9 ft. 4 in. wide. 
 It is very finely knotted ; Martin computes that there are 20 million knots. 
 
 2 Reproduced in Martin, O.C., Pis. 7, 8. The attribution to Persia of the four embroi- 
 dered carpets in silk and gold thread at Rosenborg Castle, also illustrated by Martin (Plates 
 11-14) and considered by him to have been brought from Persia with the Coronation carpet, 
 hardly carries conviction. The Chinese motives are too true to type, and the Persian motives 
 look like the work of an imitator. Their appearance strongly suggests an Indo-Chinese 
 origin. 
 
 ^ Reproduced in colours in Vienna, 0 .( 7 ., PI. 46. 
 
 32 
 
PERSIA 
 
 There is a good example, though incomplete, in the Salting Collection 
 at South Kensington,^ with floral and palmette ornament, and an inter- 
 laced border, on gold and silver grounds* This carpet bears some resem- 
 blance to one of those in the Treasury of St. Mark’s, Venice. Other 
 specimens are to be found in the principal museums of Europe and America, 
 and in the possession of wealthy collectors. Though very sumptuous, 
 they are not altogether satisfactory as carpets. They are unsuitable for 
 placing on the floor, and the effect of richness is overdone. Although 
 it is now generally agreed that they were not made in Poland, there is 
 still some mystery about their place of origin. They are not always 
 entirely true to the Persian tradition of carpet-design. There can be 
 no doubt, however, that some were made in Persia. Perhaps, when 
 more is known about carpet-making in India, evidence may be forth- 
 coming that the manufacture was taken there from Persia. Wherever 
 the weaving was done, it does not seem to have been carried on very 
 long. This sumptuous manufacture appears to have died out before 
 the end of the century which saw its rise. A carpet in silk only, without 
 metal threads, belonging to the Gobelins Museum is reproduced in 
 Plate 25. The central quatrefoil is green and the spandrels white. The 
 rest of the middle space is red. The wide border is yellow, with a red 
 outer and blue inner band. 
 
 Before returning to the later woollen carpets of Persia, reference 
 may be made to another type, woven in the same materials as those just 
 described, but by the tapestry process. The thin and delicate texture 
 may perhaps account for the small number now known to exist — hardly 
 more than a dozen altogether. 
 
 Carpets of this class, in silk, gold and silver, are occasionally enlivened 
 with human or animal forms. A fine example, with animals attacking 
 one another, besides grotesque Chinese dragons and phoenixes, is in the 
 Royal Bavarian Collection at Munich.^ 
 
 The colours used are the same as those of the silk pile-carpets, but 
 the different texture produces a much harder effect, more like mosaic- 
 work. 
 
 The best known example, and one of the most interesting, is in the 
 Louvre (Plate 26). A pointed oval in the middle contains a representa- 
 
 ^ Reproduced in colours in Vienna, O.C., PI. 96. The statement, published at the time 
 of its exhibition at Vienna in 1891, that heraldry was represented upon it was incorrect ; 
 it is still repeated. 
 
 2 Munich, Meisterwerke, PI. 55. 
 
 33 
 
 D 
 
HAND-WOVEN CARPETS 
 
 tion of a horseman fighting a monstrous serpent — evidently meant for 
 the Persian national hero Rustam^ on his horse Reksh, encountering the 
 serpent while on his way to deliver the King Ky-Kaoos from captivity 
 in Mazanderan, The story^ taken from the Shah-Namah^ the national 
 epic of Persia^ may be given in the words of Sir John Malcolm, In 
 the middle of the night a monstrous serpent, seventy yards in length, 
 came out of its hiding-place, and made at the hero, who was awakened 
 by the neighing of Reksh ; but the serpent had crept back to its hiding- 
 place, and Roostem, seeing no danger, abused his faithful horse for disturb- 
 ing his repose. Another attempt of the serpent was defeated in the 
 same way ; but the monster had again concealed himself, Roostem 
 lost all patience with Reksh, whom he threatened to put to death if he 
 again awakened him by any such unseasonable noises. The faithful 
 steed, fearing his master's rage, but strong in his attachment, instead 
 of neighing when the serpent again made his appearance, sprung upon 
 it and commenced a furious contest ! Roostem, hearing the noise, started 
 up and joined in the combat. The serpent darted at him, but he avoided 
 it, and, while his noble horse seized their enemy by the back, the hero 
 cut off its head with his sword," ^ The spandrels contain illustrations 
 of the old Arabian love-story of Laila and Majnun — the scene where 
 Laila visits the poor poet in the desert,^ 
 
 Three carpets of this kind were lent by the King of Bavaria to the 
 Exhibition of Muhammadan Art at Munich in 1910, One of them has 
 in the middle a shield-of-arms, misunderstood by the weaver, but sup- 
 posed to be intended for those of a Polish princess married in 1642 at 
 Warsaw,^ Another shows the usual hunting scene — men on horseback 
 and on foot attacking wild animals. Garden scenes fill the spandrels, 
 with figures by a stream. In the border winged attendants are playing 
 music or offering fruit to a winged prince seated on a throne, and the 
 middle panel contains other winged figures. The third carpet has a 
 conventional floral pattern. An example with a pattern of arabesques, 
 floral stems and cloud-bands is in the collection of the King of Saxony,* 
 In the later woollen carpets of Persia the floral types survive almost 
 alone. Subjects with figures and animals are not common, and when 
 they again achieve popularity during the course of the nineteenth century, 
 ^ Sketches of Persia, 1828, ch. xii. 
 
 2 Another representation of this story, on a carpet in the Musee des Arts Decoratifs, 
 Paris, has already been mentioned (p. 28). 
 
 ^ Meisterwerke, Munich, PL 60. See also Pis. 61, 62. 
 
 ^ Vienna, O.G., PL 30. 
 
 34 
 
PERSIA 
 
 the effects aimed at are more naturalistic if less happy. The varieties 
 of floral patterns are almost endless. While some tend towards the 
 direct imitation of natural life/ others show a more or less formal arrange- 
 ment of palmettes and stems covering the whole ground, A fragment 
 of a carpet reproduced in Plate 27A, made in the seventeenth century^ 
 is of the latter class. It shows the usual red colour of the central panel 
 and the dark blue border, A more elaborate example, probably made 
 early in the seventeenth century, is in the Kunstgewerbe Museum at 
 Leipzig (Plate 28), A further stage in the formalizing of the pattern is 
 shown in Plate 29, At last it breaks up into a kind of colour-mosaic, 
 often effective enough, but no longer to be regarded as an organic and 
 intelligible design (Plate 27B), 
 
 The later stages of another Persian type, already spoken of as associated 
 with Armenia, may be traced in Plate 20A, where the long leaf-forms 
 covered with floral motives dominate the whole pattern ; and in Plates 
 132A and 134, where old motives may still be traced. The double 
 prayer- carpet reproduced in Plate 30B belongs to a class of which few 
 examples are known. Knotted prayer- carpets are unusual in Persia, 
 such articles being mostly made of cotton, either embroidered or printed. 
 
 The type of pattern in which the design is composed entirely of 
 palmettes and floral stems, sometimes with the addition of cloud-bands 
 (Plates 27 and 28), is one of the best known of all. It is ascribed to the 
 province of Herat, now united to Afghanistan, though under Persian 
 rule during the flourishing days of carpet-weaving. Early in the seven- 
 teenth century the province had the reputation of making the best carpets 
 in Persia (see p, 39), The capital city, Herat, was the chief centre 
 of the manufacture. The industry flourished there until the desolation 
 of the province by Nadir Shah in 1731, when many weavers were trans- 
 ported to the western provinces. Although Herat never recovered, 
 carpet-weaving did not entirely cease, and it is still carried on there at 
 the present time. There is evidence that such carpets reached England 
 early in the seventeenth century. King James I is seen standing on one 
 of them in a State portrait, attributed to Vansomer, at Hampton Court, 
 They are also represented in pictures of the Netherlands and Spain in 
 the seventeenth century. 
 
 The carpets of Kirman have already been noticed. The French 
 traveller Chardin, who visited Persia in the second half of the seven- 
 
 ^ Example in the Victoria and Albert Museum, No. 3 — 1887. Another in Lyons Museum. 
 (R. Cox, L’ Art de decorer les Tissus, Paris, 1900, PI. 56.) 
 
 35 
 
HAND-WOVEN CARPETS 
 
 teenth century, speaks of the carpets of the provinces of Kirman and 
 Seistan. 
 
 Carpets were also made at Ispahan, though apparently not in great 
 numbers. There is little justification, as a rule, for the frequent attribu- 
 tion of old carpets to this former capital of Persia. Other well-known 
 centres of carpet-weaving in Persia have been mentioned in the preceding 
 pages. There is much uncertainty in regard to the locality of manufacture 
 of old Persian carpets, and attributions should be accepted with reserve. 
 It is seldom that they bear the hall-marks of their origin in the unequivocal 
 way of the ** Ardabil ** carpet. 
 
 In this connexion a carpet belonging to Mr. G. Mounsey merits atten- 
 tion (Plate 30A). The inscription has been interpreted by Sir Thomas 
 Arnold as meaning ** Jaushaqan work 117 ** — apparently locating the 
 carpet as Jushagan work of the year 1170 of the Muhammadan Hegira 
 (=A.D. 1757). The pattern is in red and white. The three elaborated 
 fleurs-de-lys betray European influence — perhaps Florentine rather than 
 French. It is in all respects a singular carpet. 
 
 36 
 
Chapter III 
 
 INDIA 
 
 Abul Fazl, the devoted admirer and servant of the great Mogul 
 Emperor of India, Akbar (1556-1605), and the author of the Institutes 
 of Akbar, in which the events of his reign are set down, writes as follows : — 
 
 ** His Majesty has caused carpets to be made of wonderful varieties 
 and charming textures ; he has appointed experienced workmen, who 
 have produced many master-pieces. The carpets of Iran and Turan 
 are no more thought of, although merchants still import carpets from 
 Goskhan, Khu2;istan, Kirman and Sab2rwar. All kinds of carpet-weavers 
 have settled here, and drive a^ flourishing trade. These are found in 
 every town, but especially in Agrah, Fathpur and Labor.'' ^ 
 
 In this account we have the record of the real beginning of carpet- 
 knotting in India. It is not difficult to suggest a reason why India was 
 late in the field. Pile- carpets were devised to meet the needs of colder 
 climates than that of India, and in such climates suitable wool for making 
 them can be more easily grown. Before Akbar's time the floor-coverings 
 in India would have been for the most part the cotton daris or tapestry- 
 mats such as are still made and used. The implication in Abul Fail's 
 chronicle that before Akbar's reign Persia and Turkestan Iran and 
 Turan ") provided such pile-carpets as were required to satisfy the 
 demands of wealth and luxury is confirmed from other sources. Even 
 when the new Indian factories were started, their output did not meet 
 all needs. Persia still continued to send supplies. Abul Fa2;l says they 
 came from Jushagan (near Ispahan), Khuzistan (S.W. Persia), Kirman 
 (S.E. Persia) and Sab2;war (in Khorassan, N.E. Persia), places where 
 carpets have continued to be woven down to the present day. 
 
 On the other hand, there was a small export trade in Indian carpets 
 which began, if not in the reign of Akbar himself, at any rate in that of 
 his successor. In the latter half of the seventeenth century the number 
 sent out of the country must have been considerable. 
 
 ^ The Ain i Akbari by Abul Fazl ‘Allahmi, translated by H. Blochmann, Calcutta, 
 1873, Vol. I, p. 55. He goes on to give particulars of prices and sizes in the Imperial Work- 
 shops. The writer was born in 1551, and died in 1602. 
 
 37 
 
HAND-WOVEN CARPETS 
 
 In spite of the labours of such competent authorities as Colonel 
 Hendley^ Vincent Robinson^ Sir C. Purdon Clarke ^ and others^ the subject 
 of Indian carpets still offers a field for study and investigation. Besides 
 the valuable, but somewhat equivocal, witness of Akbar's chronicler, there 
 is the account of Sir Thomas Roe, the first English ambassador to the 
 Mogul Court, who arrived in India in 1612, when Akbar^s successor Jahangir 
 was on the throne. He states that he saw Persian carpets spread before 
 the Emperor on the festival of the New Year, and he records a promise 
 of Jahangir to send Persian carpets to England. That the ambassador 
 took pains to discriminate between Persian and Indian carpets may be 
 doubted.^ 
 
 So far as actually existing specimens of the seventeenth century are 
 concerned we have chiefly those in the palace of the Maharaja at Jaipur, 
 so fully described by Col. Hendley ; a few similar carpets noted as being 
 in mosques at Berhampur and Bijapur, besides some at Ahmedabad ; 
 the Girdlers^ Company's carpet ; the Fremlin " carpet ; and a number 
 of others scattered about the world which are ascribed, on one plea or 
 another, to the factories started by Akbar. 
 
 The most characteristic of the carpets at Jaipur seem to be certainly 
 Indian, though not of Akbar's time. The design of these represents 
 rows of flowering plants each delineated separately and entire to the roots 
 as if planted in a garden, or else set in the interstices of a trellis as though 
 climbing. They were brought to the Treasury at Jaipur from the old 
 Palace at Amber near by, about the year 1875. Col. Hendley, who examined 
 the old tickets upon them, points out that some are described as Lahore 
 carpets, and that the dates, where given, are not earlier than the middle 
 of the seventeenth century, though it is not clear whether such dates are 
 those of manufacture or of the inventories. The shapes of some are 
 unusual. One is circular and others have a polygonal inset contour on 
 one side. These have been ascertained to have been specially made 
 for an apartment of the palace at Amber built in 1630. In view of this 
 evidence, it seems quite safe to assign the carpets of this class at Jaipur 
 
 ^ Colonel T. H, Hendley, C.I.E,, Asian Carpets : Sixteenth and Seventeenth Century 
 Designs jrom the Jaipur Palaces, London, 1905. V. Robinson, Eastern Carpets, London, 
 1882. Oriental Carpets, Vienna, 1892-4. F. R. Martin, Oriental Carpets, Vienna, 1908. 
 
 2 The language of these days took little heed of geographical precision in matters of 
 the kind. The minutes of the Girdlers’ Company, of 1634, describes the carpet then given 
 to them by Sir Robert Bell as a Turkey carpet, though it was known to have come from 
 India. Later in the century, the terms Chinese, Japanese and Indian are applied almost 
 indiscriminately to artistic imports from the East. 
 
 38 
 
INDIA 
 
 to the Indian looms in the seventeenth century. Two of the carpets 
 from Jaipur^ with flowers on a red ground^ have recently been brought 
 to England, They are reproduced on Plate 31, The further claim which 
 has been put form^ard that all existing carpets showing such patterns 
 are Indian seems more open to question,^ If that be so^ then very trouble- 
 some points must arise in regard to velvets^ brocades and other works of 
 art with plant designs treated in the same way. In fact, these motives 
 were long popular with Persian craftsmen. They may even have been 
 carried into India by the Persian weavers who, as all authorities agree, 
 worked in Akbar's factories, 
 
 A fine carpet showing plant-forms similar to those on the carpets 
 from Amber, and done in similar colours, is in the Salting Collection, 
 Victoria and Albert Museum (Plate 32), It will be seen that the plants 
 are each in a separate compartment. This example is incomplete. An- 
 other in the Diisseldorf Museum is probably part of the same carpet. 
 It has a broad border of floral forms resembling the middle pattern. 
 
 Another class of carpets in the Jaipur treasury shows palmettes and 
 floral stems disposed in a somewhat formal manner on a red ground, 
 with a wide border of palmettes and leaves on dark blue or green. Some- 
 times there are hunting-animals in the middle field, and occasionally 
 Chinese phoenixes and dragons. Such designs are obviously modifications 
 of the finer and freer patterns of Persia in the sixteenth century. The 
 scheme of colouring often has a quality of its own in warmth and depth ; 
 but are we to say that these phenomena were peculiar to India s' Until 
 a score of years ago, these carpets were all attributed to Herat, and 
 the attribution to that celebrated centre of Persian carpet-weaving must 
 still be allowed to have been often correct, 
 
 Olearius, who went to Persia with the embassy of the Duke of Hol- 
 stein- Gottorp about 1639, says that in the city of Herat the most hand- 
 some carpets of Persia were then made,^ At the same time, it is a plausible 
 theory that if any of the carpets actually made under Akbar exist to- 
 day, some of this class are among them, Herat was favourably situated 
 for transporting either its carpet-weavers or their products to Akbar's 
 Court. 
 
 The most completely authenticated of all Indian carpets has already 
 been briefly referred to — that in the possession of the Girdlers" Company 
 
 ^ F. R. Martin, Oriental Carpets, p. 89. 
 
 2 Adami Olearii, Reisebeschreibung nach Muskow und Persien, Hamburg, 1696. Quoted 
 by F. R. Martin, O.C., p. 69. 
 
 39 
 
HAND-WOVEN CARPETS 
 
 of London (Plate 33), The elucidation of the history of this carpet 
 a few years ago threw a new light on the problem of Indian carpet-weaving. 
 It measures about 8 yards in length by 2,^ yards in width. The coat- 
 of-arms and motto of the Company are represented in the middle. At 
 each end is a shield azure, an eagle displayed argent, in chief three fleurs- 
 de-lys or. Between each of these and the central coat is seen a bale of 
 goods with the monogram R.B. and a merchant's mark. The researches 
 in the books of the Company and in the records of the India Office recovered 
 all material facts. The arms at the ends were identified as those borne 
 by Robert Bell, master of theGirdlers' Company in 1634, and a director 
 of the East India Company. The records of the India Office showed 
 that the carpet was made for him in the royal factory at Lahore, and 
 those of the Company recorded the gift by him in the same year.^ The 
 middle of the carpet is red, the wide border being in dark blue. The 
 identification of the Girdlers' carpet helped towards that of another 
 carpet which left England for America some years ago. A coloured 
 illustration in Vincent Robinson's Eastern Carpets preserves a record 
 of the design (Plate 34). The middle has a pattern somewhat like that 
 of the Girdlers' carpet, but with the addition of hunting animals among 
 the stems. The wide border is blue, with the arms of the Kentish family 
 of Fremlin at intervals with palmettes and stems between.^ The points 
 of resemblance between this carpet and the Girdlers' prompted a reference 
 to the published literature of the East India Company, where the name 
 was found to occur several times. The conclusion that a member of 
 this family arranged for the carpet to be made in India through the agents 
 of the company seems fairly safe. 
 
 There is little else but conjecture to aid in carrying farther forward 
 the record of the factories in India founded under Akbar. An interesting 
 fragment in the collection of Dr. Friedrich Sarre of Berlin is shown to be 
 Indian by the pattern of two fighting elephants ridden by their mahouts.^ 
 Two other fragments — one in the possession of Dr. Roden of Frankfurt, 
 and the other formerly in the collection of M. Jeuniette, and now in the 
 
 ^ Information collected and published by the Company. See also Art Workers' Quar- 
 terly, Vol. Ill, July, 1904, p. 97 ; T. H. Hendley, Asian Carpets, 1905, p. 8. 
 
 2 Or Framlingham, of Hartlip, Kent. Gules, a chevron between three close helmets 
 argent plumed or. Crest, an elephant or, armed gules, gorged with a chaplet vert. The arms 
 were identified by Mr. Van de Put, and first published in The Times special Textile number, 
 June, 1913. In V. Robinson’s book (London, 1882, PI. 9) the carpet is attributed to the 
 weavers of Alcaraz in Spain. 
 
 ^ Illustrated by Martin, 0 .( 7 ., Fig. 235. His attribution of the carpet to the sixteenth 
 century seems rather too early. 
 
 40 
 
INDIA 
 
 Louvre, have fantastic patterns of heads of elephants, lions, oxen and 
 horses with diminutive serpentine bodies, and floral sprays interspersed. 
 They are probably of the seventeenth century, contemporary with Dr. 
 Sarre^s piece. A singular carpet in the Museum of Fine Arts at Boston 
 in America has, in a medley of motives, a hunting-cheetah driven to the 
 chase on a bullock wagon covered with a brightly-coloured textile, to which 
 it is chained.^ This recalls Abul Fail's account of Akbar's hunting 
 expeditions, where he says that the hunting leopards got Jushagan carpets 
 to sit on, and even had carts made for them.^ The French traveller 
 Francois Bernier, who visited India in 1656-68, refers to the practice 
 of keeping the hunting-leopards chained on a small car. 
 
 The importation of carpets into England through the East India 
 Company appears to have been considerable in the seventeenth century. 
 Great families, notably that of the Duke of Buccleuch, whose ancestor 
 the Duke of Montagu was a director of the Company, still possess fine 
 specimens. They are mostly of the ** palmette '' or Herati type. 
 Whether any were brought into India from Persia it is difficult to say, 
 but it is to be remembered that the Company traded in Persian goods, 
 thus supplanting the more difficult route of communication with Persia 
 established by the English Mucovy Company by way of Archangel in 
 the reign of Queen Elizabeth. 
 
 The weaving of pile- carpets in India appears never to have quite 
 ceased from the time of Akbar onwards. 
 
 The types of the nineteenth century are varied. On the one hand 
 there are the sumptuous silk carpets of Warangul, with floral patterns, made 
 about the middle of the century ; on the other, the exceedingly coarse 
 wool carpets, with animals, birds and floral forms, of Tanjore and Malabar. 
 A carpet made at Tanjore in the Madras Presidency is illustrated in 
 Plate 35. Its principal interest lies in the fact that the design is typically 
 Indian in character, without Persian influence. The compartments 
 enclosing the floral sprays are red, buff and pale blue. The border is 
 white. The knotting is coarse and the colours are dull. 
 
 Soon after the middle of the century, carpet-knotting was begun 
 in the jails, with the object of providing useful and remunerative 
 employment for those detained there. This object was to some degree 
 attained. The scheme had another aspect — whether it tended to raise 
 or depress the standard of the industry in India. That question need 
 
 ^ Martin, O.C., Fig. 234. 
 
 ^ Blochmann, Ain i Akbari, I, p. 287. 
 
 41 
 
HAND-WOVEN CARPETS 
 
 not be discussed here. Most of the jail-carpets ** were 
 old patterns. 
 
 The cotton pile-carpets made in several districts of 
 latter half of the last century are, it is to be hoped, unique. 
 
 copied from 
 India in the 
 
 42 
 
Chapter IV 
 
 TURKEY 
 
 Beyond all question the best carpets in the world are the Persian^ 
 whether the point of view taken be that of design, colour, or craftsmanship* 
 Little imagination is needed to appreciate the skill demanded of the 
 weaver in keeping a great carpet like that from Ardabil straight and 
 even in the making, and to give it a texture which should outlast five 
 hundred years of use* Such mastery of the technicalities of the craft 
 is often combined, as in that carpet, with resourcefulness in design and 
 true genius for colour* Every other carpet-weaving country has felt 
 the influence of Persia* In Asia Minor that influence prevailed to such 
 a degree that it almost becomes a question whether the best carpets made 
 there should properly be called Turkish* 
 
 This designation is not altogether satisfactory for the art of Asia 
 Minor, but there is no better comprehensive term. Traces of the native 
 Turkish art of their old home in Central Asia are rare, although they 
 are sometimes to be found in nomad carpets. The art was the outcome 
 of a more advanced civilization. In the course of history, successive 
 waves of humanity have been pressed into the narrow limits of Asia Minor 
 after a fashion which may be compared, in another element, with those 
 bays into which the waters of the ocean rush from time to time in a tidal 
 wave. An enumeration of the races represented in Asia Minor includes 
 Turks, Greeks, Jews, Persians, Turcomans, Tartars, Yuruks, Circassians, 
 Armenians, Nestorians, Syrians, Kurds, Cossacks, Bulgarians, Arabs 
 and gypsies. Not all these people are carpet -weavers, but several of 
 them are. The pressure of the Mongolian horsemen which set the 
 Seljuks moving westward, and after them the Ottoman Turks, drove 
 Persian craftsmen and poets to seek the same asylum, and the latter 
 brought with them a proficiency in the arts of which the Turks were 
 not slow to take advantage. In 1514 the conquest of Tabriz was effected 
 by the Turkish Sultan Selim I, and that city was again entered by his 
 successor Suleiman I twenty years later. Craftsmen were carried away 
 into Asia Minor by the conquerors. Some, indeed, have attributed 
 
 43 
 
HAND-WOVEN CARPETS 
 
 the brilliant height attained by Turkish art in the sixteenth century to 
 that event. 
 
 The Turk was a Sunni — an orthodox and literal observer of the 
 precepts in the sacred writings. He could not therefore permit the 
 representation of figures of men and animals, which the sectarian Persian 
 allowed. But for this, the most accomplished of the carpets of Asia 
 Minor might be regarded as an offshoot of Persian art. The mixture 
 of races in Asia Minor finds expression in a variety of types often having 
 little relation one to another. The ** Turkey '' carpets used in Western 
 Europe during the last five centuries are the work of the uplands of 
 Anatolia behind the port of Smyrna, whence so many were shipped. 
 For a long time these were practically the only pile-carpets brought to 
 Europe. The relative inaccessibility of Persia, and the almost insuperable 
 difficulties of transport rendered the Persian carpets practically unknown 
 until the seventeenth century, and even after they began to be imported 
 they were called ** Turkey ” carpets like the rest. 
 
 The seafaring communities of Italy had trading settlements dotted 
 along the coast of Asia Minor as early as the thirteenth century. With 
 the abandonment of these posts under pressure from the Ottoman Turk 
 during the course of the fifteenth century, commercial activity was not 
 put an end to. A lively trade still went on between the Venetians and 
 the Turks. The Italian traders also had their agents in the countries 
 of Western Europe, and by such means the carpets were obtained which 
 are seen represented in early pictures. The carpets of Asia Minor first 
 became known in the West in the fourteenth century, and in the two 
 following centuries they were imported in large numbers. Their strange 
 exotic beauty rendered them popular at once with the painters, who 
 obviously copied them with much fidelity. These pictorial representations 
 of the fourteenth, fifteenth and sixteenth centuries have greatly simplified 
 the task of classification. 
 
 The first essay on this subject was contributed by Dr. Julius Lessing 
 of Berlin. His book,^ published at Berlin in 1877, reproduces a series 
 of illustrations in colour from carpets represented in paintings by, or 
 attributed to, Jan van Eyck, Memlinc, Van der Goes, Mabuse, Holbein, 
 Ghirlandajo, Pinturicchio, Montagna, Moroni, Girolamo dai Libri and 
 others. 
 
 Most later writers on Oriental carpets have either made use of Lessing^s 
 work or made further comparisons for themselves. The subject requires 
 ^ Alt Orientalische Tep'pichmuster nach Bildern . . . des XV-XVI Jahrhundert. 
 
 44 
 
TURKEY 
 
 careful handlings for Oriental motives often have a long life, and similarities 
 between representations in old pictures and actual carpets obviously of 
 far more modern times are occasionally quite startling. Even these 
 cases afford useful evidence, for they tend to confirm the conclusions 
 made on various grounds that the old painters were remarkably true to 
 their models, and that the origin of designs still in use must be traced 
 to a remote past. An example is afforded among the illustrations in the 
 present volume. The painting at Vienna by Hans Memlinc (Plate 36) 
 shows a carpet with a central pattern almost identical with that of the 
 carpet reproduced on the same plate — yet the picture is of the fifteenth 
 century and the carpet is of the nineteenth.^ Relatively few carpets of 
 the time of the paintings are now to be seen in the land of their origin, 
 although some of the greater mosques contain important specimens. 
 On the other hand, it is not beyond the bounds of probability that a few 
 carpets depicted by the European artists mentioned above are actually 
 in existence to-day. Considerable numbers found their way into the 
 churches of Italy, where infrequent use has guarded them from destruction. 
 Many have been sold during the last forty years, and are now to be found 
 in the museums and private collections of Europe and America. 
 
 The earliest Turkish carpets we know of are in the mosque of 
 Ala- ed- Din at Konia (the ancient Iconium). That city was the capital 
 of the Seljuk Sultans of Rum, and Dr. Martin, who was the first to detect 
 their significance, attributes the carpets to the reign of the sultan whose 
 name the mosque bears (a.d. 1219-36). The question whether they 
 are as old as that may be debatable, but there need be no hesitation in 
 ascribing them to the times of the old Seljuk line at Konia, before it was 
 superseded in the fifteenth century by the Ottomans, whose capital was 
 far to the west at Brusa. 
 
 There are four carpets, all somewhat dilapidated, and two fragments. 
 The patterns are primitive diapers and interlacings of geometrical form, 
 and two of them have archaic Kufic lettering in the borders. The colours 
 are chiefly shades of blue and red, and yellow.^ 
 
 ^ A very similar carpet is reproduced in a French romance of the fifteenth century 
 at Vienna, Goeur d’ Amour epris, Cod. 2597. (See Kunst und Kunsthandwerk, V, 1902, 
 P- 307-) 
 
 2 Martin, 0 .( 7 ., PI. 30. 
 
 Their resemblance to the heraldic carpets made in Spain in the fifteenth century is 
 quite striking (see p. 68). It seems likely that some of the motives of the Spanish carpets 
 were borrowed from the successors of these Konia carpets in Asia Minor which have since 
 disappeared entirely. See also article by F. Sarre in Kunst und Kunsthandwerk, Vol. X, 
 1907, p. 503. 
 
 45 
 
HAND-WOVEN CARPETS 
 
 A carpet preserving a tradition no less remote is here illustrated 
 (Plate 37A). In its present incomplete state it shows two octagonal 
 compartments, each filled by an angular and disguised, though 
 unmistakable, rendering of a Chinese dragon and phoenix in red and 
 blue on a yellow ground* The pattern is obviously archaic, and the 
 difficulty of assigning a date to the carpet itself is considerable. Dr. 
 von Bode, who obtained this carpet many years ago in Rome, has pointed 
 out the remarkable similarity to the pattern of a carpet represented in 
 Domenico di Bartolo's fresco of the Marriage of the Foundlings,'^ 
 painted between the years 1440 and 1444 in the Hospital of Sta Maria 
 della Scala at Siena. Animal forms, even thus disguised, are rare in 
 Turkish carpets. Restrictions in regard to such representations were 
 not so rigidly observed under the Seljuks as under the later Ottomans, 
 and it is probable that the tradition of this carpet goes back to Seljuk 
 times. There is no difficulty in accounting for such Chinese motives. 
 They found their way to Asia Minor in the first instance through the 
 Mongolian conquests, but they were reinforced afterwards by the 
 immigrant Persian craftsmen. However early in origin the patterns 
 may be, few existing carpets from Asia Minor were actually made before 
 the sixteenth century. The finest in texture and the most intricate in 
 design of all Turkish carpets were made during that time. Like some 
 of the contemporary woollen pile-carpets of Persia, they are generally 
 woven on silk warps to secure a closer texture, although at times a fine 
 woollen warp is used. They show such uniformity in design and colouring 
 that it seems almost beyond doubt that all were the work of a single locality. 
 Some of them are of large si2;e. It has been suggested that the factory 
 was an Imperial one, and this seems likely, for the carpets must have 
 been too costly for the general markets.^ The location is not known ; 
 perhaps it was in the neighbourhood of Brusa, the Asiatic capital. The 
 best work was done in the sixteenth century, although some examples 
 appear to belong rather to the seventeenth. The patterns are floral 
 and very gracefully drawn, with roses, carnations, tulips, irises and 
 hyacinths, and long curving leaves, rendered in the same manner as on 
 the exquisite painted pottery of the time (see Plate 38). The colouring 
 of the ground is almost always a bright crimson of a characteristic tone, 
 though blue is occasionally used. There is often a central panel and 
 
 ^ A later Imperial factory was situated at Hereke on the eastern shore of the Sea of 
 Marmora, at a short distance both from Constantinople and Brusa. A modern Hereke 
 carpet is reproduced in Plate 97. 
 
 46 
 
TURKEY 
 
 spandrels in dark or light blue, green, yellow or white, or border-panels 
 in the same colours. Floral ornament, contrasted with the rest of the 
 design by its formal character, and arabesques generally fill these panels. 
 A carpet, unfortunately not complete, in the Victoria and Albert Museum 
 (Plate 39) has the crimson ground with a typical pattern. 
 
 Another carpet which falls within the same group is illustrated in 
 Plate 40. The distribution of the floral stems here shown is unusual 
 for a carpet, although parallels may be found in the painted tilework. 
 The ground is again red. The salience of the small white flowers is 
 due to the use of cotton instead of wool for those parts, giving a dead 
 white in place of the softer tones of the wool. 
 
 Several examples are illustrated in colours in the great Austrian book. 
 Two are in the Museum of Art and Industry at Vienna, one in the museum 
 at Leipzig, and a fourth in the Musee des Arts Decoratifs in Paris.^ 
 Another in the last-named collection is on a blue ground, which is very 
 exceptional.^ 
 
 A few prayer-carpets of this class are known. One, in the Austrian 
 Imperial Collection, was shown at Vienna in 1891, and again at Munich 
 in 1910. Two more were shown in the latter exhibition. The first 
 belongs to the Kunstgewerbe Museum, Berlin, and the second to Prince 
 Liechtenstein.® It is a point which may not be entirely without 
 significance that two of these prayer- carpets have borders from which 
 those most frequently seen on the later prayer- carpets of Ghiordes (see 
 Plate 53) are obviously derived. Thus the possibility of an origin in the 
 same locality is suggested. 
 
 A different type of carpet is illustrated in Plate 41. These contrast 
 so forcibly with the kind just described that it seems almost beyond belief 
 that both should have been made in the same part of the world at the 
 same time. The knotting is coarser and of another kind ; the colouring 
 is simpler and quite different in tone. The designs are stiff and formal 
 where the others are free and natural, and the motives are geometrical 
 rather than floral.^ Persian influence is strong in the first group. In 
 
 ^ Vienna, O.C., Pis. 40, 85, 15 and 68. See also Martin, 0 C., Figs. 328-30. 
 
 2 Reproduced in colours in Ancient Oriental Carpets, Leipzig, 1906-8, PI. 25. See also 
 Martin, O.C., Fig. 327. 
 
 ^ Oriental Carpets, PI. 14 ; Meisterwerke, Munich, Pis. 74, 75 ; Martin, Fig. 331. 
 
 ^ Dr. F. Sarre, in an illuminating paper on early Turkish carpets (Kunst und Kunst- 
 handwerk, X, 1907, p. 513) suggests that the border of the first example on Plate 41 is a 
 modified Kufic inscription. It appears rather to be a counterchange cresting like Plates 
 ii8a and 141. 
 
 47 
 
HAND-WOVEN CARPETS 
 
 the latter the type of design is Turkish, and in modified forms its continuity 
 is traceable down to the present day. Large numbers must have been 
 exported to the West, The two in question were both obtained in Italy, 
 and such carpets are not now found in Asia Minor, They are often to 
 be seen in pictures of the Italian, Flemish and German schools in the 
 fifteenth and sixteenth centuries, Ghirlandajo, Memlinc, and Holbein, 
 among others, certainly had access to carpets of the kind,^ There is 
 almost a suggestion of primitive austerity in these large and uncom- 
 promising geometrical forms, and there can be no doubt that they 
 are descended from an early type. Another geometrical type popular 
 with these painters is reproduced on Plate 42, in a carpet belonging to 
 Mr, George Mounsey, The interlaced border associates this carpet 
 also with the well-known rugs showing arabesques in the middle field 
 (see Plate 43B), The carpet illustrated in Plate 37B represents a step 
 towards greater freedom of design. Small shaped panels in dark blue 
 on a red ground are distributed over the surface. 
 
 The ** Oushak ” carpet, the real forerunner of the ** Turkey carpet ** 
 of to-day, was much in demand in Western Europe in the sixteenth and 
 seventeenth centuries. The name is derived from a town situated in 
 the middle of the carpet-weaving districts of the Anatolian uplands some 
 distance due east of Smyrna, These carpets are generally of convenient 
 si^e for transport, perhaps 6 ft, wide and twice as long, though far larger 
 examples are sometimes met with. The colours are generally brilliant 
 and pure but not in great variety, A succession of large panels occupies 
 the middle space, either in blue on red or red on blue ; sometimes there 
 are two shades of blue. One well-known form of design is illustrated 
 on Plate 43A, The large star-shaped panels are dark blue and the ground 
 is red. Carpets of this type are to be found in pictures, A well-known 
 example is in a painting by Paris Bordone, in the Academy at Venice, 
 representing the Fisherman restoring the lost ring to the Doge, King 
 Henry VIII stands on another carpet of similar style, in a portrait at 
 Belvoir Castle,^ Sometimes large pointed-ogee panels are substituted 
 for the star-forms ; they are generally placed close together, with half- 
 
 ^ A very good example is Holbein’s great painting of the “ Amabassadors ” in the 
 National Gallery, London ; another is seen in a painting by Memlinc in the same gallery. 
 Several patterns of the type are illustrated in J. Lessing’s Alt Orientalische Teppichmuster 
 nach Bildern , . . des XV-XV IJahrh., Berlin, 1877 ; see especially Pis. i, 19, 26 (Memlinc); 
 9, II (Holbein) ; 13, 14 (Ghirlandajo) ; 21 (Anguissola). 
 
 2 Connoisseur, Vol. VI, 1903, p. 67. This shows that such carpets reached England 
 before the middle of the sixteenth century. 
 
 48 
 
TURKEY 
 
 panels at the sides. An example in the Victoria and Albert Museum 
 is reproduced (Plate 44),^ 
 
 In the sixteenth and early part of the seventeenth century there 
 was a large output of these Oushak carpets. At the same time, 
 smaller rugs with a class of pattern of their own, were carried to Italy in 
 considerable numbers. These, like the Oushak carpets, were probably 
 made principally for export. There are two chief kinds of design — 
 one in which the middle is covered with repeating arabesques (Plate 
 43B), and the other with a plain middle space relieved only with a small 
 central ornament and spandrels in the corners. The colour of the ground 
 is generally brick-red, sometimes blue. Yellow enters largely into the 
 pattern as a rule, relieved by blue and white, Chinese cloud-bands are 
 not unfrequently used in the borders, repeated at close intervals, and 
 occasionally in the middle. The most usual border is the conventionalized 
 Kufic lettering, generally interlaced. The arabesques are drawn in an 
 angular manner and closely packed to cover the surface. The Kufic 
 border is found so frequently in the Italian, Flemish and German paintings 
 of the fifteenth and sixteenth centuries that there is no need to single 
 out examples. The arabesques, too, are often seen — more especially 
 in the paintings of the Venetian Lorenzo Lotto, The Kufic border 
 continued to be used until the eighteenth and even the nineteenth century 
 with little change (see Plates 63 and 64), 
 
 A small carpet in the Victoria and Albert Museum, obtained from 
 a church in Italy, has the same pattern of floral ornament and cloud-bands 
 carried through the middle and the border alike. The ground of the 
 middle is blue, and of the border red. 
 
 The demand for these carpets by the Venetians and other seafaring 
 traders calling at Smyrna was considerable. Evidences of concessions to 
 Western taste may be seen not only in the subdued colouring, but also 
 in the general uniformity of shape and size and to some extent in the 
 design, A carpet lent by Frau Limburger to the Munich Exhibition of 
 1910 had the usual yellow and blue arabesques covering the red ground 
 everywhere except in the top left-hand corner, where there was a shield- 
 of-arms of the Genoese families of Centurione and Doria impaled,^ 
 
 ^ Other examples, mostly in private possession, are illustrated in Vienna, O.G., PI. 17 ; 
 Martin, O.C., Figs. 315, 317-20 ; Meisterwerke, Munich, Pis. 70, 71. There is a carpet 
 with the star-shaped panels in the Victoria and Albert Museum (No. 138), and two with 
 ogee-shaped panels (Nos. 135 and 136). 
 
 2 Meisterwerke, Munich, PI. 72. The arms have been identified by Mr. A. Van de Put. 
 Cardinal Wolsey’s insistent demand for these carpets is referred to later (p. 76). 
 
 49 
 
 E 
 
HAND-WOVEN CARPETS 
 
 The rugs with the space of plain colour in the middle convey more 
 forcibly still the impression of having been made to suit the taste of the 
 European customer. The eye of the Oriental is not so partial to those mono- 
 tonous spaces^ which might have been relieved by a simple pattern with very 
 little extra trouble. We have in these rugs the first signs of an adaptation, 
 as early as the sixteenth century, to the exigencies of the foreign market 
 for which they were destined. In later days those influences were so 
 potent that the survival to the present day of the beautiful craft of carpet- 
 weaving as a native industry is almost to be wondered at. Yet it would 
 be hard to point out any great branch of the artistic handicrafts now 
 practised in the East with better results. It is true that we cannot hope 
 to see carpets made again to rival the great work of the past, any more 
 than we can look for a revival of the conditions under which it was 
 produced. The royal patron, with the resources of the whole kingdom 
 at his command, has given way to the commercial trader. It is due to 
 the latter to say that he has done much to keep the industry alive, but 
 whether modern conditions are favourable to the development of the 
 highest qualities in design and craftsmanship is another matter. At 
 any rate it may be justly claimed that surprisingly beautiful modern 
 carpets may still be bought at a trifling cost. 
 
 Other types of carpets made in Asia Minor are contemporary with 
 the small rugs just described. Somewhere about the end of the sixteenth 
 century a white ground was used. The choice of colour may again have 
 been a concession to Western taste. White grounds are occasionally 
 found in old Persian carpets (e.g. Plate i8), but they do not cover the 
 whole carpet, border and all ; and where used they are more broken up 
 by close patterns. The colours are chiefly red, blue and yellow. The 
 carpets illustrated on Plates 45 and 46A are of this kind. It will be noticed 
 that the former has the cloud-band repeated at intervals in the border. 
 Various explanations have been given of the nondescript forms in the 
 middle, but they are nothing more than angular renderings of the S -shapes 
 often found in Persian design. The other carpet has in the middle 
 repetitions of the double stripe and the group of three discs below (see 
 Plate 205), which appears so often in the art of Asia Minor in the sixteenth 
 century. The double stripes may have originated either in cloud-forms 
 or in the fur-markings of animals. The three discs appear to have been 
 devised, or at any rate adopted as a badge by the great Mongol conqueror 
 Timur in the latter part of the fourteenth century. 
 
 When Ruy Gonzales de Clavijo, who went with an embassy from 
 
 50 
 
TURKEY 
 
 Castile to the Court of Timur at Samarcand in 1403^ arrived at the town 
 of Kesh, 36 miles from Samarcand^ he was shown a palace which had 
 been many years in buildings and was not then finished. Over a doorway 
 he noticed the figure of a lion and a sun^ which are the arms of the 
 lord of Samarcand ; and, though they say that Timour Beg ordered these 
 palaces to be built, I believe that the former lord of Samarcand gave 
 the order ; because the sun and lion, which are here represented, are 
 the arms of the lords of Samarcand ; and those which Timour Beg bears 
 are three circles like O's, drawn in this manner and this is to signify 
 that he is lord of the three parts of the world. He ordered this device 
 to be stamped on the coins, and on everything he had.^^ 
 
 Later the ambassador mentions that the judges in the city of 
 Samarcand use ** the seal of the lord, having three marks upon it, like 
 this, 
 
 Timur penetrated into Asia Minor, but it is not necessary to connect 
 the appearance of his device there with his invasion. The use of the 
 three discs was rendered so widespread by his orders that in course of 
 time they must have become merely a trite motive of ornamentation. 
 
 Another kind of pattern, an arrangement of interlacings in star-form, 
 generally surrounded by diminutive trees and floral stems, placed in 
 radiating fashion, sometimes covers the middle of the carpet. These 
 carpets seem to form a link between those of the mosque at Konia and 
 the Anatolian examples described above. There is something in them 
 which suggests North Africa or even an approximation to the carpets 
 of Spain, but they may be Asia Minor work of the fifteenth to the 
 seventeenth centuries (Plate 47). A more intricate design, with similar 
 tendencies, is of equally uncertain origin (Plate 48). Three specimens 
 of a class of carpets not unfrequently seen, both in actual examples and 
 in pictures, are reproduced in this volume (Plates 49-51). Carpets 
 of this group are singularly uniform in design. There can be little doubt 
 that they were made in a single district of Asia Minor, though it is not 
 known where. They have acquired the name of Transylvanian carpets 
 from the accidental circumstance that considerable numbers have been 
 found in that province, which lies at the south-eastern corner of the 
 old Austro-Hungarian lands, separated by the Carpathian Mountains 
 
 1 Narrative of the Embassy of Buy Gonzalez de Clavijo to the Court of Timovr, at Samar- 
 cand, A.D. 1403-6, Translated by Clements R. Markham (London, Hakluyt Society, 1859, 
 pp, 124, 175), Dr. F. R. Martin first drew attention to this passage in connexion with 
 the appearance of the device on works of art. 
 
 51 
 
HAND-WOVEN CARPETS 
 
 from Roumania* While the latter country remained annexed to the 
 Turkish Empire these carpets were imported by way of Kronstadt, thus 
 making Transylvania a centre for their distribution farther afieldd The 
 earlier type is represented by Plate 49. This carpet resembles closely 
 a specimen in the Victoria and Albert Museum ; the chief difference is 
 that the spandrels of the latter carpet are filled with a kind of arabesque 
 design instead of the floral motives, indicating an earlier date. Examples 
 must have reached the Low Countries by the seventeenth century, for 
 they are to be found faithfully reproduced in pictures. There is a good 
 example in the London National Gallery, a painting of two men in a 
 room (called A Merchant and His Clerk by Thomas de Keyser.^ 
 The colouring of the earlier specimens is subdued and varied, the 
 central field being usually of a deep purple tone, and the border panels 
 in four or five different colours. The relatively numerous examples 
 extant afford a very good opportunity of tracing first their Persian 
 derivation, then the gradual appearance of Turkish colouring and motives, 
 and finally the complete subordination of the original design. For this 
 reason it is proposed to discuss these examples more fully. If the four 
 spandrels of the carpet reproduced in Plate 49 had been united into a 
 star, they would have formed quite a good central pattern, and this 
 treatment of the spandrels is more Persian than Anatolian in origin. The 
 method of treatment of the central field and of the spandrels, however, 
 shows a tendency to geometrical conventionali2;ation which is aided by 
 the coarse knotting employed in this type of carpet. It will be seen 
 that the middle space is occupied by two vases placed in the niches formed 
 by the spandrels and still provided with the chains which betray their 
 derivation from hanging lamps. The shape and filling of the narrow 
 border panels show quite clearly their relation to the well-known Persian 
 border pattern in which arabesques surround a flower. Specimens of 
 more recent date do not have the small star-panels in the border and 
 the narrow panels are immediately adjacent to each other. Later on 
 the central vase-pattern with the conventionalized floral stems disappears. 
 In its place we sometimes find a central ornament of a totally un-Persian 
 character, surrounded with flowers (Plate 50). In the spandrels appears 
 a geometric pattern which can scarcely be recognized as consisting of 
 
 ^ Several examples in the Victoria and Albert Museum actually came from Transylvania 
 about forty years ago. Dr. Martin states that the Transylvanians retained a percentage 
 of the carpets passing through, as a tax (O.C., p. 130). 
 
 2 Lessing reproduces the border of another in a portrait by Cornelis de Vos at the 
 Brussels Museum {Alt Or. Teppichmuster, PI. 29). 
 
 52 
 
TURKEY 
 
 flowers or leaves* After a lapse of only a few years the flowers in the 
 narrower panels of the border lose their natural character^ while the 
 arabesques become stiffened almost beyond recognition. In order to 
 leave no doubt that these carpets come from Anatolia^ we find sometimes 
 those conventionalized carnations springing from the spandrels and 
 reaching into the central panel which are so common in the Anatolian 
 prayer-carpets. 
 
 The last carpet of this type here illustrated (Plate 51) shows these 
 motives. The central panel reveals its connexion with its predecessors 
 (Plates 49, 50)^ and there are other traces of good tradition — but there 
 is hardly anything in this carpet to remind us of its original Persian 
 ancestry. 
 
 Prayer-rugs have already been briefly referred to. They were 
 nowhere so popular as in Asia Minor. 
 
 Every Muhammadan has to perform five ablutions daily, which 
 are followed by five prayers. Just as specially shaped vessels have to 
 be used for these ablutions, so have all particulars for the prayers been 
 laid down. The prayer-rug in the mosque is placed immediately in 
 front of the prayer-niche (Qiblah), which indicates the direction of Mecca, 
 towards which the faithful have to pray. For this reason the carpet 
 has a special design, which in itself is reminiscent of the niche, so that 
 the apex may be pointed towards Mecca. The true believer has to step 
 on to the farther end in order that, after bowing towards the angels on 
 the right and on the left, who record his good and his evil deeds 
 respectively, and murmuring several prayers, he may kneel down and, 
 while resting upon his hands, touch the earth, that is to say the carpet, 
 with his head. 
 
 The type from which the Anatolian prayer-carpets are derived is 
 a niche supported by one or two columns at each side, and provided with 
 a lamp of vase-form hanging from the apex, and perhaps having two 
 pricket-candlesticks below.^ Several of these constituent parts have 
 undergone radical change at the hands of the carpet designer, no doubt 
 due to successive copying through the centuries. If the earliest prayer- 
 carpets had come down to our times the original type would have been 
 shown more clearly. The niche is the most constant feature. The 
 lamp becomes a hanging vase of flowers with the chains still seen, though 
 the vase is inverted (see Plate 49). Sometimes a spray of flowers hangs 
 
 ^ See an early carved wood panel belonging to M. Gillot, in Exposition des Arts Musul- 
 mans, by G. Migeon, Paris, 1903, PI. I. 
 
 53 
 
HAND-WOVEN CARPETS 
 
 alone (Plates 53 and 54)* Quite often the lamp is replaced by an inverted 
 ewer with handle and spout, generally containing flowers (Plate 98). ^ 
 Occasionally it is transformed into a nondescript ornament (Plate 102), 
 and in a few instances a central panel is connected by a line with the 
 head of the niche, as though it too were derived from the hanging lamp. 
 The supporting columns remain as a rule in some form or another (Plate 
 55). At times they become ornamental stripes sprouting into floral 
 forms at top and bottom (Plate 54) ; or the niche head may even be 
 provided with the unsubstantial support of a mere succession of floral 
 forms (Plate 53), Floral sprays or stripes may also represent the pricket- 
 candlesticks, which are not found in an intelligible guise on existing 
 carpets. 
 
 The three best-known types of these prayer- carpets have been 
 associated with the localities of Ghiordes (Gordium i)t Kula and Ladik 
 (Laodicea), but this classification (for it cannot safely be regarded as a 
 precise statement of geographical fact) takes no account of some important 
 specimens of which the carpet reproduced in Plate 52 is an excellent 
 example. Here the spandrels are a deep red, and the niche itself a very 
 dark blue, but the colour reverts to red again below the large pale blue 
 cloud-band which in form almost repeats the niche-head, though this 
 is perhaps accidental.^ The carpet is Anatolian. In some respects it 
 resembles the small rugs imported into Italy in the fifteenth and sixteenth 
 centuries (see p. 44). It has been attributed to the end of the fifteenth 
 century by Bode, and to the middle of the sixteenth by Martin. Much 
 depends on whether the exaggerated angularity of the pattern is a sign 
 of primitive origin, or whether the impression that it may be a good 
 design breaking up, is altogether without foundation. The carpet is a 
 fine one, but the possibility of an origin in the seventeenth century cannot 
 safely be overlooked. The three classes of prayer-carpets referred to above 
 are mostly of the eighteenth and nineteenth centuries, though a few 
 examples are as early as the seventeenth. The most familiar type of 
 Ghiordes prayer-carpet is illustrated in Plate 98. 
 
 It shows a well-known border with palmettes and rosettes and 
 numerous little stripes of conventional ornament. The extra stripes 
 
 ^ A Ghiordes prayer-carpet belonging to the Duke of Buccleuch has in the niche a 
 lamp, with foliations springing from it, suspended by chains, and beneath that an inverted 
 ewer containing carnations. A Turkish prayer-carpet is usually woven upside down ; 
 this may account for the inversion. 
 
 2 There is an excellent coloured illustration in Vienna, O.C., PI. 3. See also W. von 
 Bode, V orderasiatische Kniipfteppiche, coloured frontispiece to 2nd edition. 
 
 54 
 
TURKEY 
 
 above the niche will also be noticed. The hanging lamp has become 
 an inverted ewer. A simpler and more unusual form of Ghiordes prayer- 
 carpet is shown on Plate 54. The carpet reproduced in colour on Plate 
 53 is again more elaborate. It is one of a few examples woven with 
 inscriptions. That in the two panels above the niche is as follows : 
 ** My lord^ my august Padishah — may you be joyful and happy unto 
 the days of the last judgement.'^ This inscription shows that the carpet 
 in question may have been intended as a present for a sultan or some 
 other dignitary. There is also an inscription within the niche^ as follows : 
 ** I come before thy throne heavily laden with sin and pray that my sins 
 and guilt may be forgiven me,** ^ It is in Turkish but contains a few 
 Persian words^ a circumstance which has no bearings however, on the 
 theory held by some that these carpets were actually made by Persian 
 workmen in Anatolia, because a number of Persian words, especially 
 religious terms, are used by the Turks. 
 
 An inscription in the open space of the prayer-niche is rare. The 
 suggestion of lion-masks in each alternate palmette of the border (to be 
 seen also on Plate 98) may be accidental. On the other hand, it will be 
 recalled that masks of lions are often found in the palmettes of Persian 
 carpets of the sixteenth and seventeenth centuries, or even attached 
 independently to the sterns.^ The same border is seen in the curious 
 prayer-carpet illustrated on Plate 56, although in that example the palmettes 
 are obviously quite floral in character. The elongated form of the niche- 
 head is unusual. This carpet must be classed with the Ghiordes group. 
 It belongs to Mr. George Mounsey. The hyacinth sprigs within the niche 
 may be compared with those on Mr. Mounsey's small mat illustrated on 
 Plate 46B. The mat is knotted in wools like a carpet, although the design 
 is evidently taken from the well-known silk-velvet divan covers of Asia 
 Minor. 
 
 The Kula prayer-carpets are generally inferior to those of Ghiordes. 
 The knotting as a rule is not so fine, the design shows a tendency towards 
 monotonous diapers, and the colour is paler and less pleasing, with a 
 predominance of yellow. The blue niche with the strings of blossoms 
 shown on Plate 100 is common in Kula carpets. The wide border is 
 filled with rows of carnation blossoms. It is more usual for the border 
 to be broken up into many narrow stripes ; more than a do^en may be 
 
 ^ Translated by Professor Dr. von Kaelitz. 
 
 2 See Pis. I, 10, 14 and 22 Two Ghiordes prayer-carpets are reproduced in colours 
 in Vienna, O.C., Pis. 57, 60. 
 
 55 
 
HAND-WOVEN CARPETS 
 
 counted in some instances. The carpet illustrated on Plate loi has 
 eleven^ and it is classified as a Kula^ although the little patches of landscape 
 within the niche do not properly belong to such carpets. 
 
 The difficulties which beset any local classification of the carpets 
 of Asia Minor are demonstrated in an example reproduced on Plate 57. 
 The most conspicuous motive of the pattern is the series of four small 
 landscapes within the niche, with a representation of a mosque in the 
 arch at the top. Such little plots of ground with cypress-trees and 
 buildings are generally supposed to have been intended to represent 
 a Muhammadan cemetery, and from the circumstance the name usually 
 applied to such carpets is derived. Apart from this feature, the design 
 has ingredients both of the Ghiordes and Kula types. The border, 
 with its row of lion-mask '' palmettes and flowers alternating, resembles 
 those of the Ghiordes carpets just described (Plates 52 and 56). The 
 narrow stripes outlining the border on each side, and the filling of the 
 space above the niche, are typical of Kula (see Plate 100). The carpet 
 is Anatolian, and it is unsafe to specify the actual locality of production 
 more closely. The same may be said of another carpet belonging to 
 Mr. George Mounsey, in which the landscape motive is again rendered 
 (Plate 58). The ground is bright red, with rows of houses and trees 
 in strongly contrasting colours. 
 
 The last of the three principal types of these later Anatolian prayer- 
 carpets is the ** Ladik.'^ A good example is seen on Plate 102. A row 
 of long inverted stems ending in blossoms is generally to be found below 
 the niche in this group. The device filling the niche, with its four ewers 
 precariously balanced, has lost all meaning. A yellow panel under the 
 central arch bears the date 1211 of the Hegira, corresponding to a.d. 1797.^ 
 The fantastic forms of this type of carpets are again exemplified in Plate 
 103, which has a ewer poised in the niche and two more in the spandrels. 
 Another variation, much simpler, with three distinct arches, is shown 
 on Plate 55. In some Ladik carpets even the niche is reduced to an 
 absurdity (Plates 104 and 105). There are several other types of 
 prayer- carpets of Asia Minor bearing geographical names. These are 
 described in a later chapter (see p. 155 foil.). 
 
 Carpets such as that reproduced on Plate 114 seem like intruders 
 among the carpets of Anatolia. Their strange barbaric design, their 
 exclusive use of geometrical forms and the multiplicity of hooked 
 
 1 A Ladik prayer-carpet reproduced by F. R. Martin {O.C., Fig. 340) is dated 
 A.H. iiio=A.D. 1699. 
 
 56 
 
TURKEY 
 
 contours give them an unsophisticated air which telis of nomad life. 
 They are the work of the Yuruks — nomad shepherds of Asia Minor, 
 Before turning to the consideration of the carpets of the Caucasian 
 region^ two examples of uncertain provenance should be referred to. 
 The first (Plate 59) has the form of an Anatolian rug with a niche at each 
 end, but in some respects it resembles still more nearly the carpets of 
 Western Persia, Were it not for the stiff conventionalization of the 
 floral border, and the trivial diaper-filling of the spandrels, the carpet 
 might almost have been attributed to the weavers of Kurdistan,^ It 
 was probably made in Eastern Asia Minor, The other (Plate 60) 
 conforms still more closely to Western Persian types, but the border 
 is found in Caucasian work, and it should be borne in mind while 
 considering the carpets to be discussed in the next chapter, 
 
 1 Cf. Vienna, 0.(7., PI. 66. 
 
 57 
 
Chapter V 
 
 THE CAUCASUS 
 
 In a present-day classification of Oriental carpets the segregation 
 of a group which it is convenient to describe as ** Caucasian ** is inevitable. 
 After very little acquaintance with the subject this term calls up quite 
 a definite image^ though the carpets to which it is collectively applied 
 vary as much as the people do in that part of the world. Few of the 
 carpets placed in this group are older than the eighteenth century, and 
 most are of the nineteenth, though carpet-weaving was carried on among 
 the Caucasian people considerably earlier than that. The anomaly 
 thus involved may be explained first of all by its practical convenience, 
 and again to some degree by the gradual southward advance of Russia 
 over the Caucasian region, and an apparent localisation and fusion of 
 patterns which followed. The predecessors of these '' Caucasian "" 
 carpets must be looked for among those assigned to Asia Minor, Armenia 
 and Persia. 
 
 The territories in question lie between the Caspian and the Black 
 Sea, with the Caucasus Mountains running obliquely across from the 
 one to the other. This range forms a natural boundary between Europe 
 and Asia, and until the beginning of the nineteenth century it marked 
 approximately the frontier between Russia on the one side and Persia 
 and Turkey on the other. From time to time the two latter people 
 disputed for a larger share of the southern borderland, but ultimately 
 Russia shifted her own frontier by conquest southwards, and a Trans- 
 Caucasian province of Russia was constituted. In the first year of the 
 nineteenth century the King of Georgia abdicated in favour of the Czar. 
 In 1827 during a campaign against Persia, Erivan and Tabriz were 
 conquered, and in the following year the Russian frontier was fixed by 
 Treaty at the Aras River. The boundary between the Russians and 
 the Turks was thrust forward in the next year, and in 1878 Russia obtained 
 Batum, The whole of this relatively narrow strip of land between the 
 two great inland seas is peopled, as might be expected, by a mixture of 
 racial elements. 
 
 The carpets are mostly made on the south side of the mountain- 
 
 58 
 
THE CAUCASUS 
 
 range^ but there are carpet-weavers in the narrow angle between the 
 mountains and the Caspian shore on the north side. Motives descended 
 from the old designs of Persia^ Armenia^ Asia Minor and nomad tribes 
 are to be discerned. Generally speaking, these carpets show a stiff 
 conventionalization of floral patterns, and geometrical forms much 
 bolder in scale than those of most nomad carpets. The almost playful 
 introduction of little figures of men, animals or birds, usually in odd 
 corners, is quite frequent. 
 
 The task of differentiating the various local groups now recognized 
 must be left to a later section of this book, to which it properly belongs. 
 A few general considerations are all that need be set down here. 
 
 The carpet reproduced on Plate 6i shows roughly what may be 
 expected in the group. The middle design renders the old Persian 
 palmettes and large floral motives in a fashion almost unrecognizable, 
 while the stems have been transformed into a random medley of zigzags. 
 The border is less broken up, forming a strong contrast with the middle 
 pattern. 
 
 A less drastic change is seen in another carpet (Plate 62) showing 
 the transformation of an entire pattern, not only of single motives, when 
 transplanted to Caucasian soil. The Persian stems and palmettes and 
 cloud-bands are here marshalled into a rigid order quite alien to the 
 old Persian tradition. Although there is no central medallion and no 
 corner spandrels, one need only visualize the stems, flowers and cloud- 
 bands in a freer rendering and a well-known Persian design of the 
 seventeenth century lies before our eyes. The angularity of the design 
 in this case is by no means due to clumsiness or want of technical ability ; 
 on the contrary, it is perfectly intentional. It will be noticed that the 
 ends of the cloud-bands which are repeated so often in the carpet, both 
 horizontally and vertically, have been treated as conventionalized leaves. 
 The local touch which is seldom missing in old imitations of famous 
 patterns, because they were copied in all innocence, without any intent 
 to deceive, may be seen here in the flower consisting of eight curves, 
 which is repeated five times, once almost in the middle of the carpet, 
 and again towards each corner. This motive may often be found in 
 Caucasian carpets (see Plates 125, 132A). 
 
 The carpet shown in Plate 27B is another link between the Caucasian 
 work and the traditional designs of Persia. Another carpet attributed 
 to the Caucasian region is reproduced on Plate 20A. Its elongated leaf- 
 forms on an unusually large scale, overlaid with conventionalized floral 
 
 59 
 
HAND-WOVEN CARPETS 
 
 stems, have an obvious affinity with the dragon- carpets shown on Plates 
 5 and 6, and the Armenian carpet dated a . d » 1679 (Plate 7). The border 
 is an angular rendering of one of the type shown in Plate 165* It will 
 be remembered that the extension of the Russian boundary brought 
 Armenia partly within the territories of Turkey, Persia and Russia. 
 
 A well-known group of Caucasian carpets is represented by Plates 
 63 and 64. These carpets, with their compressed and angular renderings 
 of Persian floral patterns, usually on a deep blue ground, are admirable 
 from a decorative point of view. The borders are almost invariably of 
 white interlinked bands derived from Kufic lettering. It is a curious 
 fact that the earlier forms of the border pattern are as typical of the 
 art of Asia Minor as those of the middle design are of Persia. Occa- 
 sionally carpets of this class have in the middle field an interlaced design 
 in which the forms of Kufic letters appear. The carpet illustrated on Plate 
 1 19 shows a similar middle pattern but a different border — of conven- 
 tionalized floral motives. The narrow stripes edging the borders of 
 these carpets are filled with rows of carnation blossoms turned alternately 
 to right and left — a motive found earlier in Anatolian prayer-rugs. 
 
 This edging is seen on the border of the carpet reproduced in colour 
 on Plate 121. The middle pattern with its row of bold octagons and 
 its uncompromisingly geometrical treatment belongs to a class usually 
 ascribed to the province of Shirvan on the north side of the Caucasus 
 Mountains. The carpet is not older than the nineteenth century, but 
 it is instructive to compare it with Anatolian examples such as those 
 illustrated on Plate 41. A comparison, with the former of these especially, 
 will leave little room to doubt of the derivation of the one from the other, 
 allowing for three centuries of change. It also provides a valuable com- 
 mentary on the origin of the typical modern Caucasian patterns. It is 
 also the favourite pattern of the pileless ** Soumak ” rugs, an example 
 of which is reproduced in Plate 146. The latter again should be 
 compared with the Anatolian rugs above mentioned ; in this case it is 
 the second example which shows the greater similarity. The various 
 types of Caucasian rugs of the nineteenth century, with their territorial 
 names, are dealt with in a later chapter. 
 
 60 
 
Chapter VI 
 
 EASTERN ASIA 
 (a) Turcoman Tribes 
 
 Although few existing Turcoman carpets are likely to have been 
 made before the eighteenth century, there are no carpets of more 
 primitive type. Carpets of Asia Minor and the Caucasus recall them 
 at times, and the earlier these carpets are the more likely are they to do 
 so. It is the same with the carpets reproduced in pictures ; the oldest 
 are most like them. There can be no doubt that the craft among the 
 Turcoman people dates back many centuries. Had there been an export 
 trade, it might have led to the preservation of a few early examples in 
 the West. 
 
 The Turcoman carpets follow a line of their own in matters of 
 design, colour and technique. They are made in a greater variety of 
 forms, and they serve more purposes, than those of other lands. The 
 tradition they follow is no doubt of great antiquity, and if the carpets 
 themselves are of no great age, the reason must be sought in the daily use 
 to which they were all alike subjected. They are made in the regions 
 beyond the farther shore of the Caspian Sea, in a tract extending eastwards 
 as far as Bokhara, northwards to the Sea of Aral, and southwards to the 
 boundary of Persia. There are also Turcoman nomads in Afghanistan, 
 Beluchistan, Persia and Asia Minor, most of whom weave carpets. The 
 sedentary population are not carpet-producers, but the nomads for the 
 most part are skilled in carpet-weaving, and different tribes have their 
 own special patterns. The principal equipments of their large movable 
 tents (kibitka) are made by these weavers. Rugs for the floor are never 
 large. The most characteristic production is the tent-band, usually about 
 a foot wide and sometimes as much as 50 ft. long, which is hung round 
 the tent where the sides join the roof. The designs, sometimes including 
 rude figures of men, besides camels and other animals, are usually in 
 knotted pile on a pileless ground of light tone (see Plate 177). Bags of 
 carpet-weaving are hung round the tent to contain the family belongings. 
 There are also portieres, narrow fringed borders for tent-entrances, 
 camel-collars and other articles of the same material. The work is 
 
 61 
 
HAND-WOVEN CARPETS 
 
 done by the women. The ground is almost invariably a deep purple- 
 red, except in those cases where it is pileless and of a natural pale tone. 
 The pattern is mostly in dark blue, white, and shades of red ; sometimes 
 green or yellow are added. The basis of the pattern is usually polygons, 
 star-shapes, lozenges or crosses, containing, and separated by, interlacings, 
 chequer-patterns and geometrical forms perhaps derived from floral 
 motives. The carpets are excellently well made, the wool of fine quality, 
 the dyes good, the knotting close and regular, and the work durable. 
 The different patterns are specified in a later chapter. Rugs with the 
 prayer-niche are not common, although perhaps the specimen illustrated 
 on Plate 190 may be intended to show this feature. There is a prayer- 
 rug in the Victoria and Albert Museum, with a plain niche, apparently 
 of undyed camel-hair ; the row of dark blue cones in the border is a 
 little unusual in Turcoman rugs. It almost seems as though the nomad 
 weaver had borrowed some ideas, for once, from an Anatolian prayer- 
 carpet. The Turcoman carpets made in Afghanistan and Beluchistan 
 are usually rougher and poorer in technique, and simpler in pattern 
 (see Plates 187 and 193). 
 
 (6) China and Chinese Turkestan 
 
 The history of Chinese carpets has not yet been written, and the 
 task cannot be undertaken with any approach to finality just yet. There 
 is still need for careful research on the spot and for more representative 
 collections to be formed in Europe so as to provide a field for comparative 
 study. A few Chinese carpets, picked up by travellers, were brought 
 to the West during the course of the last century. More recently still, 
 the vogue which arose for Chinese art generally focused attention on 
 the carpets and large numbers were shipped to America and Europe 
 by commercial agents. For the most part, it was as novelties that they 
 gained their popularity. 
 
 Reference has already been made to the fragments of pile-carpets 
 found within the last few years in the Gobi desert region (see p. 6). 
 Further developments must be awaited, and proper facilities for study 
 afforded, before a reliable estimate of their age and nationality can be formed. 
 
 Marco Polo, who makes frequent references to the textile industry 
 of China, says nothing about the weaving of pile-carpets there, although 
 on his way out he remarks how good the carpets of Asia Minor are. 
 
 Chinese carpets to-day form a group of very pronounced 
 individuality. The main characteristics of this group are common to 
 
 62 
 
EASTERN ASIA 
 
 a large area of Asia extending in a wide belt from the westernmost limits 
 of Chinese Turkestan at the apex of India^ through Tibet, and across 
 the northern provinces of China as far as the Yellow Sea. 
 
 There are grounds for believing that pile-carpet making among 
 those peoples is not of great antiquity. They seem rather to have picked 
 up the craft from their neighbours. The most vital point of contact 
 would be Turkestan, where they were in touch with the Turcoman and 
 Persian weavers. Carpets have long been made in Kashgar, Yarkand 
 and Khotan, but they are woven also in Turfan and in other outlying 
 parts of the province. The line of advance towards the Eastern seas would 
 be that sketched out above, and the craft appears to have spread gradually 
 from one point to another along the route. 
 
 The industry in Chinese Turkestan, Tibet and the neighbouring 
 parts of China is carried on chiefly by the women in their homes. As 
 it advances through North China it is taken up by the men, and factories 
 have been established in recent years. 
 
 The Chinese carpets usually regarded as showing the best evidence 
 of age are those in silk, or silk and gold, commonly attributed to the 
 time of the Ming dynasty, but with small justiflcation as a rule. Silk 
 is not an ideal material for carpet-knotting, and its use for such a purpose, 
 though natural in China, is probably no older there than elsewhere in 
 carpet-weaving countries. 
 
 The designs of these carpets have an archaic appearance, due largely 
 to the stiff and angular treatment of the floral motives usually forming 
 the pattern. The Chinese way of putting in the gold thread, where 
 that is used, adds emphasis to this angularity, and the pale and faded 
 tones of the silk strengthen the illusion of age. The example illustrated 
 on Plate 65 is typical. The conventionalization, to a point almost past 
 recognition, of the circular dragon-and-phoenix motive will be noticed. 
 The carpet was originally wider, with the motive in each corner and another 
 in the middle. Blue, the colour of the ground, is the prevailing tone. 
 Much of the pattern is in gold thread. In these carpets the floral motives 
 are usually closely packed to cover the whole ground. A remarkable 
 example acquired in the year 1919 by the Stockholm Museum showed 
 repetitions of two plant forms in lobed compartments on a gold ground, 
 and a border of interlaced floral stems on a silver ground, recalling Persian 
 design.^ Sometimes other motives appear as well. Blue is usually 
 
 ^ Described and illustrated by Dr. Eric Folcker in Burlington Magazine, Vol. XXXV, 
 1919, p. 61. 
 
 63 
 
HAND-WOVEN CARPETS 
 
 the dominant colour of the silk* Where gold thread is used, it is poor 
 in quality and loosely woven into the texture. The carpet partly illustrated 
 on Plate 66 a is a good example of this type. It has a row of eleven niches, 
 all containing floral patterns in silk on a gold-thread ground. Silver 
 thread forms the ground in the spandrels. There is a similar carpet, 
 but smaller, with only six niches, in the Austrian Museum of Art and 
 Industry, Vienna.^ An example of later date, without the metal threads, 
 is shown on Plate 67. 
 
 The silk and gold carpets form a strong contrast to the majority 
 of the Chinese woollen- pile carpets, which show a great variety of 
 motives, often in the bright clear colours characteristic of Chinese art. 
 At times the designs show an originality not beyond criticism. Relief 
 effects cannot be banished from carpet- design, and it is no use making 
 the attempt ; but when flower-pots, incense-vases, table-ornaments 
 and the like are rendered as though they were actual objects placed on 
 the carpet, the result is not entirely free from a suggestion of incongruity. 
 The carpet illustrated on Plate 68 is a very favourable example of this 
 class of pattern, on a white ground. The artistic sense of the Chinese, 
 when the balance is not disturbed too much by purely commercial motives, 
 is still genuine and good, and they know well how to harmonize what 
 would appear as absurdities in other hands. The handicap from the 
 beginning in regard to their carpet industry seems to have been that 
 its lineage was not very ancient, and as a late-comer it was pressed into 
 a mould fashioned for other uses. Figure-subjects have been freely 
 used in Chinese carpet-design. For the most part they represent divinities, 
 saints, sages or priests, often with an inscription in Manchurian, Mongo- 
 lian or Tibetan. These carpets are intended for use in the temples. 
 
 Most of the designs are drawn from the repertory of traditional Chinese 
 art. We find dragons and phoenixes, sometimes together, unicorns ** 
 {ch*i-lin), deer, horses, cranes, bats and butterflies. ** Corean lions ** 
 and various emblems associated with Buddhism are found, as well as 
 emblems of the Taoist philosophy. Landscape effects with trees and 
 animals are sometimes suggested, and the various forms of the long 
 life ** character are introduced. Corean lions and floral sprays in colours 
 on a red ground, form the sole motive of the carpet shown on Plate 69. 
 The dragon-headed horse (lungma) and the phoenix are represented in 
 the middle of the small rug reproduced on Plate 70B. The ground of 
 this carpet is red, and the pattern chiefly in blue and white. 
 
 1 Vienna, O.C., PI. 52. 
 
 64 
 
EASTERN ASIA 
 
 A peculiar arrangement of rocks and waves is sometimes disposed 
 round the middle space based on the inner border of the carpet^ and 
 converging towards the centre. Carpets with simpler patterns^ of floral, 
 diaper, or fret motives are often very decorative. The carpet, in v/hite 
 and two shades of blue, with a simple lotus pattern (Plate 71), is an excellent 
 example. 
 
 The Chinese, and with them the Tibetans, have used the pile-knotting 
 method for a variety of purposes. Some of their carpets are of large 
 dimensions ; others are little mats varying from about 3 ft. to only a 
 few inches square, never intended for the floor. Sometimes two mats 
 are made in one continuous texture to cover the seat and back of a chair 
 — square for the seat and shaped for the back. An example, with a floral 
 pattern in colours on a red ground, is illustrated on Plate 70A. A similar 
 double arrangement of two oblong panels is intended for a long seat. 
 Where three panels are thus woven together, the third is intended to 
 hang down in front. Saddle-cloths are also made in a curved shape. 
 Designs evidently intended to be seen upright, and not suited for a 
 horizontal position, are generally meant to be used as temple-hangings for 
 the wall or altar, or as table- hangings. A species of carpet peculiar to 
 the Chinese and kindred peoples of the Far East is made to be placed 
 round pillars, in such a way as to bring the two side edges together 
 and so complete the design. Sometimes these carpets have standing 
 figures ; more often they have dragons, which appear to be coiling 
 round the columns. There is generally a line of waves and rocks with 
 spray breaking over them below, and a festooned arrangement of jewels 
 and tassels above. The examples on Plates 72 and 73 are shown both 
 flat and in the way they are meant to be used. 
 
 The carpets of Eastern Turkestan were the first among the Chinese 
 group to find their way to Europe, and as they became familiar in the 
 West before others made farther afield were seen, a habit grew up of 
 assigning all Chinese carpets to this region. 
 
 Though preponderatingly Chinese in design and colour, they have 
 a character of their own which renders them distinguishable, as a rule, 
 from the carpets of Northern China. A typical example is shown on 
 Plate 74, and a small mat of very similar design, on a red ground on Plate 
 75A. The pattern of sea waves in the border is very common in these 
 carpets. The fret-ornament in the corners and the rosettes in the middle 
 are motives also frequently found. Sometimes these carpets are more 
 definitely based on those of Western Asia, although the colour-scheme 
 
 65 F 
 
HAND-WOVEN CARPETS 
 
 and the details of the pattern are still Chinese. A carpet shown at the 
 Vienna Exhibition of 1891 had a pattern of flowering stems in the middle, 
 partly issuing from two vases at either end, and a border of counterchange 
 cresting.^ The effect is that of an interpretation of an Anatolian design, 
 though delineated quite in Chinese fashion. 
 
 In the north-eastern provinces, Shan-si, Chi-li and Shantung, 
 the industry appears to have been well established in the eighteenth 
 century. Two small mats reproduced (Plates 75B, 66 b) are probably 
 from those provinces. The latter was brought from North China by 
 the owner. Colonel Croft-Lyons. 
 
 Many carpets find their way to Tientsin for export, and on that 
 account the term ** Tientsin carpet ** has come into use. The place 
 was merely the emporium for the Far East, like Bokhara for Middle Asia, 
 or Smyrna for the Near East. 
 
 Ninghsia and Peking have a name for making the best carpets. 
 Kalgan in Chi-li is also mentioned, and at Tsinanfu in Shantung the 
 craft is taught. Fifty years ago, Colonel Yule, the editor of Marco Polo^s 
 travels, recorded that in the city of T'ai-yuan fu, northern Shan-si province, 
 fine carpets ** like those of Turkey ** were made.^ The phrase probably 
 means no more than that the carpets were of knotted pile. 
 
 The wool of the sheep, goat, camel and yak are used. The material 
 comes chiefly from Mongolia, but local supplies are used where obtainable. 
 
 A common practice among the Chinese carpet-weavers of cutting 
 into the pile in order to round-off the angular contours produced by 
 the knotting process shows that they do not submit willingly to its limita- 
 tions. The result is that many single tufts are cut away on one side 
 down to the knot itself, thereby impairing the durability of the carpet. 
 This practice is not followed elsewhere and it tends to show that in devising 
 it the Chinese were endeavouring to adapt a foreign process of weaving 
 to their own favourite types of design. The colours used by the Chinese 
 are few in number and generally of strong tone. The ground is mostly 
 red, yellow, blue or white, and a few other colours are used. Blue is 
 greatly employed, as in their pottery ; quite frequently shades of blue 
 and white form the entire colour-scheme. The native dyes are said to 
 be by far the best, and where poor dyes have been used, it is to be assumed 
 that they were derived from foreign sources. 
 
 The brisk trade in Chinese carpets in recent times has arisen partly 
 
 ^ Vienna, O.C., PI. 49. 
 
 2 Sir H. Yule, The Book of Ser Marco Polo, 3rd ed. 1903, Book II, ch. 37. 
 
 66 
 
EASTERN ASIA 
 
 from an awakened interest in Chinese art generally. At the outbreak 
 of the war an impetus was given by the cutting off of the carpet-weaving 
 centres of Western Asia, resulting in the organi2;ation of the industry in 
 Northern China for export purposes. Experiments in setting the craftsmen 
 there to imitate the patterns of Western Asia have not been an unqualified 
 success. 
 
 67 
 
Chapter VII 
 
 EUROPE 
 (a) Spain 
 
 It has been known for many years that pile- carpets were made in 
 Spain, but the long record of the industry there, and the extent to which 
 it has been carried on, were unsuspected until recent times* On the 
 one hand, the Spanish weavers had attributed to them typical Oriental 
 carpets with devices, such as heraldry, for instance, betraying Western 
 connexions, while some of their own carpets were ascribed to the East ; 
 and on the other hand it has been hinted that they were slavish copyists 
 of Eastern work, and poor at that* Sheer weight of facts gradually 
 coming to light is correcting such views by the best of all methods, that 
 of demonstrating the truth of contrary ones* During the past forty 
 years the spoils of many churches, convents and mansions have found 
 their way into the market, and although the necessity is to be deplored, 
 there is at least one advantage — that museums have been provided with 
 the material for arriving at a truer understanding of Spanish art* So 
 far as carpets are concerned, we do not know everything yet, but the 
 main facts have been made clear* Not every pile-carpet known to have 
 come from Spain, or still there, is Spanish* Carpets were imported 
 from Anatolia ; many such have been found in Spain, and others may 
 be seen represented in old Spanish paintings* Like all other European 
 countries where carpet-knotting has been carried on, Spain learnt the 
 craft from the East, but she learnt it early, and developed at once on 
 independent lines* There is a decidedly individual character in Spanish 
 carpets, both as regards design and technique* 
 
 Some of the oldest carpets now in existence are Spanish* This 
 conclusion, which sounds improbable at first, is provided with a sound 
 historical basis by the heraldry usually found on these early carpets* 
 They are mostly abnormally long in proportion to their width, a char- 
 acteristic pointing to a local and domestic origin* A carpet may be 
 as long as the weaver desires, but its width is determined by that of the 
 loom, and domestic looms are usually narrow for convenience* It is 
 quite likely that these long heraldic carpets were convent work* Two 
 
 68 
 
EUROPE 
 
 very important examples, shown at the great exhibition of Muhammadan 
 art at Munich in 1910/ came from the Convent of Sta Clara in Valencia, 
 Old Castile, One has a shield-of-arms three times repeated in a line 
 down the middle, on a dark blue ground of diaper ornament. There 
 is an inner geometrical border, and an outer border of slim formalized 
 Kufic characters ^ with human figures, animals and birds filling the 
 spaces between the letters. Along each end outside the border there 
 is a white strip with a design of wild men attacking animals in a wood. 
 The inscribed outer border is a special feature of these early Spanish 
 carpets, and the extra white band at the ends will be met with again. 
 The arms are those of the family of Enriquez* The anchors flanking 
 the shield show that the bearer held the rank of Admiral of Castile, and 
 this places the carpet between the year 1405 and the end of the fifteenth 
 century. The other carpet generally similar, but much larger, is partly 
 reproduced on Plate 76. There are three shields, each several times 
 repeated. They refer to Marina de Ayala and Fadrique Enriquez, Admiral 
 of Castile (d. 1473)*® The carpets may therefore be dated approximately 
 in the middle of the fifteenth century. A few other carpets of the kind 
 are known. One similar to the two described, but without heraldry, 
 was shown at the Munich Exhibition.^ Two more are in the Kunst- 
 gewerbe Museum, Berlin. One of these has shields-of-arms at present 
 unidentified, on a blue ground, and the other has elaborate palmette 
 forms branching on either side from a central stem, in blue, red and white 
 on a dull red ground (Plate 77A). The latter is reputed to have come 
 from a church in the Tyrol. Both appear to be of the fifteenth century. 
 Although the Victoria and Albert Museum has not yet succeeded in 
 obtaining one of these rare carpets, some useful fragments, acquired by 
 gift from Mr. Lionel Harris, show the character of the work. 
 
 Another type, of which a few examples still exist, is no less early in 
 origin. A representative example is illustrated (Plate 78). The arrange- 
 ment of octagons enclosing geometrical forms recalls that of Anatolian 
 carpets illustrated in early pictures — those of Memlinc and his school 
 
 ^ Lent by Mr. Lionel Harris, of London. It is understood that they are now in America. 
 
 2 Sir Thomas Arnold states that there can be no certainty as to the actual significance 
 of letters on carpets of this type. 
 
 ® The identification of the arms on both carpets is due to Mr. A. Van de Put (see Bur- 
 lington Magazine, XIX, p. 344). See also Meisterwerke Muhammedanischer Kunst, 
 Munich, 1910, Vol. I, Pis. 85, 86; Museum, I, 1911, p. 431. The sizes are : (i) L. m. 5.90, 
 W. m. 2.64 ; (2) L. m. 8, W. m. 2.30. 
 
 * Meisterwerke, PI. 87. . 
 
 69 
 
HAND-WOVEN CARPETS 
 
 (Plate 36), or Holbein^ for example* The octagonal spaces which occur 
 so frequently in carpets illustrated in pictures of the fifteenth^ and even 
 the fourteenth century^ are probably primitive and convenient renderings 
 of circles* In the Spanish carpets^ the form is shown in its stiffest 
 convention^ but redeemed by the variety and brilliancy of colouring* 
 The example illustrated has a row of three large red octagons enclosed 
 by rectangular compartments* Within each octagon is a large and elaborate 
 star-device in polychrome on a bright red ground* The star is filled with 
 a variety of diaper patterns and modified Kufic letters* The space 
 around each octagon has a trellis pattern on dark green* There are 
 borders of conventional ornament on pale blue and black* Another 
 example, the property of Dr* Ludwig von Buerkel, was shown 
 at the Munich Exhibition of 1910* It had a double row of octagons, 
 but otherwise it was very similar though less varied in detail* It was 
 probably a few years later in date than the London carpet*^ The later 
 development of this pattern of octagons in Spain is exemplified in a carpet 
 with a red ground in the Victoria and Albert Museum (Plate 77B)* 
 Both the types of Spanish pile-carpets hitherto described have 
 originality enough to render them fairly easily recognisable* A third 
 group, now to be discussed, is quite unique* There is nothing in the 
 whole range of carpet-knotting like them*^ They reproduce the 
 patterns of contemporary Spanish woven silk stuffs, much enlarged in 
 scale, but with no further modification beyond that required by the 
 technique* Such patterns make considerable demands upon the skill 
 and patience of the carpet-knotter where fine lines and curves have to 
 be rendered* The adoption by the Spanish weavers of a single-warp 
 knot (see p* 95), perhaps invented for the very purpose, made the task 
 easier, though it involved great manipulative skill* The earliest-known 
 example of the class is the fine carpet in the Victoria and Albert Museum 
 with a pattern found in several variations among Hispano-Moresque silks. 
 Two lions stand to right and left of a pointed fruit device from which 
 branch downward two long stems which curl round to enclose the lions, 
 uniting above where they end in elaborate foliations* The design is 
 in red and yellow on a dark blue ground* There is a narrow border 
 of dragons, with interlinked necks, in yellow* This carpet has the early 
 Spanish characteristic of inordinate length in proportion to its width* 
 
 ^ Meisterwerke, PI. 88 ; Museum I, 1911, p. 429. See also F. R. Martin, O.C., Fig. 342. 
 
 2 The nearest approach, in principle though not in detail, is made by the English 
 carpets yet to be described. 
 
 70 
 
EUROPE 
 
 The collection of Spanish carpets in the Victoria and Albert Museum 
 is the largest and most varied existing^ and it includes other remarkable 
 examples* Reference must be made to a few of them in tracing the 
 later development of the craft* One of these (Plate 79) has a variation 
 of the familiar ** artichoke ** pattern in green and blue on a red ground* 
 The white outer edging at the two ends, with the pattern of primitive 
 tree-forms and birds, links this carpet with the long heraldic ones described 
 above* It was probably made in the last years of the fifteenth century* 
 A very fine strip, with the lobed pattern so familiar in Italian and Spanish 
 silk velvets of the fifteenth century, belongs to the Hon* H* D* McLaren, 
 M*P* The colours are unusually subdued, on a black ground (Plate 
 80)* Reverting to the Museum collection, a carpet with another render- 
 ing of the "" artichoke ** pattern in green, has a flame- coloured ground, 
 of glowing red dying off into orange — a most beautiful effect of colour* 
 A typical example of the sixteenth century has a fine blue-and-white 
 damask pattern in the middle, with five yellow shields superimposed 
 (Plate 81)* Upon that in the middle is the Sacred Trigram I H S, while 
 those in the corners have the familiar memento mori, a skull and cross- 
 bones* The conventional renaissance border is principally in red and 
 yellow* 
 
 In the seventeenth century the design of Spanish carpets becomes 
 a little uncertain* On the one hand there are the true successors of 
 the frankly Spanish carpets of the sixteenth century, and side by side 
 with these there are deliberate adaptations of the Turkish carpets which 
 were probably to be found in considerable numbers in the peninsula at 
 the time* One example in the Museum has a copy of a floral Turkish 
 carpet-design in the middle, done in blues and greens and yellows, and 
 a border of double-headed eagles* Another with an Oriental design 
 throughout has the word ** Trinidad " wrought into the border — no 
 doubt representing the name of the church or convent for which it was 
 made* The carpet reproduced in colours on Plate 82 is perhaps one 
 of the Spanish adaptations of an Eastern design* 
 
 An interesting carpet of the early eighteenth century given to the 
 Museum by the late Sir Charles Dilke returns to the heraldic tradition* 
 The central field is taken up with a shield-of-arms and an inscription 
 apparently referring to the title Vi^conde de los Villares, created in 1708*^ 
 Two other types of carpets were made in the Spanish peninsula 
 in considerable numbers* One has a looped pile surface in wool or silk 
 
 1 The identification of the arms is due to Mr. A. Van de Put. 
 
 71 
 
HAND- WOVEN CARPETS 
 
 on a linen ground. These were made in the Alpujarra Mountains near 
 Granada and perhaps in other districts. As they are not hand-knotted 
 they need not be described in detail. Others are boldly embroidered 
 on linen and therefore they do not concern us here, 
 
 (6) Poland 
 
 Some reference to the question of carpet-knotting in Poland has 
 already been made, in connexion with the mistaken attribution to Poland 
 of the Persian silk-and-gold carpets. There is ample evidence, however, 
 that pile-carpets were made in Poland, 
 
 Both Polish and Persian carpets are mentioned in old inventories, 
 and examples undoubtedly of Polish origin exist to-day. Two are 
 preserved in the Museum at Lemberg, One has shields-of-arms and 
 the date 1698, The other has a floral pattern. Another carpet with 
 heraldry is in a private collection at Cracow,^ 
 
 The carpet illustrated here (Plate 83) is the property of Dr, Friedrich 
 Sarre, The pile is wool, and the pattern is in colours on a white ground. 
 The general appearance of these Polish carpets recalls the English work, 
 but the treatment of the design is more angular, 
 
 (c) Finland 
 
 The making of hand-knotted carpets probably dates back several 
 centuries among the peasantry of Finland, The question whether the 
 carpet-knotting method is indigenous there, or whether it was borrowed 
 from neighbouring peoples is considered as debatable by investigators 
 of the national arts of the country. 
 
 Considerable numbers of pile-carpets of local origin exist to-day. 
 They are invariably of modest dimensions, mostly about the si^e and 
 proportions of a hearthrug. The knotting is not very fine, and the 
 carpets consequently have a rough and fleecy appearance. The patterns 
 are mostly simple peasant designs based on floral motives, done in bright 
 and varied colours. One of these rugs usually formed part of a bride's 
 trousseau, and many have the initials of the owner and the date. Examples 
 with dates earlier than the middle years of the eighteenth century appear 
 to be unknown, but the craft goes back very much farther. 
 
 At one time the work declined, but in recent years efforts have been 
 made, not altogether without success, to revive it. The example here 
 
 1 Orientalisches Archiv, II, 1911-12, Pis. 11-13. 
 
 72 
 
EUROPE 
 
 illustrated, with the date 1799, is in the Victoria and Albert Museum 
 (Plate 84). The colours are red, green, blue and white on a black ground. 
 The broad border is in blue, white and black on a red ground ; there 
 is a narrow edging in blue. Tapestry-woven mats of Finnish origin 
 are known, although they are rare. Others have been made in Norway 
 and Sweden, but they may be more suitably classed with tapestries, and 
 they hardly come within the scope of the present work. 
 
 (d) France 
 
 The makers of hand-woven carpets in France, as elsewhere in 
 Europe, borrowed the processes from the East. In regard to design 
 they followed a course no less independent and original than the English 
 craftsmen. The real beginning of carpet-knotting in France cannot 
 be traced to an earlier date than the opening years of the seventeenth 
 century. A guild of carpet-makers existed in Paris four centuries before, 
 but nothing now remains of their handiwork, and we are even ignorant 
 of its nature. 
 
 In the year 1601 a craftsman, Jehan Fortier, claimed to be the 
 originator of a process of carpet-making which was described as being 
 after the manner of Turkey and the Levant, and a similar claim was 
 made by Pierre Dupont in 1605. Privileges were granted for carrying 
 on the work. A factory was established by Henry IV in the Louvre 
 under Dupont, and artists of repute were commissioned to make designs. 
 These were probably from the beginning in the style of contemporary 
 French decorative art, like those made under Louis XIV and later. 
 In consequence, French carpets have a very distinct originality. 
 
 The factory of the Savonnerie, which has provided a generic name 
 for all French hand-knotted carpets, was founded in 1626. Its site was 
 a building originally used as a soap-works, on the banks of the Seine at 
 Chaillot, then a suburb of Paris, but since united with the city. Simon 
 Lourdet, a pupil of Dupont’s, was the first director. In 1672 the Louvre 
 atelier was transferred to the Savonnerie, where work for Louis XIV 
 was already in progress. Ninety great carpets were ordered for the 
 Louvre, many measuring as much as 30 ft. in length by 16 ft. or more 
 in width. These commissions kept the looms at work for the greater 
 part of the long reign of Louis XIV. The designs were characterised 
 by the magnificence and splendour associated with that monarch’s name. 
 They were based largely on the acanthus foliage and scrollwork belonging 
 
 73 
 
HAND-WOVEN CARPETS 
 
 to the period, forming a setting for the royal arms and emblems, the 
 trophies of arms, the gods and goddesses and other devices. The effect 
 aimed at, and attained, can only be properly gauged in their appropriate 
 surroundings, lying on the floors of the great saloons of Louis XIV, and 
 associated with the ponderous boule-work and inlaid furniture of the time. 
 More than thirty of these great carpets are still in the French Mobilier 
 National. 
 
 By the kind dispositions of the administrator, M. Dumonthier, and 
 with the sanction of the French Government, a fine series of these carpets, 
 numbering eleven in all, was placed on public view in the Victoria and 
 Albert Museum in London, in the years 1912 and 1921. Two of these 
 carpets in the Mobilier National are here illustrated, with the permission 
 of M. Dumonthier (Plates 85, 86). 
 
 The existence to-day, in a remarkably good state of repair, of so 
 many of these carpets witnesses to the high quality of the craftsmanship 
 and materials. 
 
 Work at the Savonnerie went on during the course of the eighteenth 
 century, and under the First Empire. Numerous carpets in the Mobilier 
 National are of those times. 
 
 In 1825, after a career of two centuries at the Savonnerie, the works 
 were united with the national tapestry-factory of the Gobelins. In these 
 new quarters carpets were made for the palaces of the Louvre, the Tuileries, 
 the Elysee, St. Cloud, Compiegne and Fontainebleau, for the Cathedral 
 of Notre Dame and for the Pantheon. 
 
 The work is still carried on at the Gobelins. The modern carpets 
 not infrequently exceed the older in weight and size. At times their 
 designs are based on earlier traditions, in order to adapt them to their 
 destined entourage. 
 
 74 
 
Chapter VIII 
 
 ENGLAND 
 
 The making of pile-carpets by hand in England followed very closely 
 upon the first importation of Oriental examples. In the sense in which 
 the term carpet ** is used to-day^ and in any sense cognate to it^ carpets 
 were unknown in this country five hundred years ago. In the West 
 their gradual adoption has kept pace with progressive ideas of comfort. 
 In the East they are the outcome of the conditions of daily life^ and they 
 have always been indispensable. There are very few carpets now in the 
 world more than five hundred years old^ but that fact provides no clue 
 to the antiquity of the carpet^ since such goods were liable by their nature 
 to perish in the using. Whatever our uncertainties may be in regard 
 to the East^ we can be pretty sure that practically nothing was known in 
 England of the pile-carpet before the reign of Henry VII at the earliest, 
 if not that of his successor. There are many illustrations of interiors 
 in English paintings and miniatures executed before those times, but in 
 none of them is a pile-carpet represented. As late as the fifteenth century 
 a flooring of stone or earthenware tiles, or, perhaps, wood-boards, was 
 considered to need no covering for the sake of comfort or appearance. 
 The occupants of the more modest dwellings probably had often to be 
 content with the bare earth ; and, indeed, some of our country churches 
 were no better off at a much later time. A foot-cloth,"^ a piece of rich 
 stuff of some kind, would be laid down before the altar of a church or 
 before the throne in the royal presence-chamber. 
 
 These cloths, spread out in special places on occasions of ceremony, 
 were only in a very limited sense the predecessors of the modern carpet. 
 They did not entirely meet the problem of the treatment of the floor 
 in mediaeval times. In the houses of the well-to-do, sweet rushes, hay, 
 straw, foliage, fragrant herbs or flowers were scattered over the floor. 
 These could be renewed at will, and the broom might be more or less 
 drastically used, at regular or irregular intervals, according to the taste 
 of the individual. 
 
 An illustration from a celebrated manuscript in the Library of 
 Lambeth Palace, entitled Dictes and Sayings of the Philosophers, a Latin 
 
 75 
 
HAND-WOVEN CARPETS 
 
 book translated into English by Earl Rivers^ shows the translator offering 
 the book to King Edward IV. The cloth of honour is brought forward 
 so as to pass under the feet of the king^ while loose rushes, painted bright 
 green in the picture, cover the whole of the rest of the floor.^ 
 
 Even after another century had passed, at the end of Eli2;abeth's 
 reign, the state of affairs was very much the same. Paul Hentzner, a 
 German who came to London in 1598, states that Queen Elizabeth's 
 presence-chamber at Greenwich was strewn with hay. In fact, the interior 
 economy of our houses seems to have been in harmony with their outward 
 appearance. We recall the picturesque exclamation of the Spaniard in 
 Queen Mary's time : '' These English have their houses made of sticks 
 and dirt, but," he is careful to add, " they fare commonly as well as the 
 king." 
 
 Rush-matting, although used to cover the floor by our nearest 
 neighbours, the French, at the beginning of the fifteenth century, does 
 not appear to have been adopted in England before Henry VIII's reign, 
 and its general use cannot be ascribed to an earlier time than that of 
 James I. One of the earliest representations of rush-matting in an English 
 picture is in a portrait of Henry VIII at Belvoir Castle.^ The rushes 
 are simply plaited in a diagonal fashion. In this reign the pile- carpet 
 first comes on the scene. Cardinal Wolsey was probably the first 
 Englishman, for a subject at any rate, to bring pile-carpets into use. They 
 had long been known, though only as rare and valuable commodities, 
 on the Continent. A minister who, if rumour spoke truly, fed his aspiring 
 soul with dreams of the papal tiara, and who at one time would have 
 rejoiced to secure the imperial crown for Henry, must not be chargeable 
 with insular habits. He must be equipped in the fashion of the times, 
 and carpets must be got. The foreign trade of the country being then 
 largely in the hands of Continental merchants resident in London and 
 the seaports, he could apply to the Venetian factors ; but that course 
 would be tedious and slow. The steps he did take lifted the question 
 out of the sphere of commerce into that of international diplomacy. The 
 dispatches and reports of Sebastian Giustinian, then Venetian Ambassador 
 in England, show how much significance that dignitary attached to a 
 request from the Cardinal for some carpets.^ Negotiations went on for 
 
 ^ Allen, Selections from Erasmus, p. 126, gives a criticism of the practice. 
 
 2 The Drapers’ Hall is said to have had mats in the Chequer Chamber, and rushes 
 in the hall, in 1495, but we have no information as to the material of the mats. 
 
 3 Four Years at the Court of Henry VIII. Selections of dispatches, etc. Vol. II., 
 pp. 198 foil. 
 
 76 
 
ENGLAND 
 
 more than a year. In June^ 1518^ Giustinian writes home to the Signory 
 from Lambeth that the Cardinal had promised to take him before the 
 Council and obtain audience for his arguments in regard to the repeal 
 of the duties on Candian wines imported into England by the Venetian 
 traders. ** After these colloquies^'^ the ambassador proceeds, ** his 
 right reverend lordship requested me very earnestly to contrive with the 
 magnifico the captain and the masters that, paying for the same, he might 
 have certain choice carpets and some other articles, but above all the 
 carpets. I told him that I did not know whether there were any, but 
 that if there were, his lordship should have them. I suspect he will 
 not be accommodated, which will prove of serious detriment to us ; 
 whereas had he received twelve or fifteen small handsome carpets, he 
 would have been extremely satisfied. Should your Excellency think 
 fit, you might see either to forwarding them by land, or promise that 
 he should receive some by the next galleys.'^ In November of the same 
 year the ambassador writes again from Lambeth. The Cardinal, he 
 says, was extremely angry with the Venetian merchants in London, who 
 appear to have done something which led him to believe that they thought 
 too lightly of his authority in the kingdom. Giustinian had been to 
 see him with the object of arranging the dispute. He appears to have 
 been in some degree successful. The Cardinal sent for the merchants, 
 who offered him seven very handsome Damascene carpets. He was 
 willing to accept this present from the ambassador, but not from the 
 merchants. In the end, he agreed to regard them as the joint gift of 
 both. These carpets were but a drop in the ocean. The autumn of 
 the next year came, and still the request for the carpets, and the contingent 
 question of the duty on the Cretan wines remained as before. Giustinian 
 is by this time back in Venice. In October, 1519, he makes a report 
 in the Senate on his legation in England. The Cardinal is still very 
 anxious, he says, for the Signory to send him one hundred Damascene 
 carpets, for which he has asked several times. The Senate is urged to 
 make the gift, and even if the Signory does not choose to incur the expense, 
 the London factory will take it on themselves. The gift might easily 
 settle the affair of the wines of Candia, whereas it would be idle to discuss 
 that matter further until the Cardinal receives his hundred carpets. In 
 October, 1520, Wolsey received sixty carpets from Venice. Doubtless 
 they were similar in character to that shown in the full-length portrait 
 of Henry VIII at Belvoir Castle, to which reference has already been 
 made. The king stands on a carpet with a pattern of arabesques in 
 
 77 
 
HAND-WOVEN CARPETS 
 
 large compartments. A royal portrait- groups painted by Holbein^ in 
 the Privy Chamber at Whitehall, after that palace had been handed over 
 to the king by Wolsey, perished in the fire of 1697-8, but copies made 
 before that date show the king standing on a carpet somewhat similar 
 to that at Belvoir.^ Still another portrait of Henry, in Earl Spencer's 
 collection at Althorp, claims our attention. The king is seated at a table 
 with the Princess Elizabeth ; the jester Somers stands behind them. On 
 the table is spread a small Oriental rug. It will be remembered, of 
 course, that for a long time after Oriental carpets were first brought 
 into Europe it was far more common to use them as tablecloths than 
 as floor- coverings. Evidence on this point meets us continually in 
 Flemish, Dutch, and Italian pictures of the fifteenth to the seventeenth 
 century. 
 
 The actual making of pile-carpets in this country followed hard 
 upon their first importation from the East. A carpet represented in part 
 on Plate 87A has in the middle the royal arms of England, with the initials 
 of Queen Elizabeth and the date 1570. On the left are the arms of the 
 borough of Ipswich, and on the right those of the family of Harbottle. 
 The whole of the pattern is thoroughly English and typical of Elizabethan 
 times. In particular, the borders of honeysuckle and oak-stems are to 
 be noticed. The carpet is the property of Lord Verulam. It is probably 
 the oldest existing example of carpet-knotting in England. Attempts 
 at carpet-weaving may have been made at an earlier date, even under 
 Wolsey himself ; for among his household goods at Hampton Court 
 were several woollen table-carpets of English making." The patterns 
 of these are described, and they are not inconsistent with the idea 
 that the carpets may have been of knotted work ; but, still, we can- 
 not be sure. It may be taken for granted that, in view of the wide 
 disparity as regards the cost of labour, the English craftsman in the 
 days of Elizabeth was no better able to enter the lists against the skilled 
 Oriental weaver than he is to-day. But for all that, the English worker 
 had his own sphere, and the craft, once taken up, was long kept 
 alive. 
 
 Before the end of Elizabeth's reign, the English Turkey (or Levant) 
 Company had begun direct trading with the Eastern Mediterranean, 
 and carpets were more easily obtained. Some of these were copied, 
 more or less faithfully, in England. A carpet in the Victoria and Albert 
 Museum has a stiff geometrical pattern (Plate 88). The shield-of-arms 
 
 ^ There is a copy by Remigius van Leemput at Hampton Court. 
 
 78 
 
ENGLAND 
 
 in the middle of three of the borders is that of Sir Edward Apsley* An 
 unusual feature is the inscription, along one of the short sides, in beautiful 
 Elizabethan characters in knotted pile, like the rest of the carpet. It 
 reads as follows : Feare God and keepe his commandements made 
 
 in the yeare 1603/' The knighthood was conferred in that year, and 
 the carpet may have been made to commemorate the event. The design 
 is Eastern in origin, but the carpet is English. In the same museum 
 there is a small panel bearing the arms of Queen Elizabeth, with her 
 initials and the date 1600 (Plate 87B). The design is typical throughout 
 of the English art of the time. It is not a large piece, and it could hardly 
 have been intended for the floor. This knotting process came to have 
 the name of ** Turkey work ** in England. The knot is quite a simple 
 one — the same as that generally used in Turkish carpets. Panels made 
 in this way were often used for upholstery, and numbers of them exist 
 to-day.^ 
 
 The carpet reproduced in colours on Plate 89 is a very fine example 
 of the time of James I. The date 1614 will be found in the border, and 
 the whole design is characteristically English of the period. The carpet 
 is in the possession of Lady Hulse. There is an English carpet of similar 
 design at Knole. A second example at Knole has the arms of the Countess 
 of Dorset, who died in 1645. 
 
 English travellers got to Persia early in Elizabeth's reign, and a 
 little later an effort was made to find a Persian carpet-weaver who might 
 be induced to come to England. 
 
 A chapter in Hakluyt’s Voyages is entitled Certaine directions 
 given ... to M. Morgan Hubblethorne, Dier, sent into Persia, 1579/^ 
 In this chapter we read as follows : In Persia you shall finde carpets 
 
 of coarse thrummed wool, the best of the world, and excellently coloured : 
 those cities and towns you must repair to, and you must use means to 
 learn all the order of the dyeing of those thrums, which are so dyed as 
 neither rain, wine, nor yet vinegar can stain. ... If before you return 
 you could procure a singular good workman in the art of Turkish carpet- 
 making, you should bring the art into this realm, and also thereby increase 
 work to your Company.” 
 
 We have no evidence that Hubblethorne secured his Persian 
 workman. Carpet-knotting in the old style went on, however, until 
 
 ^ In the Inventory of 1679 there were mentioned four Turkey-work carpets in the 
 Wardrobe, and twelve Turkey-work chairs and one Turkey-work carpet in the lower offices. 
 (Mrs. Charles Roundell, Ham House, pp. 50, 51.) 
 
 79 
 
HAND-WOVEN CARPETS 
 
 the middle of the seventeenth century. By that time the country was 
 richer^ and the trade of the East India Company having developed^ there 
 was no real difficulty about getting carpets from India and Persia, though 
 it was a costly business. The Girdlers' carpet, as already stated (p. 39), 
 was obtained through the East India merchants in 1634. The quantity 
 of carpets of this Indo-Persian class still to be seen in old English houses 
 bears witness to the extent of this commerce, and, incidentally, to the 
 national wealth in the latter half of the seventeenth century. As a 
 consequence, the industry at home seems practically to have died out ; 
 and when, as we shall see in a moment, it was revived in the middle of 
 the eighteenth century, it was regarded as a new industry. 
 
 Meanwhile, experiments in carpet-knotting ** after the manner 
 of Turkey and the Levant ** had been made in France, as already related. 
 Two craftsmen in the Savonnerie at Chaillot, having some difference 
 with the administration, removed to London in 1750, and began making 
 a carpet in a room at Westminster. They had raised some money by 
 subscription, but they soon got into difficulties. In the end they applied 
 to a fellow-countryman in London, Pierre Parisot, a tapestry-weaver. 
 Parisot succeeded in interesting the Duke of Cumberland, and engaged 
 the men in 1751. The factory was removed to Paddington, and the 
 first carpet made was presented by the Duke to the Princess of Wales. 
 After two years the works were removed to Parisot's tapestry-factory 
 at Fulham. Great expectations were raised by the undertaking, but 
 its career was short. The entire works of the Fulham factory were sold 
 by auction in London in 1755. It is the usual story of such enterprises 
 in England. The work was costly, and there was only a limited demand. 
 Perhaps had the factory been able to pull through the first few years 
 of effort and financial straits, its career might have been assured; but 
 even that is doubtful.^ 
 
 The models and appliances thus fell into other hands, and so the 
 failure of Parisot^s venture must have helped towards the success which 
 attended other efforts made about that time. 
 
 In 1756, the year following the sale of the Fulham works, the Society 
 of Arts offered a premium for making carpets in England in imitation 
 of those made in Turkey and Persia. The name of Thomas Moore, of 
 Chiswell Street, Moorfields, is the first on the list of recipients. A carpet 
 he produced to the Society was considered to be ** in many respects 
 equal, and in some respects superior, to those imported from Persia 
 ^ See the Gentleman’s Magazine, Vol. XXIV, 1754, p. 385. 
 
 ■ 80 
 
ENGLAND 
 
 and Turkey/' ^ He received a premium of £25 in 1757^ and a like sum 
 was awarded to Thomas Whitty of Axminster» The next year Whitty 
 received £25 again ; while Passavant of Exeter received an equal amount. 
 Again^ the following year (1759) the Society made an award of £50^ giving 
 £30 to Whitty and £20 to William Jesser of Frome. 
 
 More than twenty years later the good results of these awards were 
 apparent. There is a note in the first volume of the Society's T ransactions 
 (1783) that by them the manufacture of carpets is now established 
 in different parts of the kingdom, and brought to a degree of elegance 
 and beauty which the Turkey carpets never attained." That expression 
 of opinion is coloured by the tastes of the day, which would have 
 discovered more beauty in the Graeco- Roman elegancies of the brothers 
 Adam than in anything the East ever had, or ever could have, produced. 
 But we are in a position to judge for ourselves. 
 
 Not much appears to be known of the subsequent career of Jesser's 
 factory. In regard to Passavant's at Exeter, the carpet illustrated on 
 Plate 90 was doubtless made there. It bears the inscription Exon 
 and the date 1757. A carpet very similar in character, but different 
 in design, is at Petworth House. It has the same inscription, with the 
 date 1758. 
 
 The other two recipients of the Society of Arts' awards, Moore 
 and Whitty, attained a greater celebrity. Moore seems to have attracted 
 the favourable notice of Robert Adam, and this brought him commissions 
 for carpets for mansions built or enlarged by that celebrated architect. 
 The carpet illustrated on Plate 91 is in the red drawing-room of Syon 
 House. The design is in the style which we have learned to associate 
 particularly with the brothers Adam. There can be no doubt as to the 
 craftsman responsible for the weaving. His name, Thomas Moore," 
 with the date 1769, is woven into the border. At Osterley, another 
 house associated very extensively with Robert Adam's activities, the 
 Earl of Jersey possesses several carpets which are shown by the records 
 to have been made by Moore. The designs for four of these are in the 
 Soane Museum. Those for the Tapestry Room and Etruscan Room 
 are dated 1775 ; that for the State bedroom is dated 1778. There is 
 no date on the design for the drawing-room. The Soane Museum has 
 an extensive collection of Robert Adam's designs for carpets. 
 
 Thomas Whitty's factory outlasted the others. The industry which 
 
 ^ A Concise Account of the Rise ... of the Society. . . . By a Member of the said 
 Society, 1763, p. 58. 
 
 81 
 
 G 
 
HAND-WOVEN CARPETS 
 
 still flourishes at Wilton traces its origin to him. He first started carpet- 
 making at Axminster^ in the Court House near the churchy in 1755. 
 There carpets continued to be made for about eighty years. That factory 
 was succeeded by Moody's^ at Wilton^ where carpet-making has gone 
 on ever since. 
 
 Among the Axminster products were some very elaborate and costly 
 carpets made for the Pavilion at Brighton. The expenditure on the 
 building and equipment of this royal residence was very lavish, and the 
 carpets were in keeping with the general scale of magnificence. An 
 enormous carpet was made for the Saloon about 1823. It was shaped 
 at the ends to fit the room. A portion is still preserved at Buckingham 
 Palace. The pattern is on a light ground, and shows the influence of 
 the Chinese taste so conspicuous in the decoration of the Pavilion. 
 
 In another carpet, made for the Music-Room, the pseudo- Chinese 
 style is still more marked. It is now only a fragment, but an account 
 is given by Nash and Brayley,^ who describe it as one of the largest 
 in the kingdom, its dimensions being 61 ft. by 40 ft., and its weight about 
 1,700 lb. It was made at Axminster to fit the room, costing £700. 
 The Saloon carpet cost £620. Another carpet, for the Banqueting Room, 
 was made at Axminster to Mr. Jones' design, costing £735. 
 
 An account of the carpet-knotting industry in this country would 
 lack its natural counterpart if all reference were omitted to the weaving 
 of carpets in Scotland and Ireland. The manufacture of carpets in the 
 south-west of Scotland was carried on with much success in the second 
 half of the eighteenth century. The industry has had an unbroken and 
 prosperous career ever since. It seems likely that hand-knotting was 
 not adopted there at first, but in 1831 the Trustees for Manufacture in 
 Scotland awarded two premiums — of £150 and £30 — for four Turkey- 
 carpets to a Kilmarnock firm. These are said to have been the first 
 of that luxurious and costly type manufactured in Scotland.^ Hand- 
 knotted carpets continued to be made in the south-west of Scotland for 
 about twenty years. Early in the present century efforts were made 
 to start carpet-knotting among the fisher-folk of Sutherland and Caithness, 
 but in the end it was found impossible to induce the girls to settle down 
 to regular work. The looms were removed to Glasgow, where they 
 
 1 Illustrations of Her Majesty's Palace at Brighton, 1838, p. 9. 
 
 2 This is completely outclassed by a carpet lately made in Donegal, weighing 2J tons. 
 ® British Association Reports, Glasgow, September, 1876. The Textile Industries, 
 
 by James Baton, pp. 204-6. 
 
 82 
 
ENGLAND 
 
 were kept going until the outbreak of the war. There are still deft fingers 
 ready to make carpets to-day, whether knotted after the Turkish or the 
 Persian manner. But the mainstay of the carpet industry of Scotland 
 has ever been the use of those ingenious contrivances for producing a 
 pile-surface by mechanical means. 
 
 When we come to Ireland we must begin a little farther back. Irish 
 ** rugs ** were in demand in this country in the sixteenth and seventeenth 
 centuries. But the rug ** of those days was a rough material, shaggy 
 in appearance, perhaps like a modern blanket. Holinshed has a story 
 of a man who went to a bear-baiting in London on a frosty morning 
 wearing a Waterford rug. ** The mastiffs,^' says the Elizabethan 
 chronicler, ** had no sooner espied him than they set on him, thinking 
 him to be a bear.'' It is a long stride from the Waterford rug of Queen 
 Elizabeth's days to the Donegal carpets of our own times. Perhaps 
 the chief link connecting them is the quality of the Irish wool. The 
 hand-knotted carpet industry was introduced, or revived, in Ireland 
 about twenty years ago. A factory was opened at Killybegs in 1898, 
 and the work was soon extended to other centres in North and South 
 Donegal. Before long, hundreds of workers were employed. Some 
 very good results have been obtained. An imposing product of the 
 Donegal factories is in the Library of Australia House in London. It 
 measures 46 ft. 6 in. by 23 ft. The general tone of the ground is brown. 
 There is a fine border of wattle and vine-leaves. This carpet is matched 
 by others of the same manufacture now in this country. Others are 
 also to be found in Scotland and Ireland and on the Continent, as well 
 as in Canada, the United States, Egypt and South Africa. Fine carpets 
 have also been made in Kildare and in Queen's County. 
 
 In conclusion, the carpets made by our great craftsman William 
 Morris must not be forgotten. It was in 1879 that his first experiments 
 in carpet-knotting were made in Queen Square. Then carpet-looms 
 were set up at Hammersmith, and finally the work was transferred to 
 Merton. A careful study of the old productions of the East was the 
 foundation of his work. ** They show us the way to set about designing 
 such things," he said. Modern carpets, ** while they should equal the 
 Eastern ones in material and durability, should by no means imitate 
 them in design, but show themselves obviously to be the outcome of 
 modern and Western ideas." 
 
 Morris made some noble carpets. The large carpet made for the 
 Earl of Carlisle's drawing-room at Naworth, was finished in 1881. It 
 
 83 
 
HAND-WOVEN CARPETS 
 
 took nearly a year to make^ and ** weighed about a ton/' Another fine 
 carpet was made for the late Earl of Portsmouth. It was arc-shaped, 
 made so as to fit the place for which it was destined at Hurstbourne. 
 The floral pattern is interrupted by three large shields bearing the arms 
 of the Earl and the Countess. 
 
 84 
 
PART II 
 TECHNICAL 
 
■ I V I . ■ '(.' .’ ' /■'''■ ''.Iff ^ 
 
 ' ■ ■ ■ ' ■‘..■■'Y't ii 
 
 ;•■.■*•■ . '■■■' •> ■'• "• 
 
 '■ ' . -f ;/•: i. -Vc.v?<?a’v. 
 
 •; •; . \ ••ivy*.:- ■■;■■ ■' 
 
 
 'iS 
 
 4'« '•? N<*'SM 
 
 Ilf 
 
 
 .v-?:p 
 
 w. 
 
 ' •■'■ !'.<■ r':\'"'w4U 
 
 ■■■' : V 
 
 ■ “^ ■( !* K.s,,.. •■'.I.'Vj 
 
 ..'wv 
 
 ■'I-- .■■.'"iV;'l'!'{&t 
 
 
 
 W'*i 
 
 
 
 Im 
 

PLATE S3. 
 
 Ghiordes prayer-carpet, late 17th cen- 
 tury. (See page 54.) 
 
Chapter I 
 
 THE TECHNIQUE OF HAND-WOVEN CARPETS 
 (a) Knotting and Weaving 
 
 It is essential to students to have some knowledge of the technique 
 of carpets^ for not only does it help in understanding and appreciating 
 at its proper worth any particular specimen standing alone^ but it is 
 also of the utmost importance in grouping together kindred pieces^ and 
 in associating with such groups, when made, any fresh carpet that may 
 be met with* In this latter respect technique is of even more importance 
 than design itself, at least so far as the finer subdivisions are concerned* 
 The connoisseur who identifies a carpet by its general appearance really 
 pays more attention to technique than he is usually aware of, for the 
 same design looks very different when it is rendered in various textures ; 
 and the texture, apart from the design, produces an impression more 
 valuable for purposes of diagnosis than may be reali2;ed* This 
 impression is made much more vivid and reliable by an understanding 
 of the knotting, weaving and other technical details of the specimen 
 under consideration* Apart from this, moreover, it is easy to understand 
 that a weaver is more inclined to change his designs than his methods* 
 The demand of the market may be for certain patterns at the 
 time popular ; or the beauty of rugs made in other places and following 
 other traditions may appeal to him so strongly that he will copy their 
 designs ; but there are no such reasons to induce him to depart from 
 his accustomed methods of weaving* So far as unorgani2;ed production 
 is concerned the technique is a very reliable indication of the place or 
 people responsible for the rug* On the other hand, in the case of a highly 
 organi2;ed factory system, both the design and technique of some other 
 district may be copied, and then the question of the identification of 
 the product may be a very difficult or unanswerable one* Luckily it 
 is in these circumstances that the answer is of the least consequence* 
 
 In the first place one requires to distinguish a hand-made from 
 a machine-made carpet* It is not easy to give a simple rule to decide 
 the point ; but the inspection of a few pieces, especially from the back, 
 soon makes the difference obvious even to the untrained eye, and the 
 
 87 
 
HAND-WOVEN CARPETS 
 
 perusal of the following pages describing the making of hand-knotted 
 carpets should afford positive indications by which the matter can be 
 placed beyond all possible doubt. Leaving the question of machine- 
 made carpets for a subsequent brief mention^ the whole field of hand-made 
 carpets can be divided into two great classes — those with smooth faces, 
 and those with a pile. Both classes are textile fabrics ; which means 
 that they consist essentially of two sets of threads — the warp and the 
 weft — which cross each other at right-angles and are interwoven so as 
 to make a coherent tissue. The smooth-faced carpets consist of warp 
 and weft alone : the pile carpets have, in addition, short extra pieces 
 of thread knotted to the warp-threads, so that their free ends stand up 
 and form a surface similar to that made by the blades of grass in a meadow. 
 
 Like other textile fabrics carpets must be made on a frame, which 
 is known as a loom. This consists essentially of two horizontal beams 
 or rollers, kept a proper distance apart by side-pieces. The threads of 
 the warp are stretched between these two beams and all lie parallel to 
 each other and usually in a vertical position. When they are placed 
 on the loom the weaving can begin. The weft-threads are put in one 
 by one : the first thread passes in front of alternate warp-threads (suppose 
 the first, third, fifth, etc.), and behind the remaining warp-threads (the 
 second, fourth, sixth, etc.) ; the second weft-thread reverses the arrange- 
 ment of the first, passing behind the odd-numbered and in front of the 
 even-numbered warp-threads ; the third weft-thread repeats the arrange- 
 ment of the first, the fourth of the second, and this alternate reversal 
 is carried throughout the whole of the fabric (Fig. c, p. 92). 
 
 The loom need not be an elaborate structure. In its simplest form 
 it may consist of two round pieces of wood, tied or otherwise fixed to 
 two adjacent trees, one piece being near the ground and the other as 
 much higher as the length of the warp requires. On such a loom the 
 necessary tension of the warp can be obtained by wedging the beams 
 further apart, or by tightening the cords which lash them to the trees. 
 A more convenient arrangement is to have for beams two rollers capable 
 of rotation, fixed in a permanent framework (Fig. a). This plan has 
 the advantage that the loom need not be the full height of the carpet ; 
 for a supply of warp can be wound round the upper beam (thus often 
 called the warp-beam) and unwound as the finished part of the carpet 
 is, bit by bit, rolled round the lower, or cloth-beam. With such an 
 arrangement the weavers can sit all the time in about the same position, 
 though it is still often found convenient to accommodate them on a plank 
 
 88 
 
THE TECHNIQUE OF HAND-WOVEN CARPETS 
 
 that can be placed at various heights on the rungs of two ladders. At 
 the Victoria and Albert Museum there is a Persian Carpet-loom which 
 was in actual use in a house in Kirman a few years ago. It has on it 
 a small rug, partly finished, and it is accompanied by all the usual 
 implements of the carpet weaver (Plate 92A). A typical form of Turkish 
 loom accommodating several weavers is shown in Plate 923. The loom 
 is not always vertical, though in the majority of cases it is so. Among 
 nomad weavers, and especially in the districts of Central Asia, a 
 horizontal loom is quite common. The two beams rest on the ground 
 and mostly have slots in them through which the side-pieces pass, the 
 whole arrangement strongly resembling the common embroidery-frame. 
 The weavers squat on the finished part of the carpet, which lies a few 
 inches above the ground but can easily be supported by placing under 
 it pieces of wood or other objects. The necessary attitude is, however, 
 an uncomfortable one for persons unaccustomed to it. 
 
 The laborious method of guiding the weft-thread in its proper course 
 round each warp-thread separately can be avoided by a simple contrivance 
 added to the loom. A thick stick is thrust through the warp in exactly 
 the same manner as a weft-thread, only it is placed far above the place 
 where the actual weaving is taking place. This stick divides the warp- 
 threads into two sets or leaves f and forms what is called a shed (Fig. a). 
 All the odd-numbered warps are in one leaf and all the even-numbered 
 in the other. The weft-thread, wound on a shuttle, can now be easily 
 passed between the two leaves of the shed and at once takes its right 
 position in relation to the warp. For the next passage of the weft a reverse 
 shed has to be made. This is effected by attaching all the warp-threads 
 of the back leaf of the shed to a strong rod, by means of short cords of 
 equal length — an arrangement called a heddle. If then the rod is drawn 
 forward or suitably rotated, the warp-threads attached to it will be drawn 
 in front of the other set and so a reverse shed is formed, through which 
 the weft-thread can be passed as before (Fig. b). It is clear that one shed 
 is permanently set on the loom, but the other can be temporarily substi- 
 tuted for it at will. 
 
 Some other accessories of the weaver must be mentioned here. 
 One is a heavy comb-like implement for beating the weft down. Its 
 teeth pass between the warp-threads, and a few blows consolidate the 
 latest rows of weft and knots with the finished part of the carpet. For 
 severing the pile-yarn a knife is mostly used, which is held in the right 
 hand while the knot is tied with the left. Shears with long and slightly- 
 
 89 
 
HAND-WOVEN CARPETS 
 
 curved blades are employed for trimming the pile of the carpet to a 
 uniform level. Balls or skeins of coloured yarn for the pattern hang 
 in a row from the upper beam or heddle-rod within easy reach of the 
 weavers. All these accessories are illustrated in Plate 92. 
 
 As implied already 
 smooth-faced carpets are 
 simpler in construction than 
 pile-carpets and^ conse- 
 quently, It is convenient 
 to describe the details of 
 their technique first. They 
 are called Kilims (Gilim), 
 and the most common kind 
 — which are essentially sim- 
 ilar in technique to the 
 tapestry- hangings of Europe 
 — are often called Kara- 
 mani, after the province 
 of Karaman in Southern 
 Asia Minor, though their 
 production is by no means 
 confined to that locality 
 but indeed extends through 
 almost the whole of the 
 carpet-making regions of 
 the East. 
 
 The warp does not 
 show in the finished Kilim, 
 but the pattern is entirely 
 formed by the weft which 
 is beaten down sufficiently 
 to hide the former com- 
 pletely. To make a varied 
 pattern the weft-threads of course must be of different colours and so 
 cannot be passed right across the loom, but each thread is taken only 
 so far as that particular colour is needed and then passed back again, 
 and so on until a patch of colour of the desired shape is formed, the 
 whole pattern being composed of many such coloured patches. If in any 
 place the boundary between two patches of colour is vertical (i.e. parallel 
 
 90 
 
 FIG. A.— CARPET-LOOM. 
 
THE TECHNIQUE OF HAND-WOVEN CARPETS 
 
 to the warp)^ it is clear that a vertical slit must result there (Fig. c)^ for 
 no weft-thread passes from one patch to the other to bind the fabric 
 together. This defect is commonly avoided by designing the pattern so 
 that vertical lines either do not occur or else are very short, but the 
 weavers of Karaman 
 and Kurdistan often 
 make a feature of this 
 very limitation and allow 
 an open-work effect to 
 give an added beauty to 
 their work. 
 
 From the manner 
 of their manufacture 
 Kilims are practically 
 alike both back and 
 front. Sometimes, it is 
 true, short ends of the 
 weft lie loose at the 
 back ; but more often 
 these are inwoven so as 
 to make the two sides 
 indistinguishable. The 
 peculiarities of the tech- 
 nique of Kilims have 
 a great influence on the 
 patterns adopted. Fine 
 details are rarely seen, 
 and the constant use 
 of diagonal or stepped 
 outlines is a feature 
 which mostly enables 
 them to be distinguished 
 
 , FIG. B.— LOOM WITH SHED REVERSED. 
 
 at a glance. 
 
 Smooth-faced carpets of another kind are known as Soumaks. These 
 differ from ordinary Kilims and from all other carpets in the way the 
 weft is interlaced with the warp. Each weft-thread is passed forwards 
 in front of four and then backwards behind two warp-threads (Fig. d). 
 Thus the stitches, as they may be called, overlap each other on the front 
 of the carpet, and as they usually slope in opposite directions in consecutive 
 
 91 
 
HAND-WOVEN CARPETS 
 
 rows^ a kind of herring-bone effect is produced. At the back^ the shorter 
 stitches form a series of ridges ; but the back is not intended to be seen 
 and no attempt is made to keep it free from loose ends of yarn. The 
 normal texture of a Soumak has often to be slightly modified near the 
 boundaries of the pattern^ for the long stitch across four warp -threads 
 would sometimes interfere with the finer details. Another difference 
 between Soumaks and other Kilims is that the former have after every 
 one or two rows of stitches an extra weft-thready which passes right across 
 the carpet but is not visible when the latter is finished. 
 
 The fineness of texture of Kilims depends upon the closeness of 
 the warp -threads, which are usually spaced so that from five to twenty- four 
 
 fall within one inch. The 
 weft-threads lie consider- 
 ably closer together than 
 this, but whereas each 
 warp-thread forms in the 
 fabric a fine rib, which 
 is easily perceptible, the 
 exact si^e of the weft does 
 not to any considerable 
 extent affect the appear- 
 ance of the carpet. It 
 may be said that there 
 are about six times as 
 many weft -threads as 
 warp -threads in a given 
 distance. 
 
 Kilims are made more 
 quickly than pile-carpets and are generally lighter in substance, as a 
 result of which their wearing qualities are not equal to those of the latter. 
 In the East they are largely used on the floor, but in Western countries, 
 where perhaps the conditions are more severe, they are often reserved 
 for hangings or furniture coverings, for which purposes they are admir- 
 ably adapted. 
 
 Useful and beautiful as the smooth-faced carpets are, they are much 
 excelled in importance by the larger class of carpets with a pile surface. 
 In these the warp and weft foundation is completely hidden, as far as 
 the front is concerned, by the more or less upright pile. The effect 
 of use is gradually to wear away this pile, but as long as any is left the 
 
 92 
 
 FIG. C.— KILIM OR TAPESTRY WEAVING. 
 
THE TECHNIQUE OF HAND-WOVEN CARPETS 
 
 vital threads forming the foundation of the carpet are protected from 
 serious injury. Even the back of the carpet, though it gets far less wear 
 than the front, is protected by the pile-yarn passing round the warp- 
 threads. This is by no means the only advantage that a pile gives. In 
 a Kilim the surface is composed of the fibres of the yarn lying flat in the 
 plane of the carpet ; in a pile-carpet it is the countless cut-ends of the 
 fibres that are presented to the eye, an arrangement which gives to the 
 surface a quality that no other 
 can. The colours of the yarn 
 become richer and more trans- 
 parent ; the contours of the pat- 
 tern are made softer and more 
 mysterious, and the tone of each 
 coloured area not only varies 
 subtly from point to point, but 
 often changes completely as the 
 spectator views it from different 
 positions. 
 
 The pile is formed from 
 short pieces of yarn, rarely more 
 than one inch long, tied to, or 
 twisted round, the warp-threads 
 — the whole piece, including the 
 free ends, usually being called a 
 knot or tuft. The knots are tied 
 in rows right across the carpet, 
 each row being finished and 
 secured by weft-threads before 
 another row is begun. With few 
 exceptions the knots are attached 
 to two adjacent warp-threads, 
 but there are various ways in which they may be tied. In all Eastern 
 carpets, and nearly all Western ones, only two kinds of knots are found. 
 They are known as the Ghiordes, or Turkish knot, and the Sehna, or 
 Persian knot. The names are those of carpet-weaving villages in Asia 
 Minor and Western Persia respectively. 
 
 In the case of the Ghiordes knot the yarn lies across and in front 
 of the two warp -threads and its free ends pass round behind them and 
 then emerge again between them. As each pair of warp-threads has 
 
 93 
 
HAND-WOVEN CARPETS 
 
 its own knot it will be understood that consecutive warp-threads have 
 either two, or no yarn- ends passing forward between them, and that 
 what may be called the collar of the knot is seen lying across a pair of 
 threads (Fig, e). 
 
 In the case of the Sehna knot the yarn completely encircles one 
 warp-thread of a pair ; one yarn-end emerges between them, while the 
 other passes behind the second warp-thread and comes to the front 
 outside it (Fig, f). Thus, with the Sehna knot one yarn-end emerges 
 
 between each pair of warp-threads right across the loom, and the collar 
 of the knot is seen crossing one warp-thread only ; in both knots the 
 free ends come below the collar so that the pile inclines to the lower 
 or first-woven part of the carpet* 
 
 The Ghiordes knot is symmetrical from side to side while the Sehna 
 knot is unsymmetrical ; it follows that the latter may be tied in two ways, 
 for the encircled warp-thread may be either the right-hand or left-hand 
 one of the pair, as seen by the weaver. The right-hand Sehna knot is 
 much more common than the left-hand, and its pile tends to lie towards 
 the left, while the pile of the left-hand inclines towards the right (Fig* 
 g). If the knot is known to be of the Sehna variety, passing the hand 
 across the carpet will usually tell at once of which kind it is. 
 
 When the warp-threads are spaced very closely on the loom it may 
 be impossible for them to lie all in one plane, in which case alternate 
 threads fall towards the back, and the warp lies on two levels. 
 Sometimes the depression of the back set is only slight, but it may be 
 
 94 
 
THE TECHNIQUE OF HAND-WOVEN CARPETS 
 
 so pronounced that only half the warp-threads are seen at the back of 
 the carpet. A Ghiordes knot tied on a warp on two levels may incline 
 
 either to the right or to the left ; but 
 in the case of the Sehna knot^ the en- 
 circled warp-thread is always the one 
 lying nearer to the front of the carpet 
 (Fig. h). In a tightly- woven carpet 
 with a warp on two levels it is often 
 very difficult to be sure which kind of 
 knot is used, because in either case two 
 yarn- ends emerge between every pair of 
 warp-threads of the front set. If the 
 knot is a Ghiordes, a little probing will 
 usually show its collar crossing two warp- 
 threads ; but if this cannot be found, 
 then, though a Sehna knot may be strongly suspected, the proof may 
 be quite difficult. It may be said, how- 
 ever, that the Ghiordes knot is rarely 
 found with a warp on very different 
 levels, and a Sehna knot still more 
 rarely with a warp strictly on one level. 
 
 Both these knots are sometimes found 
 tied on four instead of on two threads, 
 though rarely throughout the whole 
 of a carpet. In such a case the knot 
 itself is not really different, for it only 
 means that two warp -threads are taken 
 together and treated as one. 
 
 The adoption of one or other of 
 these knots is rather due to custom than 
 to any inherent advantage of either. It 
 is true that carpets of very fine texture 
 usually have the Sehna knot, and dis- 
 tinctly coarse ones almost invariably 
 have the Ghiordes, but it is quite pos- 
 sible to use the latter for very close 
 knotting. 
 
 A third kind of knot, tied on a 
 single warp-thread, is much less com- 
 
 95 
 
 FIG. G.— LEFT-HAND SEHNA KNOT. 
 
HAND-WOVEN CARPETS 
 
 mon than the two already described, being found only, but almost 
 exclusively, in early Spanish carpets* The yarn completely encircles 
 the single warp-thread, one end crosses the other at the back and then 
 the ends come to the front, one at each side (Fig. i). It is tied only 
 on alternate warp - threads ; and on that account, as will be seen later, 
 is very easy to recognize. 
 
 When a row of knots, of whatever kind, has been tied, one or more 
 weft-threads are put in. The most usual number is two, in which case 
 the thread is taken from side to side right across the loom, and then 
 without being severed taken back through the reverse shed. If only 
 one shoot (as it is called) of weft is made after a row of knots, then 
 
 the shoot after the next row is 
 taken through the reverse shed. 
 Three or more shoots of weft 
 may be found after a row of 
 knots, but the larger numbers are 
 not often used throughout the 
 whole of a carpet. 
 
 The weft when beaten down 
 usually takes a more or less sinuous 
 course in the thickness of the car- 
 pet, for the warp is too strong and 
 tight to be much deflected from a 
 straight line. If the warp lies on two 
 levels the weft -threads through one shed are necessarily more sinuous than 
 those through the reverse shed ; and if the levels are very distinct then 
 it will happen that the weft-thread through one shed is quite straight 
 and undeflected while the other is very sinuous indeed. In many of the 
 finer Persian carpets with a warp on two levels, a common plan is to 
 have uniformly three shoots of weft after each row of knots, the middle 
 shoot being very sinuous and the other two straight (Fig. h). 
 
 To ensure a carpet having even edges and being free from bulges 
 or puckers, and of equal width throughout, it is necessary to allow the 
 weft a proper and uniform amount of slack and then beat it down so 
 that its tension is the same in each part of the width of the carpet. 
 
 When a line of weft, consisting of one or more shoots, is completed 
 the carpet is ready for another row of knots. The Ghiordes and Sehna 
 knots are tied on the same pairs of warps as before, that is they come 
 vertically above each other : on the other hand the single-warp knots 
 
 96 
 
 i/W M L'\ k'' 
 
 FIG. I.— SINGLE. WARP KNOT. 
 
THE TECHNIQUE OF HAND-WOVEN CARPETS 
 
 are tied on those warps left blank in the previous row, and so fall into 
 diagonal instead of vertical lines. A carpet with the single-warp knot 
 therefore has the easily-noticed peculiarity that diagonal lines in the pat- 
 tern are much smoother in contour than those in a truly vertical direction. 
 
 The fineness of knotting in a pile carpet is a feature of great 
 influence in determining the beauty of its texture and the amount of 
 labour necessary for its production. It depends primarily upon the 
 closeness of spacing of the warp-threads on the loom. The number 
 of these to the linear inch may vary from about eight to more than sixty, 
 giving from four to thirty or more knots to the inch in a horizontal line. 
 The number of knots in a vertical line (i.e., in the direction of the warp) 
 depends upon the fineness of the yarns employed, the number of shoots 
 of weft after each row of knots and also upon the thoroughness with 
 which the beating-down is carried out. These details are mostly arranged 
 so that the height of the knots is roughly equal to their width, but it 
 mostly happens that the latter dimension is the smaller and the ratio of 
 height to width is commonly about four-fifths, though sinking as low 
 as one-half in some groups of carpets. 
 
 The simplest way of estimating the fineness is to compute the number 
 of knots in a square inch of surface. This number will vary from about 
 1 6 to nearly i,ooo, though in the vast majority of carpets it lies between 
 50 and 100. If there are more than 100 knots to the square inch the 
 texture may be considered fine. Carpets with more than 300 knots are 
 rarely met with and there appears to be no advantage to the appearance 
 of the carpet in pushing the fineness beyond 400 knots to the inch, though 
 phenomenal pieces of such texture are occasionally seen. As far as 
 durability is concerned, a coarser carpet, provided it is well knotted and 
 not extremely coarse, is more satisfactory than a finer one. 
 
 Whether a bit of yarn just long enough for a knot is used in making 
 it, or whether, as is a common practice, the end of a long piece is taken 
 and cut off when the knot is tied, it is impossible to ensure all the free 
 ends being of the same length, so that a final operation is to clip the 
 pile level with shears. Here the Chinese show some of their peculiar 
 ingenuity, for by cutting down the knots nearest to the contours of the 
 pattern, they cause a distinct furrow to be formed which not only serves 
 to round off the angularities of the contour but also throws the design 
 into relief. This effect of relief is often increased by trimming somewhat 
 closer the whole of the pile of the ground. 
 
 In carpets other than Chinese, though the usual practice is to have 
 
 97 
 
 H 
 
HAND-WOVEN CARPETS 
 
 the pile as even as possible, yet its length varies considerably in different 
 specimens* In some rugs, notably those from Sehna and Ghiordes, 
 the pile is so short that the collars of the knots are scarcely hidden* In 
 such cases the rug itself is thin, and the pile stands upright, which makes 
 the surface lustreless and harsh to the touch* 
 
 The Nomadic tribes of the Caucasus and Kurdistan, on the other 
 hand, prefer a long and sometimes almost shaggy pile, a choice which 
 results in thick and mostly very durable rugs* Such a pile of necessity 
 lies flat, and exhibits to the utmost the lustre of the wool* A long pile 
 of a dull wool, as seen too much in India, has the least satisfactory 
 appearance of any* 
 
 The treatment of the sides and ends of a carpet still remains to be 
 mentioned* At the sides the returning weft-threads by themselves form 
 a selvedge and prevent the fabric from becoming unravelled, but in use 
 they would soon get worn through* It is customary, therefore, to have 
 at the edge one or more stouter warp-threads and to leave these free 
 from knots, but to serve them with extra turns of weft, and then to fortify 
 the whole with an overcasting of extra threads which may form one or 
 several cord-like lines along the side* 
 
 As the ends of the warp-threads are loose when the carpet is removed 
 from the loom, unravelling is a danger which must be provided 
 against* Usually before the knotting begins a piece of plain, or of 
 tapestry-weaving is done at the beginning of the carpet, and a similar 
 piece is added at the top when the knotting is finished* When the carpet 
 is taken from the loom the loose ends of the warp are either braided in 
 various ways or they are taken in groups of about four to eight, which 
 groups are next taken in pairs and knotted together, and this process 
 often is repeated until there are several rows of knots* In either case 
 the carpet usually terminates in a loose fringe* The lower end of the 
 warp may consist of loops instead of cut-ends, owing to the way in which 
 it was originally put on the loom* These loops may be treated as above, 
 but an alternative plan when they exist is to fill them up with a few rows 
 of ordinary weaving so that a true selvedge is formed* This cannot be 
 done at the upper end of the carpet, for it would be almost impossible 
 to have the loops of warp there all of the same length* 
 
 Many smaller points of technique, as well as some variations from 
 the normal methods described above, may be very useful in identifying 
 carpets ; but they can be mentioned more conveniently with the groups 
 in which they are found* 
 
 98 
 
THE TECHNIQUE OF HAND-WOVEN CARPETS 
 
 The structure of machine-made carpets is generally quite different 
 from that discussed above* It is notoriously difficult to devise a machine 
 to tie a knot, though recently it has been found possible to reproduce 
 the Ghiordes knot on a machine, and the method is employed commercially. 
 The process most in use, however, is one founded on the technique of 
 velvet-weaving. In this, besides the foundation-warps, other warps 
 of much greater length are woven into the fabric and allowed to stand 
 up in a series of loops. These are cut through in ** Wilton ** and left 
 uncut in the ** Brussels ** carpets. In other methods loops of the pile- 
 yarn are held in position by being pinched in various ways by the foundation 
 threads : or lengths of chenille are first made separately and then woven 
 with other foundation threads in such a way that they lie in the face 
 of the fabric. It is by such a method that the ** patent Axminster ** 
 carpets were made. 
 
 (b) Group Characteristics of Early Carpets, 
 
 The technique of the oldest carpets known does not differ in 
 essentials from that of those of the present day, and all the details found 
 in them have already been described in the preceding chapter, but it 
 may still be useful to take each of the well-defined early groups in turn 
 and give a short account of their technical construction. 
 
 The early Persian carpets with a woollen pile may be divided into 
 two groups — those made in the periods roughly corresponding to the 
 sixteenth and seventeenth centuries. 
 
 Persian carpets of the earlier of these two groups invariably have 
 the Sehna knot. The warp is on two levels and there are three shoots 
 of weft after each row of knots. The warp is sometimes of silk, but 
 more often of cotton. The middle or sinuous shoot of weft is either 
 of silk or cotton, and the other two shoots are mostly of cotton but not 
 infrequently of wool and occasionally of silk. The texture is very fine, 
 there being on the average about 230 knots to the square inch. The 
 pile is generally of wool, though some metal thread, inserted by the 
 tapestry method, is found in a few examples. 
 
 The selvedges and ends of these early carpets rarely remain now, 
 but it appears that the sides were usually finished with a single over- 
 casting, and that the ends had a simple warp-fringe — in fact that the 
 treatment was the same as that of the best modern Persian carpets. The 
 number of different colours used was fairly constant and rather large, 
 being on the average nine or ten. The colours most frequently occurring, 
 
 99 
 
HAND-WOVEN CARPETS 
 
 placed in order of the amount used of each, are : — crimson : dark blue : 
 yellow : orange : ochre : white : blue : light blue : green : black. 
 Other colours, placed in order of frequency of occurrence, are : — light 
 red : green-blue : red : brown : pale green : light crimson : pale 
 
 brown. In this and the following lists, shades and colours not included 
 are occasionally found. 
 
 Persian Carpets of the later group also have the Sehna knot and 
 a warp on two levels and in most respects resemble the earlier ones. 
 The warp, however, is generally of cotton : the sinuous shoot of weft 
 is mostly of cotton and the other two shoots are either of cotton or wool. 
 The texture is not so fine as in the earlier group, but still the average 
 number of knots to the square inch is as high as 175. Metal thread is 
 not used. The average number of colours used is nine or ten as before. 
 Those most frequently occurring, placed in order of the amount used, 
 are : — dark blue : crimson : blue : white : yellow : light blue : green : 
 orange : red : black. Other colours placed in order of frequency of 
 occurrence, are : — green-blue : ochre : red : brown : purple : pale 
 
 green. 
 
 It will not be supposed, of course, that the foregoing descriptions 
 apply to the average standard of the carpets made in the periods in 
 question, for they are, and can only be, based on the carpets which have 
 come down to us and which were therefore the most precious or the most 
 durable of those made. 
 
 The carpets usually called Indo-Persian (see p. 39) which in point 
 of date probably centre about the beginning of the seventeenth century, 
 or a little later, also have the Sehna knot, a warp on two levels and three 
 shoots of weft. The warp is either of silk or cotton, and for the weft 
 cotton is generally used for all three shoots. There are on the average 
 about 165 knots to the square inch. The pile is of wool with occasionally 
 a little white cotton. The colours, which are very bright, rich and 
 transparent, number ten on the average. Those most frequently occurring, 
 placed in order of the amount used, are : — crimson : deep green- blue : 
 green : yellow : orange : dark blue : white : pale crimson : blue : 
 light blue : black. Other colours, placed in order of frequency of 
 occurrence, are : — pale brown : deep yellow : deep purple : light red : 
 purple : dark brown. 
 
 The carpets of Asia Minor, of the kind called Ushak, which date 
 from the fifteenth to the seventeenth centuries, are invariably woven 
 with the Ghiordes knot. The warp is of white wool and lies on one 
 
 100 
 
THE TECHNIQUE OF HAND-WOVEN CARPETS 
 
 level or on two levels very slightly separated. The weft is of red wool 
 and there are two shoots after each row of knots. The pile is of wool 
 with occasionally a little white cotton^ and the texture is rather coarse. 
 The sides and ends have rarely been preserved, but it seems that the 
 former were usually finished with a selvedge consisting of several cords. 
 Traces of wide coloured webs at the ends of many carpets make it appear 
 that this was the usual way of treating them. 
 
 The earlier of these Asia Minor carpets — that is, those up to about 
 the end of the sixteenth century, have on the average about 70 knots 
 to the square inch. The number of different colours is on the average 
 seven. Those most frequently occurring, placed in order of the amount 
 used, are : — red : dark blue : yellow : white : black. Other colours, 
 placed in order of frequency of occurrence, are : — light blue : green : 
 green-blue : dark blue : light red : purple. 
 
 The later of these Asia Minor carpets have an average of about 
 85 knots to the square inch. The average number of different colours 
 is seven as before. Those most frequently occurring, placed in order 
 of the amount used, are : — red : dark blue : blue : yellow : green : 
 light blue : black. Other colours, placed in order of frequency of occur- 
 rence, are : — green-blue : white : purple. 
 
 The Asia Minor carpets commonly called Transylvanian differ 
 in some respects from those already described. The warp is mostly 
 dyed, yellow being the usual colour ; and the weft is more often yellow 
 than red. The average number of knots to the square inch is 70. There 
 are six different colours on the average. Those most frequently occurring, 
 placed in order of the amount used, are : — red : yellow : white : blue : 
 black. Other colours, placed in order of frequency of occurrence, are : — 
 light blue : purple-red : light red. 
 
 A group of Turkish carpets dating from the sixteenth century are 
 woven with the Sehna knot, which is never used in other Turkish rugs. 
 They may be divided into four classes, but it is not certain that 
 carpets of the last two were made in Turkey (see p. 51) though 
 the technique considered alone points that way. 
 
 Those of the first class (see p. 46) have a warp of very fine yellow 
 wool, apparently that of the Angora goat. The warp has a right-handed 
 twist, and lies on two levels. There are three shoots of weft, which 
 is of wool usually yellow in colour. The pile is of wool, and there are 
 on the average 85 knots to the square inch. The average number of 
 different colours is seven. Those most frequently occurring, placed 
 
 lOI 
 
HAND-WOVEN CARPETS 
 
 in order of the amount used, are : — crimson : yellow : yellow- green : 
 blue : green : light blue : white* Other colours, placed in order of 
 frequency of occurrence, are : — deep yellow : dark blue : black. 
 
 The carpets of the second class (see p. 47) resemble those of the 
 first in most respects. The warp (also with a right-handed twist) is, 
 however, often of silk, and is crimson, green or yellow. The knotting 
 is much finer, there being on the average 205 knots to the square inch. 
 The average number of different colours is six. Those most frequently 
 occurring, placed in order of the amount used, are : — crimson : yellow : 
 light blue : white : green. Other colours, placed in order of frequency of 
 occurrence, are : — blue : dark blue : black : yellow-green : deep yellow. 
 
 The third class consists of carpets with arabesque patterns of the 
 kind shown in Plate 48 (see also p. 51). Technically, they resemble 
 the first group very closely though the texture tends to be rather finer. 
 The average number of different colours is five or six. Those most 
 frequently occurring, placed in order of the amount used, are : — crimson : 
 green : light blue : yellow : white. Other colours, placed in order 
 of frequency of occurrence, are : — blue : black : purple. 
 
 The fourth class comprises carpets with arabesque pattern of the 
 kind shown in Plate 47 (see also p. 51). Apart from having the Sehna 
 knot, they resemble in technique the ordinary Ushak carpets. The 
 warp is of coarse white wool with the usual left-handed twist, and lies 
 on one level. There are two shoots of woollen weft — mostly red in colour. 
 The knots average about 80 to the square inch. The average number 
 of colours is seven or eight. Those most frequently occurring, placed 
 in order of the amount used, are : — red : light blue : blue : white : 
 yellow : black. Other colours, placed in order of frequency of occurrence, 
 are : — brown : purple : dark blue. 
 
 Spanish carpets of the sixteenth and seventeenth centuries are made 
 entirely of wool and the single-warp knot is generally used. The warp 
 is white, and the weft, which usually consists of three or four untwisted 
 strands, is red, white or yellow. The average number of knots to the 
 square inch is no. The sides are finished with a narrow web of plain 
 weaving, which no doubt was turned under the carpet. The ends, appar- 
 ently, had wide webs and fringes. There are on the average about six 
 colours. Those most frequently met with, placed in order of the amount 
 used, are : — red : blue : yellow : green : white : light blue. Other 
 
 colours, placed in order of frequency of occurrence, are : — dark green : 
 light red : dark brown : bright yellow : black : orange. 
 
 102 
 
THE TECHNIQUE OF HAND-WOVEN CARPETS 
 
 The carpets made in England during the sixteenth and seventeenth 
 centuries were clearly based, as far as their technique is concerned, upon 
 Asia Minor models, though important differences are found. The warp 
 is of hemp, or perhaps coarse flax, and it lies on one level. The weft 
 is of the same material as the warp and there is one shoot after each row 
 of knots. The Ghiordes knot is used, and the pile is of wool. The 
 texture is rather coarse, there being on the average about 65 knots to 
 the square inch. The sides were finished with a narrow band of plain 
 weaving and there appears to have been a similar web at the ends in most 
 cases. The colours used were often bright and each was used in several 
 shades so that the total number of differently- coloured wools was high. 
 Apparently in the seventeenth century there was a tendency to replace 
 the hempen warp with one of wool, and in still later times the warp was 
 more generally of the latter material. 
 
 The following are descriptions of the technique of those carpets 
 illustrated in this book, which it has been possible to examine closely. 
 In each case the colours have been placed, as far as could be judged, 
 in the order of the amounts occurring in the particular specimens described. 
 After a colour the most important parts of the design where it is used 
 are sometimes specified and in this connexion it may be noted that the 
 border-stripes are numbered from the outer edge of the carpet, 
 
 Persian Carpet (Victoria and Albert Museum), 15 th-! 6th 
 
 century, 18 ft, X 9 ft, 10 in, (See Plate i and p, 19,) 
 
 Warp : Two-ply green silk. On two levels, 45 to i inch. 
 Weft : White silk. Three shoots after each row of knots. 
 
 Knots : Wool, Sehna, 21 to i inch, 470 to the square inch. 
 Colours : Nine, Crimson (field ; outer part of 2nd border) : dark 
 blue (inner part of 2nd border) : orange (ist border) : white 
 (3rd border) : blue : light yellow : light green : black : light 
 brown, 
 
 Persian Garden Carpet (Wagner Collection), ibth-iyth century, 
 17 ft. 5 in, X 14 ft, 2 in. (See Plate 3 and p, 12,) 
 
 Warp : Four-ply white cotton. On two levels. 33 to i inch. 
 Weft : Wool, or occasionally silk, of various colours. Three shoots 
 after each row of knots. 
 
 Knots : Wool, Sehna, 14 to i inch, 240 to the square inch. 
 Colours : Nine. Blue (field) : light blue : white : yellow (border) : 
 crimson-red : blue-green : brown : dull green : black. 
 
 103 
 
HAND-WOVEN CARPETS 
 
 Armenian Carpet (Victoria and Albert Museum). lyth-iSth 
 century. 6 ft. ii in. x 6 ft. (See Plate 5B and p. 16.) 
 
 Warp : Two-ply white wool. On two levels. 16 to i inch. 
 
 Weft : Red wool. Two shoots after each row of knots. 
 
 Knots : Wool. Ghiordes. 12 to i inch. 100 to the square inch. 
 
 Colours : Eight. Red (field) : blue : white (ist border) : yellow : 
 light blue : dark blue : brown-black : purple. 
 
 Armenian Carpet (Victoria and Albert Museum). Probably 17th 
 century, ii ft. X 6 ft. 4 in. (See Plate 8 and p. 16.) 
 
 Warp : Two-ply whitish brown wool. On two levels. 18 to i 
 inch. 
 
 Weft : Red-brown wool. Two shoots after each row of knots. 
 
 Knots : Wool. Ghiordes. 9 to i inch. 80 to the square inch. 
 
 Colours : Seven. Brown-black (field) : white (border) : blue : 
 
 yellow : crimson-red : light blue : purple. 
 
 Persian Carpet (Victoria and Albert Museum). i6th century. 
 19 ft. 10 in. X 13 ft. (See Plate 9 and p. 16.) 
 
 Warp : Four-ply white cotton. On two levels. 22 to i inch. 
 
 Weft : Buff cotton. Three shoots after each row of knots. 
 
 Knots : Wool. Sehna. ii to i inch. 120 to the square inch. 
 
 Colours : Ten. Crimson (field) : ochre (2nd border) : dark blue : 
 white : green-blue (3rd border) : blue : green : light blue : 
 light red : black. 
 
 The Ardabil Carpet (Victoria and Albert Museum). Persian ; 
 dated 1540. 34 ft. 6 in. x 17 ft. 6 in. (See Plate ii and p. 17.) 
 
 Warp : Two-ply yellow silk. On two levels. 35 to i inch. 
 
 Weft : Yellow silk. Three shoots after each row of knots. 
 
 Knots : Wool. Sehna. 19 to i inch. 340 to the square inch. 
 
 Colours : Ten. Dark blue (field) : crimson (4th border ; long 
 panels in 2nd border) : white (3rd border) : black (2nd border) : 
 orange-red (ist border) : yellow (central medallion and corner 
 pieces) : green (round panels in 2nd border) : blue : light blue 
 (pattern of ist border) : light crimson. 
 
 Persian Carpet (The Earl of Ilchester). ibth-iyth century. 20 ft. 
 
 X 8 ft. (See Plate 12 and p. 20.) 
 
 Warp : Six-ply white cotton. On two levels. 28 to i inch. 
 
 104 
 
THE TECHNIQUE OF HAND-WOVEN CARPETS 
 
 Weft : Pale red cotton. Three shoots after each row of knots. 
 
 Knots : Wool. Sehna. i6 to i inch. 230 to the square inch. 
 
 Colours : Thirteen. Dark blue (field) : deep crimson (2nd border) : 
 blue-green (pattern on ground of field) : light crimson (pattern 
 on ground of 2nd border) : buff (ist and 3rd borders) : yellow 
 (pattern on ground of ist and 3rd borders) : white : light 
 blue : crimson : brown : blue : black : light blue- green. 
 
 Persian Carpet (Victoria and Albert Museum). Probably 17th 
 century. 7 ft. 7 in. x 5 ft. 5 in. (See Plate 14 and p. 21.) 
 
 Warp : Yellow silk. On two levels. 54 to i inch. 
 
 Weft : Yellow silk. Three shoots after each row of knots. 
 
 Knots : Wool. Sehna. 27 to i inch. 730 to the square inch. 
 
 Metal Thread : Gilt-silver strips wound on a. core of white or yellow 
 silk. Tapestry- woven on the upper layer of the warp, three 
 shoots being equivalent to one row of knots. 
 
 Colours : Eleven. Crimson (medallion and border panels) : dark 
 blue (field) : green (2nd border) : light crimson (ist border) : 
 white (3rd border) : light blue : brown : yellow : light green : 
 black : orange. 
 
 Persian Carpet (Victoria and Albert Museum). 17th century. 
 24 ft. 6 in. X 9 ft. 5 in. (See Plate 19 and p. 24.) 
 
 Warp : White cotton. On two levels. 31 to i inch. 
 
 Weft : Two shoots of white wool and one shoot of white cotton 
 after each row of knots. 
 
 Knots : Wool. Sehna. 12 to i inch. 180 to the square inch. 
 
 Colours : Eleven. Dull crimson (field) : dark blue (outer border) : 
 blue (inner border) : white : yellow : green-blue : light red : 
 light blue : light purple : black : purple-brown. 
 
 Persian Carpet (Victoria and Albert Museum). 17th century. 
 5 ft. 5 in. X 10 ft. 4 in. (See Plate 20B and p. 25.) 
 
 Warp : White cotton. On two levels. 18 to i inch. 
 
 Weft : White cotton. Three shoots after each row of knots. 
 
 Knots : Wool. Sehna. 7 to i inch. 60 to the square inch. 
 
 Colours : Nine. Red : blue : yellow : white : green : dark blue : 
 light crimson : blue-green : black. The compartments are 
 
 variously coloured upon no fixed plan. 
 
 105 
 
HAND-WOVEN CARPETS 
 
 Persian Carpet from the Palace of the Forty Columns at Ispahan 
 (fragments at the Victoria and Albert Museum)* Late i6th century* 
 Original size 71 ft* X 33 ft* (See Plate 21 for design and p* 25.) 
 
 Warp : Three-ply white cotton* On two levels* 24 to i inch* 
 Weft : White cotton* Three shoots after each row of knots* 
 
 Knots : Wool* Sehna* 15 to i inch* 180 to the square inch* 
 Colours : Eight* Crimson (field) dark blue (2nd border) : yellow- 
 green : white (ist and 3rd borders) : light crimson : black : 
 brown (?' black of a different dye)* 
 
 Persian Prayer- Carpet (Mr* George Mounsey)* i8th century* 
 5 ft* II in* X 4 ft* 5 in* (See Plate 30A and p* 36*) 
 
 Warp : White cotton* On two levels* 21 to i inch* 
 
 Weft : White cotton* Three shoots after each row of knots* 
 
 Knots : Wool* Sehna* 9 to i inch* 100 to the square inch* 
 Colours : Two* White (ground) : copper-red in various shades* 
 
 Indian Carpets (Mr* Lionel Harris)* Early 17th century* A* 
 Shaped piece, 14 ft* 7 in* x 9 ft* B* Square piece, 24 ft* 2 in* x 9 ft* 2 in* 
 (See Plate 31 and p* 39*) 
 
 Warp : White cotton* On two levels* 26 to i inch* 
 
 Weft : Pink cotton* Three shoots after each row of knots* 
 Knots : Wool* Sehna* 13 to i inch* 170 to the square inch* 
 Colours : Nine* Crimson (field) : Green-blue (2nd border) : light 
 blue : white : deep purple : yellow : light crimson : black : 
 dark green* 
 
 Indian Carpet (Victoria and Albert Museum)* Early 17th century* 
 4 ft* 8 in* X 2 ft* II in* (See Plate 32 and p* 39.) 
 
 Warp : Two-ply silk (red, green or white)* On two levels* 46 
 to I inch* 
 
 Weft : Red silk* Three shoots after each row of knots* 
 
 Knots : Wool* Sehna* 26 to i inch* 600 to the square inch* 
 Colours : Twelve* Crimson (field) : yellow-buff (serrated leaves) : 
 light green (border) : blue-green : dark blue : white : blue : 
 light red : deep yellow : light blue : purple : dark purple* 
 
 Indian Carpet (The Girdlers' Company)* Dated 1634* 24 ft* 
 X 8 ft. (See Plate 33 and p* 40.) 
 
 106 
 
THE TECHNIQUE OF HAND-WOVEN CARPETS 
 
 Warp : Probably cotton* 30 to i inch* 
 
 Weft : Probably cotton* 
 
 Knots : Wool* Sehna* 16 to i inch* 240 to the square inch* 
 Colours : Twelve* Crimson (field) : dark blue (4th border) : 
 
 white : blue : light crimson : light blue : yellow : light brown : 
 very light crimson : dark brown : light green : purple-brown* 
 
 Asia Minor Carpet (Victoria and Albert Museum)* i6th century* 
 9 ft* 6 in* X 5 ft* 6 in* (See Plate 37B and p* 48,) 
 
 Warp : Two-ply white wool* On two levels* 17 to i inch* 
 
 Weft : Red wool. Two shoots after each row of knots* 
 
 Knots : Wool* Ghiordes* 10 to i inch* 85 to the square inch* 
 Colours : Ten* Red (field) : dark blue (panels) : green : white : 
 light red (border) : dark blue-green : light blue : blue : black : 
 greenish yellow* 
 
 Asia Minor Carpet (Victoria and Albert Museum)* i6th century* 
 7 ft* X 6 ft* 7 in* (See Plate 39 and p* 47*) 
 
 Warp : Four-ply yellow wool* On two levels* 32 to i inch* 
 Weft : Crimson wool* Two shoots after each row of knots* 
 Knots : Wool. Sehna* 13 to i inch* 210 to the square inch* 
 Colours : Eight* Crimson (field) : deep yellow : light blue (ist 
 border) : white : yellow (2nd border) : yellow- green : green : 
 dark-blue* 
 
 Asia Minor Carpet (Victoria and Albert Museum)* i6th century. 
 9 ft* 4 in* X 8 ft* 3 in* (See Plate 40 and p* 47*) 
 
 Warp : Two-ply green silk* On two levels* 41 to i inch* 
 
 Weft : Crimson silk* Two shoots after each row of knots* 
 
 Knots : Wool and white cotton* Sehna* 16 to i inch* 330 to 
 the square inch* 
 
 Colours : Seven* Deep crimson (field and border) : white : dark 
 blue (panels in border) : yellow : green : light blue : black* 
 
 Asia Minor Carpet (Mr* George Mounsey)* i6th century* 6 ft* 
 II in. X 6 ft. I in* (See Plate 42 and p* 48*) This carpet is made 
 up from a larger one^ of which there is another piece in the Circulation 
 Collection of the Victoria and Albert Museum* 
 
 Warp : White wool* On one level* 15 to i inch* 
 
 107 
 
HAND-WOVEN CARPETS 
 
 Weft : Red wool. Two shoots after each row of knots. 
 
 Knots : Wool. Ghiordes. 9 to i inch. 70 to the square inch. 
 
 Colours : Nine. Red (alternate squares of the field) : dark green 
 (alternate squares of the field) : white (pattern of border) : 
 deep crimson-red (ist border) : green : black : light blue : 
 blue : greenish yellow. 
 
 Asia Minor Carpet (Victoria and Albert Museum). i6th century. 
 
 5 ft. 8 in. X 3 ft. 7 in. (See Plate 43B and p. 48.) 
 
 Warp : White wool. On one level. 16 to i inch. 
 
 Weft : Red wool. Two shoots after each row of knots. 
 
 Knots : Wool. Ghiordes. 12 to i inch. 95 to the square inch. 
 
 Colours : Seven. Red (field ; ist border) : yellow (pattern on 
 ground ; 2nd border at sides) : white (pattern of 2nd border) : 
 green-blue (2nd border at ends) : light blue (3rd border) : 
 blue : black. 
 
 Asia Minor Carpet (Victoria and Albert Museum), ifith-iyth 
 
 century. 17 ft. 4 in. X 8 ft. 2 in. (See Plate 44 and p. 49.) 
 
 Warp : White wool. On two levels only slightly separated. 18 to 
 I inch. 
 
 Weft : Red wool. Two shoots (sometimes three) after each row 
 of knots. 
 
 Knots: Wool. Ghiordes. ii to i inch. 100 to the square inch. 
 
 Colours : Six. Red (field and 2nd border) : dark blue (central 
 medallion, pattern on ground, and ist border) : dark green- 
 blue (side medallions) : yellow (3rd border) : light blue : black. 
 
 Asia Minor Mat (Mr. George Mounsey). 17th century. 
 
 3 ft. 6 in. X 2 ft. I in. (See Plate 46B and p. 55.) 
 
 Warp : Two-ply white wool. On two levels. 17 to i inch. 
 
 Weft : Brown wool. Two shoots after each row of knots. 
 
 Knots: Wool. Ghiordes. ii to i inch. 90 to the square inch. 
 
 Colours : Eight. Red (field) : white (end-border) : blue : light 
 
 red : light blue : green : purple : grey-brown. 
 
 Oriental Carpet (Mr. George Mounsey). i6th century. 6 ft. X 
 
 4 ft. 6 in. (See Plate 47 and p. 51.) 
 
 108 
 
THE TECHNIQUE OF HAND-WOVEN CARPETS 
 
 Warp : White wool* On two levels. 22 to i inch. 
 
 Weft : Brown and red wool. Two shoots after each row of knots. 
 Knots : Wool. Sehna. 8 to i inch. 80 to the square inch. 
 Colours : Nine. Red (field) : light blue (2nd border) : blue (ist 
 and 3rd borders) : white : yellow : brown : black : dark 
 
 blue : purple. 
 
 Oriental Carpet (Mr. George Mounsey). i6th century. 
 6 ft. 3 in. X 4 ft. 10 in. (See Plate 48 and p. 51.) 
 
 Warp : Fine yellow wool. On two levels. 23 to i inch. 
 
 Weft : Red wool. Two shoots after each row of knots. 
 
 Knots : Wool. Sehna. 10 to i inch. 120 to the square inch. 
 Colours : Six. Crimson (field and ist and 3rd borders) : green : light 
 blue (central medallion and 2nd border) : yellow : white : black. 
 
 Asia Minor Prayer-carpet (Mr. George Mounsey). Late 17th 
 century. 5 ft. x 3 ft. 9 in. (See Plate 56 and p. 55.) 
 
 Warp : Two-ply white wool (ends dyed yellow). On one level. 
 17 to I inch. 
 
 Weft : Red wool. Two shoots after each row of knots. 
 
 Knots : Wool. Ghiordes. 8 to i inch. 65 to the square inch. 
 Colours : Five. Red (field) : yellow (2nd border) : white (ist 
 border and spandrels) : black : light blue. 
 
 Asia Minor Carpet (Mr. George Mounsey). i8th century. 
 13 ft. I in. X 3 ft. 7 in. (See Plate 58 and p. 56.) 
 
 Warp : White wool. On two levels not much separated. 17 to 
 I inch. 
 
 Weft : Red wool. Two shoots after each row of knots. 
 
 Knots : Wool. Ghiordes. 10 to i inch. 160 to the square inch. 
 Colours : Eight. Red (field) : green (border) : purple (extra end- 
 borders) : dull yellow : blue : white : black : dark blue. 
 
 Chinese Carpet (Lady Cunliffe). 17 ft. ii in. x 4 ft. (See Plate 
 66a and p. 64.) 
 
 Warp : White or pale blue cotton. On two levels. 19 to i inch. 
 Weft : White cotton. Three shoots after each row of knots. 
 Knots : Silk. Sehna. 6 to i inch. 60 to the square inch. 
 Metal Thread : Woven on the upper leaf of the warp by a process 
 similar to Soumak weaving. 
 
 109 
 
HAND-WOVEN CARPETS 
 
 Colours : Seven* Rose-red, faded (border) : blue : green : yellow : 
 light blue : white : black* The ground of the niches is in 
 gold, and of the spandrels in silver thread. 
 
 Chinese Silk Carpet (Lieut*-Col* G* B* Croft Lyons, F*S*A*)* 
 
 5 ft* X 3 ft* 2 in* (See Plate 66b and p* 66*) 
 
 Warp : White cotton* On one level* 17 to i inch. 
 
 Weft : White cotton* Two shoots after each row of knots* 
 
 Knots : Silk* Sehna* 8 to i inch* 70 to the square inch* 
 
 Colours : Four* Yellow (field) : green-blue (border) : blue (pattern 
 on border) : black* 
 
 Chinese Carpet (Victoria and Albert Museum). i8th century. 
 
 6 ft* 2 in* X 8 ft* 5 in* (See Plate 68 and p* 64.) 
 
 Warp : Four-ply white cotton* On one level* 14 to i inch. 
 
 Weft : White cotton* Two shoots after each row of knots* 
 
 Knots : Wool* Sehna* 7 to i inch* 55 to the square inch. 
 Clipped round some of the contours* 
 
 Colours : Eight* White (field) : dark blue (3rd border and edging) : 
 rose-red (ist border) : light rose-red : yellow : blue : deep 
 yellow : light purple-red* 
 
 Chinese Carpet (Victoria and Albert Museum). i8th century. 
 
 9 ft. 3 in* X 7 ft* I in* (See Plate 69 and p* 64.) 
 
 Warp : Four-ply white cotton* On one level* 12 to i inch. 
 
 Weft : White cotton* Two shoots after each row of knots. 
 
 Knots : Wool* Sehna* 5 to i inch* 30 to the square inch* 
 
 Colours : Six* Rose-red (field) : dark blue (2nd border and edging) : 
 yellow (ist border) : blue : white : deep yellow* 
 
 Chinese Knotted Seat-Cover (Victoria and Albert Museum). 
 
 ifith-igth century. 4 ft* 10 in* x 2 ft* 3 in* (See Plate 70A and p* 65*) 
 
 Warp : Three-ply white cotton* On one level* 13 to i inch. 
 
 Weft : White cotton. Two shoots after each row of knots. 
 
 Knots : Wool. Sehna* 7 to i inch* 45 to the square inch. 
 Clipped round many of the contours. 
 
 Colours : Eleven. Orange-brown (field and border) : dark blue : 
 blue : light purple-red : light red : yellow-green : greenish 
 
 yellow : light blue : white : grey : purple-grey. 
 
 no 
 
THE TECHNIQUE OF HAND-WOVEN CARPETS 
 
 Chinese Carpet (Victoria and Albert Museum), i8th century. 
 4 ft. 4 in. X 2 ft. 2 in. (See Plate 70B and p. 64.) 
 
 Warp : Four-ply white cotton. On one level. 13 to i inch. 
 
 Weft : Cream-coloured cotton. Two shoots after each row of 
 
 knots. 
 
 Knots : Wool. Sehna, 7 to i inch. 45 to the square inch. 
 
 Colours : Eight. Salmon-red (field) : dark blue (border) : blue : 
 yellow : deep yellow : white : light blue : light salmon-red. 
 
 Chinese Carpet (Victoria and Albert Museum). Probably i8th 
 century, ii ft. 8 in. x 8 ft. 4 in. (See Plate 71 and p. 65.) 
 
 Warp : White cotton. On one level. 13 to i inch. 
 
 Weft : White cotton. Two shoots after each row of knots. 
 
 Knots : Wool. Sehna. 5 to i inch. 35 to the square inch. 
 
 Colours : Six. White (field) : dark blue (pattern on fields ist border 
 and edging) : blue : deep yellow : apricot-red : yellow. 
 
 Chinese Pillar Carpet (Victoria and Albert Museum). i8th 
 century. 8 ft. 3 in. x 3 ft. 3 in. (See Plate 72 and p. 65.) 
 
 Warp : White cotton. On one level. 12 to i inch. 
 
 Weft : White cotton. Two shoots after each row of knots. 
 
 Knots : Wool. Sehna. 6 to i inch. 38 to the square inch. 
 Clipped round many of the contours. 
 
 Colours : Eleven. Purple-red (ground) : yellow : deep yellow : 
 greenish yellow : white : red : light red : very light red : 
 blue : dark blue : grey-brown. Some colours are hard to 
 distinguish owing to fading. 
 
 Chinese Pillar Carpet (Victoria and Albert Museum). i8th 
 century. 7 ft. 8 in. x 4 ft. 2 in. (See Plate 73 and p. 65.) 
 
 Warp : White cotton. On one level. 13 to i inch. 
 
 Weft : White cotton. Two shoots after each row of knots. 
 
 Knots : Wool. Sehna. 5 to i inch. 32 to the square inch. 
 
 Clipped round many of the contours. 
 
 Colours : Twelve. Yellow (ground) : dark blue : white : light 
 blue : red : greenish yellow : deep yellow : light red : light 
 yellow : purple-red : very light red : grey-black. Some colours 
 are hard to distinguish owing to fading. 
 
 Ill 
 
HAND-WOVEN CARPETS 
 
 Eastern Turkestan Rug (Victoria and Albert Museum). iSth-igtli 
 century. 3 ft. 7 in. x 2 ft. 9 in. (See Plate 75B and p. 66.) 
 
 Warp : Five-ply white cotton. On two levels. 15 to i inch. 
 
 Weft : Three-fold white cotton (some brown wool). Three shoots 
 after each row of knots. 
 
 Knots : Wool. Sehna. 6 to i inch. 40 to the square inch. 
 
 Colours : Seven. Red (field) : blue (2nd border) : yellow : dark 
 brown (ist and 3rd borders) : ochre : light blue : white. 
 
 Spanish Carpet (Victoria and Albert Museum). i6th century. 
 14 ft. II in. X 6 ft. 10 in. (See Plate 77B and p. 70.) 
 
 Warp : Two-ply white wool. 25 to i inch. 
 
 Weft : White wool. One shoot after each row of knots. 
 
 Knots X Wool. Single-warp, ii to i inch. 135 to the square 
 inch. 
 
 Colours : Eight. Red (field and 2nd border) : dark green (ist 
 border) : white : blue : yellow : orange : light blue : black. 
 
 Spanish Carpet (Victoria and Albert Museum). 15th century. 
 
 6 ft. 8 in. X 4 ft. (See Plate 78 and p. 69.) 
 
 Warp : Two-ply white wool. 21 to i inch. 
 
 Weft : Threefold white wool. One shoot after each row of knots. 
 
 Knots : Wool. Single-warp. 10 to i inch, no to the square 
 
 inch. 
 
 Colours : Seven. Red (field) : blue (2nd border) : dark green- 
 blue : yellow : black (ist border) : white : orange. 
 
 Spanish Carpet (Victoria and Albert Museum). Late 15th century. 
 
 7 ft. 6 in. X 5 ft. 4 in. (See Plate 79 and p. 71.) 
 
 Warp : White wool. 20 to i inch. 
 
 Weft : White wool. One shoot after each row of knots. 
 
 Knots : Wool. Single-warp. 1 1 to i inch. 1 10 to the square 
 
 inch. 
 
 Colours : Seven. Red (field) : green : dark blue : white (3rd 
 
 border and ist extra end-border) : blue (2nd border) : black 
 (ist border and 2nd and 3rd extra end-borders) : yellow. 
 
 Spanish Carpet (the Hon. H. D. McLaren, C.B.E., M.P.). Late 
 15th century. 15 ft. X 2 ft. ii in. (See Plate 80 and p. 71.) 
 
 Warp : White wool. 16 to i inch. 
 
 112 
 
THE TECHNIQUE OF HAND-WOVEN CARPETS 
 
 Weft : White wooL One shoot (fourfold) after each row of knots. 
 Knots : Wool. Single- warp. lo to i inch. 8o to the square 
 inch. 
 
 Colours : Seven. Black (ground) : light blue : yellow : white : 
 light red : purple : green. 
 
 Spanish Carpet (Victoria and Albert Museum), i6th century. 
 9 ft. 7 in. X 5 ft. II in. (See Plate 8i and p. 71.) 
 
 Warp : White wool. 33 to i inch. 
 
 Weft : Red wool. One shoot (threefold) after each row of knots. 
 Knots : Wool. Single- warp. 16 to i inch. 270 to the square 
 inch. 
 
 Colours : Six. Blue (field) : red (border and panels) : yellow 
 
 (pattern on border and panels) : white (pattern on field) : light 
 blue : green. 
 
 Finnish Rug (Victoria and Albert Museum). Dated 1799. (See 
 Plate 84 and p, 73.) 
 
 Warp : White flax. On one level. 10 to i inch. 
 
 Weft : Fine white wool. Four shoots after each row of knots. 
 Knots : Wool. Ghiordes. 2 to i inch, ii to the square inch. 
 There is also a pile at the back of the rug^ formed by white 
 woollen knots tied after every 8 shoots of weft. 
 
 Colours : Eight. Rose-red (border) : black (field) : blue : light 
 blue : white : green : yellow : light crimson. 
 
 Two French Savonnerie Carpets. Late 17th century. (Mobilier 
 National^ Paris.) 15 ft. 10 in. X 29 ft. and 12 ft. 7 in. X 28 ft. 9 in. 
 (See Plates 85 and 86 and p. 74.) 
 
 Warp : Flax. On two levels. 22 to i inch. 
 
 Weft : Flax. Three shoots after each row of knots. 
 
 Knots : Wool. Ghiordes. 10 to i inch, no to the square inch. 
 Colours : Many. 
 
 English Knotted Panel (Victoria and Albert Museum). Dated 
 1600. I ft. 9 in. X 3 ft. 6 in. (See Plate 87B and p. 79.) 
 
 Warp : Three-ply hemp. On one level. 13 to i inch. 
 
 Weft : Three-ply hemp. One shoot after each row of knots. 
 Knots : Wool. Ghiordes. 8 to i inch. 55 to the square inch. 
 
 113 
 
 I 
 
HAND-WOVEN CARPETS 
 
 Colours : Nine, Dark green (ground) : blue : crimson : red : 
 
 yellow : white : orange : light crimson : black, 
 
 English Carpet (Victoria and Albert Museum), Dated 1603, 
 16 ft, 8 in, X 7 ft, 8 in, (See Plate 88 and p, 79,) 
 
 Warp : Hemp, 20 to i inch. 
 
 Weft : Hemp, One shoot after each row of knots. 
 
 Knots : Wool, Ghiordes, 9 to i inch, 95 to the square inch. 
 Colours : Seven, Dark green-blue (field) : green : yellow : white : 
 crimson : blue : orange, 
 
 English Carpet (Lady Hulse), Dated 1614, ii ft, 9 in, x 8 ft, 
 5 in, (See Plate 89 and p, 79,) 
 
 Warp : Hemp, 14 to i inch. 
 
 Weft : Hemp, One shoot after each row of knots. 
 
 Knots : Wool, Ghiordes, 9 to i inch, 60 to the square inch. 
 Colours : Many, Shades of green, yellow, red and blue : purple : 
 white : black, 
 
 English Carpet made at Exeter, Dated 1757, 14 ft, 9 in, x 
 
 12 ft, (See Plate 90 and p, 81.) 
 
 Warp : White wool. On two levels, 19 to i inch. 
 
 Weft : Wool of various colours. Two shoots after each row of 
 knots. 
 
 Knots : Wool, Ghiordes, 8 to i inch, 80 to the square inch. 
 Colours : Many bright colours. The ground is blue, 
 
 English Carpet at Syon House (The Duke of Northumberland), 
 Dated 1769, 34 ft, 5 in, X 14 ft, 2 in, (See Plate 91 and p, 81,) 
 
 Warp : White wool. On two levels not much separated, 9 to 
 I inch. 
 
 Weft : Black (some coloured) wool. Two shoots after each row 
 of knots. 
 
 Knots : Wool, Ghiordes, 4 to i inch, 17 to the square inch. 
 Colours : Twenty or more, 
 
 (c) Materials 
 
 It may be stated that nearly every raw material that can be used 
 for textile purposes has been employed in the manufacture of carpets, 
 
 1 14 
 
THE TECHNIQUE OF HAND-WOVEN CARPETS 
 
 but many are so rarely met with that they can be left out of practical 
 consideration. The important textile fibres — wool, cotton and silk, 
 made into yarn — constitute the whole substance of most carpets, and 
 if to these be added flax, hemp, jute, and metal thread, every other 
 material can be neglected. Of the above substances, wool is by far the 
 most important. It wears well, dyes easily and is pleasant in appearance, 
 and as it can be used for warp, weft and pile, many carpets are made 
 of it alone. The wool of both sheep and goats is used, and in 
 practice a distinction need not be made between the kinds. The natural 
 colour of wool ranges from white to black, and in substance it passes 
 by imperceptible gradations from fine fibres, easily mistaken for silk, 
 to long, coarse hair resembling that of the horse. The wool, or hair, 
 as it is more often called, of the camel is also largely used, but it is not 
 easy to dye and is mostly left in its natural colour. When wool is used 
 for warp, that of a comparatively coarse staple is chosen, and it is usually 
 spun into a rather hard-twisted two-ply yarn, which is sometimes dyed, 
 though more often left white, or whatever its natural colour may 
 be. On the other hand, a woollen weft is mostly dyed — red or 
 brown being favourite colours — and is only slackly twisted, or perhaps 
 may consist of two or more strands not twisted together at all. For 
 the pile, wool stands pre-eminent, and it is here that the finest and most 
 lustrous kinds are most esteemed. It is, of course, mostly dyed, and 
 consists of yarn of two or more strands, not tightly spun. 
 
 After wool, cotton is the fibre most used. For the warp it is usually 
 undyed and consists of several strands spun together. For the weft 
 it is often dyed blue or pink and is loosely spun, an exception being that 
 the very sinuous shoot between two straight ones (Fig. h) must be 
 a fine and tightly-spun yarn. Although admirably suited for warp and 
 weft, cotton is less satisfactory for the pile. It wears well, but is dead 
 in appearance and tends to look dirty in course of time ; and though 
 it can be dyed without difficulty the colours fade and lose their beauty 
 sooner than those on wool. For these reasons it is seldom used for the 
 pile, though it is found in the rugs of Multan, and sometimes, in small 
 quantities, in old Turkish rugs, where a pure white colour is required. 
 
 Silk is a material adaptable for all purposes, though it has not the 
 wearing qualities of wool. Even in the East it is costly and can only 
 be used in expensive carpets. Sometimes a rug made entirely of silk is 
 seen, but more often, even when the pile is of silk, the warp and weft 
 are of cotton, which in modern rugs is often mercerized. On the other 
 
 115 
 
HAND-WOVEN CARPETS 
 
 hand^ in many of the finest old Persian carpets a silken warp and weft 
 is found associated with a woollen pile. For this there is a sound reason 
 in the fact that a very fine yarn of reeled silk is stronger than a 
 woollen one of similar si2;e^ whereas a coarser woollen yarn^ even if not 
 so strong as a silken one, has at least an ample margin of strength. 
 
 Of the less common materials, flax, or linen, is occasionally substituted 
 for cotton in Eastern carpets, but its occurrence may be regarded as almost 
 accidental. Coarse flax, or perhaps hemp, was the material used for 
 the foundation of the earliest English knotted carpets. Jute is only 
 used in very inferior hand-made carpets, though of wide employment 
 in machine-made ones. The last material on the list — metal thread 
 — stands on a different footing from the rest. It is sometimes used 
 in parts of the surface of a carpet to give an unusually rich appearance. 
 The thread consists, as a rule, of silk or other yarn covered over with 
 a twisted strip of silvered or gilt metal. It is never used for the foundation 
 of a carpet and is never knotted, but whether employed in a Kilim or in 
 a pile-carpet it is inserted either by the tapestry method, as used for the 
 weft of an ordinary Kilim, or else by a process very similar to that of 
 Soumak weaving. Metal thread had the disadvantage of tarnishing badly 
 and is almost unknown in modern pile-carpets, 
 
 (d) Dimensions and Shape 
 
 Even the size and shape of a carpet allow some conclusions to be 
 drawn with regard to its provenance or classification. Very large carpets 
 are certainly only made by non-nomadic peoples : whereas a want of 
 regularity in shape, such as a difference between the widths of the ends 
 or a tendency to lie unevenly, is evidence that the carpet in question 
 was made by nomads, A carpet woven on a roughly-made loom, or 
 one which had to be taken out of the loom and remounted several times, 
 or rolled up for weeks on end, can scarcely be as regular in appearance 
 as one made under more stable conditions with better appliances. The 
 size of the tents used by nomadic people limits the dimensions of their 
 carpets, as it is impossible for them to contain a tall or wide loom. 
 Moreover, nomadic people have no need themselves for large and heavy 
 carpets which are difficult of transport, and nomad carpets were, until 
 quite recently, only manufactured for their own use. For this reason 
 the size of their carpets is narrowly limited ; the width varying from 4 to 
 5|- ft., and the length rarely exceeding two and a half to three times the 
 width. But even the carpets of non-nomadic people, as long as only 
 
 116 
 
THE TECHNIQUE OF HAND-WOVEN CARPETS 
 
 their own consumption was catered for, tended to have certain definite 
 proportions. For instance, the rooms of Persian houses are for certain 
 reasons generally oblong in shape, and a conventional arrangement of 
 the floor-covering becomes customary (Fig. j). I is the show-piece 
 {khali), the borders of which are covered by the other three carpets in 
 order to preserve it, with the result that the long and narrow carpets 
 II and III — known as kanara, or in English often as runners — are 
 largely used in passing up and down the room. IV, the divan- cover 
 (sedjadeh) is used most of all, for members of the household and visitors 
 sit upon it. 
 
 As the European demand increased, 
 and as the custom grew for buyers 
 to commission carpets in Persia and 
 Asia Minor, they naturally gave the 
 local workers not only instructions as 
 to the pattern but also as to the shape 
 and si2;e they required. Carpets so 
 ordered were designed to correspond 
 with European requirements, and to- 
 day carpets of any proportions and 
 dimensions may be obtained by any- 
 one who orders them. 
 
 Whereas rugs for use on the floor 
 are generally rectangular in shape, 
 other articles, such as bags, saddle- 
 covers or tent-bands, are naturally 
 of such shapes and sizes as their special use dictates. Even rugs for 
 Oriental use are sometimes found with rounded corners, re-entrant angles 
 or other peculiarities of shape, and were obviously made thus in order 
 to fit into special positions. In the Victoria and Albert Museum there 
 is a fragment of a Persian border which, running partly at an angle of 
 forty-five degrees across the warp and weft and partly parallel to the 
 weft, apparently came from the corner of a shaped carpet. 
 
 An average size for kanara, or runners, is about 3 X 15 ft. A very 
 common size for rugs is about 4I X ft., most prayer-rugs being of 
 these dimensions. Larger than this, rugs can be found of all sizes up 
 to about 15 X 20 ft., above which size specimens are rarer. That still 
 larger carpets were made even centuries ago is witnessed by the great 
 Ardabil carpet, which measures 17^^ X 34^ ft., and the carpet made for 
 
 117 
 
 
 zr 
 
 
 r" 
 
 
 “J 
 
 I 
 
 1 
 
 1 
 
 1 
 
 1 
 
 1 
 
 1 
 
 \ 
 
 1 
 
 1 
 
 IL ' 
 
 1 
 
 1 
 
 1 
 
 1 
 
 1 
 
 1 
 
 1 
 
 1 
 
 ! 
 
 ! 
 
 1 
 
 1 
 
 1 
 
 z 
 
 1 
 
 1 
 
 1 
 
 1 
 
 1 
 
 1 
 
 I 
 
 1 
 
 I 
 
 ! ^ 
 
 1 
 
 1 
 
 1 
 
 1 
 
 1 
 
 1 
 
 1 
 
 1 
 
 1 
 
 1 
 
 1 
 
 1 
 
 1 
 
 
 1 
 
 FIG. J.— PLAN OF PERSIAN ROOM. 
 
HAND-WOVEN CARPETS 
 
 the Palace of the Forty Columns at Ispahan^ which, judging from the 
 scale and pattern of the remaining fragments and the si2;e of the hall where 
 it was placed, must have been four times the size of the Ardabil carpet, 
 
 (e) Colour and Dyeing 
 
 The chief attraction of Oriental carpets is perhaps to be found in 
 the charm of their colouring. This view is supported by a study of 
 some of the Nomadic rugs, whose simple geometrical patterns, though 
 inoffensive and even well-balanced, possess little artistic interest when 
 seen in monochrome, but, when rendered in colour with Oriental skill, 
 at once become endowed with a beauty at once satisfying and difficult 
 to analyse. An incredible richness of colour and wealth of colour-com- 
 bination is to be found in Oriental rugs. Between the dark Beluchistan, 
 whose sombre character is all the more noticeable because of the white 
 patches it contains, and the bright Ghiordes with its delicate prismatic 
 hues, one may find in rugs every colour with which a kind Heaven has 
 endowed the flowers of our fields. And as the flowers, however brilliant 
 and varied their colours, always harmonize with each other, so the colouring 
 of old Oriental rugs never strikes us as glaring, in spite of its many and 
 powerful contrasts. 
 
 To say that the Oriental has an instinctive power of managing colour 
 is not to give him more than his due, but it must be remembered that 
 instinct is of little service when traditional methods are abandoned, and 
 even when his instinct is admitted it goes a very short way towards explaining 
 how his effects are produced. 
 
 One reason why harshness of contrast is avoided is that strongly 
 contrasting colours are usually separated by a line of neutral tint, such 
 as black or white, or of some intermediate colour ; but though this device 
 may be effective against a close inspection, it can scarcely be sufficient 
 when the carpet is viewed from a distance, as then the effect of such a 
 line would be imperceptible. 
 
 A better explanation probably lies in the fact that the colours 
 generally used are far darker in tone and much less vivid in hue than 
 is realized by the eye. A knot removed from a carpet and viewed apart 
 from it looks far greyer and darker than the patch of colour of which 
 it formed part ; and if an attempt is made to match a colour in a rug by 
 the eye alone, unaided by actual contact of the yarns, it is surprising 
 what an error can be made. What looks a pure white in a carpet will 
 often turn out to be, when contrasted with white paper, a half-tone of 
 
 ii8 
 
THE TECHNIQUE OF HAND-WOVEN CARPETS 
 
 considerable depth. Manufacturers complain of the great difficulty they 
 find in getting the whites of their carpets rights and this difficulty possibly 
 is due to the fact that even experience shrinks from using yarn so much 
 darker than the whites appear in the model which is being copied^ or 
 are wished to look in the new carpet. The extreme crudeness to be seen 
 in many machine-made carpets will often be found on a critical examination 
 to be due not so much to an incorrect disposition of the colours as to 
 an excess of brilliancy in the tones of the dyes. 
 
 It is not, however, only by avoiding excessive contrasts that 
 harmonious results are achieved. The colours must be in correct 
 proportions and in patches of the right si2;e, and in these respects the 
 Oriental owes much to his willingness to follow tradition. He uses 
 much the same dyes as his fathers, he reproduces the same paper designs, 
 or copies successful rugs without materially altering the scale or pro- 
 portions of their patterns ; and though the small changes he makes may 
 gradually lead to the evolution of new types, yet he is always ready to 
 abandon any modification which does not result in improvement. The 
 Oriental does not indulge in that wild striving for originality now so pre- 
 valent in the West, which, though sometimes leading to fresh beauty 
 and inspiration, is too often rewarded only by disappointment. When 
 all this is said, it still is likely that many old carpets were less pleasing 
 when new than they are to-day. Time, bringing with it fading and wear, 
 must have had much effect in adding softness and subtlety to many of 
 the more brilliant carpets, but it certainly will not do to ascribe their 
 delightful effect chiefly to time's mellowing influence. 
 
 There is no mystery about the nature of the dyestuffs used in the 
 past, even if we do not know the exact details of their manipulation ; 
 and, of course, many of the tales told — for instance, that the wonderful 
 colours were due to the grinding up of rubies and sapphires — will not 
 bear serious consideration. 
 
 The blues are probably invariably derived from indigo, which, 
 as is well known, is obtained by steeping the leaves of the indigo plant 
 in water. The reds are produced by madder — the root of a plant — 
 treated with various mordants ; or by Kermes, which consists of the 
 dried bodies of insects that live on the oak or on a species of cactus. In 
 later times cochineal, the Mexican equivalent of Kermes, has been 
 imported as a satisfactory substitute. No natural yellow dyestuff is quite 
 as permanent as those for red and blue, but saffron, Persian berries and 
 turmeric have all been used successfully. Catechu (or Cutch) gives a 
 
 119 
 
HAND-WOVEN CARPETS 
 
 good brown colour^ and black may be obtained by logwood in conjunction 
 with an iron salt such as ferrous sulphate. This last dye has a somewhat 
 corrosive effect upon wool, and in course of time may cause it to disappear 
 completely. Natural black wool may be used instead, but like all undyed 
 animal fibres will fade rather quickly. 
 
 The secondary colours are produced by dyeing successively with 
 two of the primaries, but they never seem to last quite as well as the 
 latter or to be quite so rich, and it is no doubt on this account that they 
 are used in comparatively small proportions in old carpets. 
 
 Though the general nature of the dyestuffs used is fairly constant 
 throughout the East, yet various districts specialise in the production 
 of certain shades of colour, either from choice or because local peculiar- 
 ities, such as the quality of the water or facilities for obtaining particular 
 grades of dyestuffs, make the production of such shades easy. The 
 brick-reds of Asia Minor, the rich crimsons of Persia, and the deep wine- 
 reds of Central Asia are notable instances ; but there are many subtler 
 colours which are highly characteristic of, and often peculiar to, certain 
 districts, and accordingly most useful for purposes of identification. 
 
 European manufacturers assert, and not without justification, that the 
 methods of dyeing in Europe are far superior to the primitive ones of 
 the East. It is a fact that carpets are hardly ever found without differences 
 of shade in colours which were intended to be the same throughout, 
 and often the difference is so great that, if it did not occur in a way quite 
 unconnected with the design, as for instance in the plain ground, it might 
 be supposed that different hues were really intended. 
 
 The explanation of this phenomenon is that the primitive methods 
 employed in dyeing are far too unscientific to ensure the same result 
 on different occasions. It is difficult for nomadic tribes to carry supplies 
 of wool sufficiently large to obviate the necessity for dyeing more before 
 the carpet is finished, but even if they do so, their custom, forced upon 
 them by necessity, of using small pots for vats, prevents them from mixing 
 much of the dye at one time. These difficulties are still present, though 
 in a lesser degree, with the sedentary weavers, so that the irregularities 
 of dyeing are less noticeable in their work. 
 
 The differences of colour then, which can be seen like stripes running 
 right across the carpet, are by no means intentional but are inseparable 
 from the whole method of production. This element, which would be 
 found annoying in a modern European carpet, gives an additional charm 
 to the simple product of the East, just as we delight in the twisted streets 
 
 120 
 
THE TECHNIQUE OF HAND-WOVEN CARPETS 
 
 and crooked roofs and corners of an old town, whereas the slightest 
 irregularity or the smallest incline in the straight front of a new building 
 would be intolerable to the eye. 
 
 Leaving out the unintentional variations of colour, it will be found 
 that the number of different hues and shades in a single Oriental carpet 
 is by no means large, ranging as it does from four to about twelve. 
 A very large proportion of the area of a carpet is worked in three or four 
 colours, and three or four others may be used in moderate quantity; 
 but if still more are employed they are almost negligible in amount. 
 
 In spite of all their disadvantages it is generally admitted that 
 the Oriental dyer of the past produced colours which for beauty and 
 permanence have never been excelled ; but unfortunately the superiority 
 of the Oriental dyes does not exist now in the same degree as it did in 
 bygone days. The later colours are often very glaring, and many of 
 them will not stand washing or exposure to the sun. This deterioration 
 is the result of Western influence, and its beginnings can be traced fairly 
 accurately to the years 1865-70. During that period aniline dyes, 
 which were invented in 1859, began to be used in the East. Their 
 brilliancy, the ease with which they could be mixed, as well as their 
 comparative cheapness, seduced the Oriental dyer just as they did his 
 brother in the West. These brilliant colours, unfortunately, are often 
 not permanent ; it is probably safe to say that those first introduced 
 never were. Many of them fade quickly in sunlight and come off when 
 washed or even wetted. Naturally the employment of such colours 
 in a carpet is simply disastrous, and the disadvantage has been recognised 
 both in the West and the East* In 1910 an embargo was placed in 
 Persia upon the import of chemical dyes either in solid or liquid form, 
 but, nevertheless, these materials are still used considerably, largely on 
 account of the ease with which they can be applied. 
 
 The worthlessness of the early aniline dyes for producing permanent 
 colours led to a prejudice against all synthetic dyes, of which the aniline 
 colours are but one small branch, a prejudice which is very slow to dis- 
 appear. And yet the chemists have made immense strides since Dr. 
 Perkin discovered mauve. The chemical nature of many of the vegetable 
 dyestuffs (including indigo and madder) has been discovered ; it is now 
 possible to make them synthetically on a commercial scale, and the product 
 is every bit as good as the original, except perhaps so far as impurities 
 in the latter may give a pleasing irregularity. More important is the 
 fact that new dyes are being discovered which surpass the older ones 
 
 121 
 
HAND-WOVEN CARPETS 
 
 in point of resistance to light and other influences, and which supply 
 the designer with fresh colours, even if the colours are not more beautiful 
 than those already known. In spite of these facts — possibly in ignorance 
 of them — it is very common still for writers to use the arguments levelled 
 against the aniline dyes of more than forty years ago ! It is so certain 
 that the future lies with synthetic dyes that it seems a pity to entertain 
 any prejudice on the subject or to do anything which is likely to retard 
 their proper development. At the same time, unsound dyes, whether 
 natural or chemical, should, of course, be discouraged in every way, 
 and buyers can do much to abolish the cruelly wasteful system by which 
 good materials and much labour are thrown away in producing an article 
 which will not stand the test of time. 
 
 The vividness of some modern dyes, quite apart from the question 
 of their soundness, and the consequent garishness of rugs in which they 
 are employed, has led to the baneful practice of chemically washing 
 the surface in order to tone down the colours and produce immediately 
 a mellow effect. The processes are kept secret as far as possible, but 
 they seem to consist of application of acids or of sea-water. The 
 immediate effect is often pleasing, and no doubt some of the milder 
 agents are not harmful ; but, unfortunately, it happens that the wool 
 is seriously damaged when the treatment has been severe. At its best 
 the method produces an effect which is better left to the slower influence 
 of time, and chemically washed rugs should be avoided if possible, though 
 some classes of the best modern rugs are practically never left quite 
 untouched. 
 
 It is rather consoling to reflect that even in older times many of 
 the troubles of the present must still have been existent. The best-dyed 
 rugs fade, and many magnificent old rugs may be seen in which some 
 of the colours have completely changed. Also there seem to be few 
 freshly dyed yarns which are quite unacted upon by water. Very fine 
 and undoubtedly genuine silk rugs of the seventeenth century were 
 recently sold in London, of which the colours came off easily on a damp 
 cloth. It is probable that they had never been cleaned. The usual 
 fate of a rug some centuries old is to be washed many times, and though 
 at first the colours may have come off to a slight extent, eventually no 
 such effect would be perceived, and the lightest colours would by that 
 time have become so toned that the slight contamination would be 
 invisible. It is not fair to condemn utterly a modern rug because its 
 pure whites have become slightly tinged with the adjacent colours. 
 
 122 
 
Chapter II 
 
 THE DESIGN OF CARPETS 
 
 A discussion of the design of carpets can well start with the 
 consideration of the ways in which carpets are mostly used, for the 
 proper treatment of the design must be based upon such a consideration* 
 In the first place it should be remembered that a carpet is usually placed 
 flat on the floor and that it is meant to be walked upon, that it is generally 
 seen from some distance away and that it is liable to be looked at from 
 all directions. 
 
 The first of these conditions points to the adoption of a design 
 which will not be impaired by being fore-shortened, and the second 
 indicates a conventional treatment as being the most suitable. A 
 naturalistic treatment always gives a certain suggestion of relief to a 
 design, and if the realism is carried to the extent of showing the shadows 
 of the objects portrayed, then it is impossible to avoid the uncomfortable 
 feeling that one has to walk on an uneven surface. 
 
 The third condition — that the carpet will often be seen from a 
 distance- — leads to the conclusion that the pattern must be fairly bold 
 to be effective. It does not follow that the carpet will never be near 
 the eye. As it is avowedly a work of art, it may at any time be examined 
 closely, and, apart from this, in the East, at all events, it is customary 
 to sit on the floor, in which case no great distance separates the carpet 
 from the eye. There is accordingly no logical reason against having 
 fine detail in the design as long as it is secondary to the general arrangement 
 of the masses. 
 
 In the next place, as the carpet may be seen from all sides, it is 
 advisable to avoid any important feature in the design which must, in 
 order to be appreciated, be viewed from a particular direction. This 
 rules out all pictorial subjects on a large scale from consideration as being 
 suita le for inclusion. If small pictorial subjects, human figures, animals 
 or trees are included, then they should be repeated several times in the 
 composition with different orientations, so that some will be upright 
 from any point of view. An exception may be made to this practice 
 if the piece is intended to be seen only from one direction. The chief 
 
 123 
 
HAND-WOVEN CARPETS 
 
 instance of this is the prayer-rug^ which is always deliberately planned 
 so that its design is best seen when it is in use* Tent-bags or door- 
 hangings would be other legitimate exceptions ; but as it happens that 
 these are mostly made by people who favour geometrical ornament, 
 the liberty that might be claimed is rarely exercised. Exceptions are 
 also found where they are perhaps not justified (see Plate 58), but in 
 these cases the effect of the rug is displeasing when seen from the wrong 
 point. 
 
 Another reason why deliberately pictorial subjects are unsuitable 
 is that parts of a carpet are often hidden by furniture or other objects, 
 and if part of the picture cannot be seen, the whole effect may be spoiled. 
 On the other hand, repeating patterns and devices, symmetrical in them- 
 selves or symmetrically arranged, are especially suitable for decorating 
 an object which may be partly concealed, for when such are employed 
 the mind can follow the design even where it eludes the eye. 
 
 The above argument in favour of a conventional, bold and symmetrical 
 treatment in the design of carpets may perhaps appear sufficient, but 
 it is not necessary to rely upon theory alone, for the conclusion is amply 
 borne out by the immemorial practice of the East. 
 
 The technique has also a certain influence upon design. It has 
 already been explained how in Kilims the change of colour in the weft 
 causes vertical slits in the fabric, which are mostly avoided by the 
 adoption of slanting or stepped outlines. This does not occur in pile 
 carpets, but there are limitations of another nature. The surface of a 
 pile-carpet is not, from the designer's point of view, capable of indefinite 
 subdivision, but it may be regarded as atomic in structure, and the 
 atom or unit of surface is a single knot. Every design must be interpreted 
 in relation to this unit. The knots each occupy a small square, or more 
 often a small oblong space, and they are arranged regularly in rows and 
 columns. The narrowest line, therefore, that can be made is one con- 
 sisting of a single line of knots ; and curves can only be represented 
 by a succession of straight lines, alternately vertical and horizontal, 
 whose shortest length is the height or width of one knot. These 
 considerations at once put a limit to the fineness of detail and the delicacy 
 of drawing which is possible with a knotted surface. The size of each 
 knot is dependent upon the fineness of the texture, so that the finer the 
 texture the smaller are the details that can be rendered and the smoother 
 are the curves. 
 
 This is no doubt the chief reason why the designs are more 
 
 124 
 
THE DESIGN OF CARPETS 
 
 conventional in the coarser carpets* A blossom that can be delicately 
 rendered in a fine carpet may, in one of coarse texture, be so angular 
 in outline as to become a mere geometrical rosette, so that in such 
 circumstances there is considerable temptation to rely exclusively upon 
 geometrical forms. 
 
 The colouring of carpets, no less than the design, tends to be 
 conventional, and perhaps a correspondence can be traced between 
 the degree of conventionalization in each. In the finer and more 
 naturalistic carpets the blossoms and leaves usually have their proper 
 colours ; in those rather more conventional there is a tendency to keep 
 the colours, red, white, yellow and blue for the flowers, green or 
 black for the leaves, and green, brown or black for the stems ; but in 
 carpets with geometrically drawn floral forms, the colours will be found 
 used indiscriminately and without any relation to the objects portrayed. 
 
 A most striking aspect of the designs of Eastern carpets — one not 
 peculiar to them among works of art and yet developed perhaps to 
 the greatest extent in their case — is the way in which the same 
 patterns are repeated again and again in rugs of a given period, and even 
 are handed down with slight modifications from age to age. It is not 
 often that a rug is found which has no well-known feature in the design, 
 and in fact the matter can be put more strongly, for it can be asserted 
 that very few carpets are met with in which every detail of the pattern 
 cannot be found with very little search in some other rug. 
 
 To illustrate this, it may be said that a collection of about 200 border- 
 stripes may be made which will include at least 95 per cent, of all those 
 to be found in Persian rugs, whether ancient or modern. 
 
 This persistence of pattern which runs throughout the art of the 
 East is due to the inclination of the craftsman to let well alone. It 
 might be thought at first that such a system would be unsatisfactory 
 because giving a feeling of want of variety in design and poverty of invention 
 in the weavers. Such a feeling, if it exists at all, is entirely outweighed 
 by the pleasure of meeting with designs already known and in tracing 
 the gradual changes that take place in them in the course of time. 
 The familiar has as much charm as the novel, and a design sanctioned 
 by custom can rarely be without merit. If all carpets were made with 
 entirely different designs, the greatest interest of the carpet-collector 
 would have vanished. 
 
 The adherence to type is, of course, most manifest in the rugs of 
 the same district, or made by people of the same race, and so is of great 
 
 125 
 
HAND-WOVEN CARPETS 
 
 value in the identification of carpets. It is rather the details of the design 
 than the general character that give the most help in this problem, but 
 one principle, which is of general application, is that geometrical designs 
 are most favoured by the nomadic and, consequently, the least cultured 
 people ; while a floral basis and especially a naturalistic treatment are 
 usually associated with a sedentary population having superior technical 
 appliances, and a greater incentive to the rendering of intelligible forms 
 induced by their manner of life. 
 
 An analysis of the design of carpets, having regard to the general 
 arrangement of the pattern as well as to the treatment of the details, is 
 of great use in acquiring familiarity with patterns found in the various 
 parts of the carpet-making world, and in learning the true principles 
 upon which the most pleasing designs are based. 
 
 The general arrangement and massing of the pattern of a carpet 
 may be considered quite apart from the nature and drawing of the details, 
 and will be a useful preliminary to the review of the latter. A great 
 majority of carpets are rectangular in shape, and the natural place first 
 to put ornament in an empty rectangle is round the edge. Thus at 
 once is the border introduced, and it will soon be noticed that nearly 
 every carpet has a border separated off from the rest. In Oriental carpets 
 the border goes round the whole carpet and is separated from the field 
 by straight boundary lines. Generally it is divided into from two to 
 about ten parallel bands, each of which contains a repeating pattern. 
 These bands (which may be called ** border-stripes run continuously 
 round sides and ends of the carpet, and generally each has the same 
 pattern throughout. The end borders have the same number of rows 
 of knots in their width as the side borders have, but as the knots are 
 mostly shorter in a vertical direction, it follows that the end borders 
 are generally narrower than the side borders, and it may be remarked 
 in passing that a quick comparison of the widths will usually indicate 
 the proportions of each knot. It very rarely, if ever, happens in Eastern 
 carpets that the ornament of either border or field encroaches across 
 the hard line that separates them, though in Western carpets this is done 
 very frequently. Sometimes there will be an extra and external border- 
 stripe at one or both ends of the carpet, but it is not common and only 
 occurs in rugs of Central Asia and very rarely in those of Turkey. An 
 extra internal border-stripe is also found at the ends of some Asia Minor 
 prayer-rugs. 
 
 The kinds of patterns found in the border-stripes may now be briefly 
 
 126 
 
THE DESIGN OF CARPETS 
 
 reviewed. In the narrowest stripes, consisting of a few rows of knots 
 or even a single row, there may be very simple geometrical patterns, 
 such as squares of alternate colours, Vandykes, barber^s-poles and the 
 like, all of which are found even in rugs of a naturalistic type. 
 
 In the wider stripes, three main systems of arrangement must be 
 considered. In the first, the space is filled with a succession of detached 
 devices, which may either be all alike, or of different kinds, mostly only 
 of two occurring alternately (Plate 198 (3 and 5) ). When these devices 
 are in the nature of leaves or sprigs, they are often arranged in contrary 
 diagonal directions so as to give the effect of a continuous organic line. 
 (Plates iiiA, and 196 (ii) ). In the second system there is a wavy stem 
 running along the border, from which leaves, blossoms, buds or other 
 devices depend (Plate 202 (i to 7) ). Such an arrangement is very common 
 in Persian, less so in Caucasian, and comparatively rare in Turkish 
 carpets. 
 
 The third system is less common than the first two. In it a 
 succession of devices extends along the centre line of the border-stripe, 
 each, as it were, growing out of the last one (Plate 193). These devices 
 may appear quite geometrical, but probably are mostly derived from 
 a floral origin. 
 
 Other patterns which might be included in the previous types are 
 noticeable rather on account of the nature of the devices than upon their 
 arrangement. Among these may be mentioned borders with inscriptions, 
 which range from the legible verses and texts of many Persian specimens, 
 to the debased Kufic letters found in old Turkish and later Caucasian 
 carpets (Plates 14, 43B, 63). Many border patterns also are based upon 
 the S-form, which takes very varied shapes (Plates 194 (22), 196 (5, 9), 
 198 (6) ). Other borders have a succession of panels, often of two shapes, 
 which themselves can be filled with ornament of various kinds (Plate 
 49). More rarely the space is covered with an all-over diaper pattern, 
 as in some Kulas (Plate 100). 
 
 A somewhat different border-stripe is made by a line passing along 
 the middle on such a course that the space on one side of it is not only 
 the counterpart but is also of the same form as that on the other. The 
 two sides are usually rendered in different colours and the pattern may 
 be called ** reciprocal ** or ** counterchanged ** (Plates 199 (17) ; 202 
 
 (20)). 
 
 The field of a carpet — that is, the space left when the border has 
 been subtracted — may be decorated in very many ways, the most obvious 
 
 127 
 
HAND-WOVEN CARPETS 
 
 of which is to fill it with an all-over repeating pattern which may be 
 held to include^ as an extreme case, a perfectly plain surface* After 
 a plain field, the simplest patterns include small spots or geometrical 
 diapers, which from a short distance can hardly be distinguished from 
 an undecorated ground* The spots may by degrees develop into floral 
 or geometrical figures of more or less elaboration* Among such figures 
 may be mentioned, as of very frequent occurrence, floral sprigs and 
 cone-devices* Great variety may be given to these patterns by the 
 arrangement of the colouring ; for example, the devices in the same rows, 
 whether vertical, diagonal or horizontal, may be coloured alike (to give 
 a striped effect), or many other systems may be adopted* 
 
 Other repeating patterns are afforded by the large group of lattices ; 
 either constructed of simple lines or of elaborate scrolls, stems or leaves* 
 The interspaces of such lattices are often occupied by detached devices 
 of various sorts* Some rather elaborate patterns are so frequently used 
 as to have acquired special names which are very useful for speedy 
 reference* 
 
 One of these, known as the ** Herati,*' consisting of an assemblage 
 of rosettes, blossoms and small, curved, serrated leaves, based on a vague 
 framework of stems, may be seen illustrated on Plates 147, 148 and 155* 
 This is found only in Persian rugs or, in a debased form, in a few others 
 based on the Persian* Another pattern of rosettes, blossoms and leaves, 
 more formally arranged, is called the ** Mina Khani ** and is illustrated 
 on Plates 149, 150 and 166* Still another is the ** Guli Hinnai,"' consisting 
 of small upright sprigs of flowers* The last two are equally Persian 
 in character with the Herati* A somewhat similar pattern, found very 
 often in Indian rugs, consists of a lattice with several different kinds 
 of floral sprigs in the various compartments* A repeating pattern 
 peculiar to early Turkish rugs is shown on Plate 43B* In this the arabesques 
 always seem to be in yellow on a red ground* 
 
 There are other patterns which, while resembling those already 
 mentioned, do not repeat regularly throughout the field* Among these 
 is a very famous one, known as the ** Shah Abbas,'' because it was so 
 common in Persian rugs of his time* It consists of an assemblage of 
 palmettes, blossoms, leaves and cloud-bands, arranged evenly, but 
 without formal repetition, throughout the field (see Plate 28)* Such 
 a pattern is capable of infinite modifications, and may introduce among 
 the other motives vases, animals, trees and many other devices* 
 
 Still more elaborate patterns of curving stems and flowers are very 
 
 128 
 
THE DESIGN OF CARPETS 
 
 common in the early Persian silk carpets, and in the modern rugs of 
 Tabriz and Kashan, though they rarely fill the whole field (see Plate 
 
 A less common scheme than many mentioned above is the division 
 of the field into narrow parallel bands, either vertical or diagonal, each 
 of which contains a pattern of the same character as those of carpet borders 
 (see Plate 164A). 
 
 Sometimes much more irregularly arranged patterns are found 
 distributed fairly evenly over the field ; but such usually occur only in 
 the more barbaric rugs. In Caucasian carpets, for instance, it is quite 
 common for the field to be filled with small geometrical or floral devices 
 without any mutual connexion, scattered capriciously all over (see Plate 
 
 117)- 
 
 When the ornament is not distributed regularly throughout the 
 field, there must be part of the latter to which extra prominence is given, 
 and the parts which naturally call for such treatment are, first of all, 
 the middle of the rug, and, next, the four corners. Accordingly, it is 
 often found that there is in the middle a panel or medallion, of more 
 or less compact form, but with a tendency to be longer in the direction 
 of the length of the rug. Medallions, as all such devices may be 
 conveniently called, of many shapes occur ; the commonest being 
 circular, hexagonal, ogival or lobed. Sometimes there are two or more 
 concentric medallions ; and very frequently the medallion has small 
 projecting pieces or panels at each end, when it is sometimes spoken 
 of as a pole-medallion (see Plates 9, 169). 
 
 The corners of the field are often filled with shaped pieces resembling 
 the quarter of a central medallion, and not infrequently these corner- 
 pieces repeat the pattern contained in the central medallion of the same 
 rug. 
 
 Corner-pieces may extend until they meet at the ends and even 
 at the sides of the field, so as to form either a kind of arch at each end, 
 or a shaped inner contour to the field. Such a shaped contour gives 
 an effect approaching that of a central medallion, but can usually be 
 distinguished on account of its being, as it were, obviously moulded 
 from without and not from within. 
 
 That part of the field not occupied by the central medallion and 
 corner-pieces may be left plain or filled with any of the repeating or 
 irregular ground patterns already described ; and the space within the 
 medallion and corner-pieces is also sometimes filled with such patterns, 
 
 129 
 
 K 
 
HAND-WOVEN CARPETS 
 
 but is more often occupied with less formal ornament having more relation 
 to the space it fills* 
 
 In a Kanara. or any carpet whose length is considerably more than 
 the widths instead of a single central medallion^ there often are two or 
 more in a line, and these are not always detached but may be connected 
 by a narrow neck (Plate 174). 
 
 The multiplication of medallions can be carried to a greater extent 
 still, and they may occur in other parts of the field than on the centre- 
 line* A very common pattern of the Turkish carpets of the sixteenth 
 and seventeenth centuries was one which included one or more large 
 medallions on the middle line and several incomplete medallions of another 
 design against the boundaries of the field (see Plate 44)* 
 
 The number of medallions may be increased until the resulting 
 design merges into the class of repeating patterns* A good example of 
 this is afforded by the rugs of Central Asia which so often have a pattern 
 of repeated octagons, with diamond-shaped figures between them (see 
 Plate 179)* 
 
 In ail the schemes of design already mentioned the field has been 
 treated as a whole, but it can also be divided into several separate panels, 
 each with a design of its own, though not necessarily different from the 
 others* The most common plan is to divide the length of the field into 
 three or more panels, as in Plates no, 136, but it can also be divided into 
 four panels by cross-lines, as in the Katchli '' rugs of Central Asia 
 (Plate 182), and of course many other systems of subdivision are possible, 
 though not often seen* 
 
 Allusion has already been made to the important type of design 
 with an unsymmetrical field found in prayer-rugs* These are carried 
 about by the devout Moslem and used by him when at prayer* At one 
 end of the field there is an arch, which forms a niche-shaped space, occupy- 
 ing most of the field* When in use, the arched end of the rug is pointed 
 in the direction of Mecca, the holy city, and the worshipper also faces that 
 way* The ground of the niche is often left plain or lightly filled with 
 ornament, but the spandrels above the arch are more richly decorated* 
 The rugs of Asia Minor usually have a wide border, and, consequently, 
 a field so long in proportion to its width that a niche occupying the whole 
 of it would appear unduly attenuated* In the prayer-rugs it is therefore 
 usual to reduce the length of the field by having one or more extra 
 horizontal border-stripes, or panels, at the ends inside the real border 
 (see Plate loi)* 
 
 130 
 
THE DESIGN OF CARPETS 
 
 The designs of the great majority of Oriental rugs will be found to 
 be arranged in accordance with the methods described above ; and 
 though many other schemes of design are possible none of them will 
 be met with at all frequently. 
 
 Most of the devices seen in carpet patterns require no explanation, 
 as they are either simple geometrical forms or else based upon ordinary 
 natural objects ; but a few of them, either from the obscurity of their 
 origin or on account of their known symbolical meaning, deserve some 
 mention. First among these may be placed the cloud-band, which was 
 derived from China, and was very largely used in the Persian carpets 
 of the sixteenth century and onwards (see Plates 29, 205), This motive 
 was repeated again and again, until it passed out of easy recognition, 
 but probably still leaves its trace on many traditional patterns. It is 
 possible that the widely used S-form is derived from it. The palmette 
 (see Plates 18, 204) is very obviously a floral form. It may be derived 
 from the palm, though its exact origin is uncertain. The name, however, 
 is a useful one, as the motive is of such very widespread occurrence. 
 
 Another motive of great importance has no very satisfactory name, 
 but is perhaps most commonly called the cone, or cone-device. The 
 most plausible hypothesis is that it is derived from the inflorescence of 
 the date-palm and symbolizes the principle of new birth. It is found 
 in Persian and Indian art from very early times, but does not appear 
 in carpets which can safely be ascribed to an earlier period than the 
 eighteenth century. In Kashmir, where it is nearly always included 
 in the far-famed shawls, it is called ‘‘ Buta,'" In the West it has been 
 variously known as the pear, river-loop, or crown jewel. This device 
 is drawn in many ways. In its simplest form it is merely a comma- 
 shaped blot, and from this it passes by many gradations through the 
 plain angular form seen in Shirvan rugs (Plate 195 (12)), and the simple 
 assemblage of florets of Sehnas and Sarabands (Plate 156), to the highly 
 elaborated floral masses of Bakus (Plate 132B) and Shirazis, Two 
 peculiar forms are the archaistic outlined one with a long bent neck, 
 seen in some Karabaghs (Plate 142), and the compound ones of Khorassans 
 (Plate 202 (13)), 
 
 Another device seen in Causcasian rugs would be quite impossible 
 to interpret if it could not be traced from the debased Kufic 
 inscriptions of some sixteenth-century Turkish rugs (Plates 136, 199 (16) ), 
 This in the most primitive form still remaining has been translated as 
 signifying '' Good Health,'' 
 
 131 
 
HAND-WOVEN CARPETS 
 
 The lamp which is represented in many prayer-rugs confirms the 
 theory that their design is derived from the gateway of, or arch in, a 
 mosque, where a lamp would be a common object. 
 
 Another device often seen on prayer-rugs is the comb, which is 
 a symbol of cleanliness — so important a virtue to the Moslem. Conven- 
 tional hand-prints are seen in the spandrels — where the hands would 
 naturally be placed — of the Beluchistan prayer-rug illustrated on Plate 
 192, 
 
 A symbol seen in some of the fine sixteenth-century Turkish carpets 
 woven with the Sehna knot, consisting of three balls and two wavy lines, 
 is discussed on page 51, 
 
 Much is often said about the symbolism in Oriental rugs, and it is 
 quite certain that many of the devices have indeed a symbolic meaning ; 
 but it should not be supposed that whenever these devices are used, 
 their symbolic aspect was at all prominent in the mind of the weaver. 
 The significance attached to them in the first instance became gradually 
 obscured with repetition until at last they are merely convenient ornamental 
 motives. It has already been explained that fresh motives are rarely 
 devised for rugs, but that existing ones are usually copied ; so that the 
 occurrence of symbolic devices in Oriental carpets has no more significance 
 than that of roses, lions or fleurs-de-lys usually has in Western art. 
 
 Though many of the devices found in carpets have a wide range 
 throughout the East, yet there are some that, either in their essence or in 
 their particular rendering, are restricted to certain of the great geographical 
 divisions. 
 
 In Plates 194 to 202 will be seen devices that are peculiar, or 
 almost peculiar, to either the Turkish, Caucasian, Central Asiatic or 
 Persian groups of carpets ; and the comparison of the details of a specimen 
 under examination with the devices contained in these plates will mostly 
 allow the first step to be taken towards its correct classification. 
 
 Plates 203 to 205 contain motives found in early Persian carpets. 
 On Plate 205, however, the ** Badge of Tamerlane ** in the upper right- 
 hand corner and the somewhat similar device in the bottom row, are 
 peculiar to the fine sixteenth-century Turkish carpets with the Sehna 
 knot. The third device with black roundels — at the middle of the 
 plate — is found in a Persian carpet and may be connected with the other 
 two, but possibly the resemblance is merely accidental. 
 
 132 
 
Chapter III 
 
 PRACTICAL CONSIDERATIONS 
 (a) Purchase 
 
 The properties of good and genuine carpets have been explained 
 in some detail in the foregoing chapters^ but it may still be useful to 
 summarize the considerations to be borne in mind when purchasing 
 them. Each of the many factors which, taken all together, make up a 
 carpet should receive due and methodical attention. These factors 
 may be stated to be : Design and colouring ; manner of weaving ; 
 
 quality of materials ; nature of dye stuffs ; condition ; genuineness. 
 Though all these factors are important, they affect to different degrees 
 a carpet according as it is new, obviously second-hand, or professedly 
 antique. For example, if a carpet is new the question of its genuineness 
 or condition hardly arises, but the nature of its dyestuffs is the most 
 important question ; if the carpet is more than fifty years old there need 
 be no anxiety about the dyes employed, but its condition must be very 
 carefully scrutinized ; and, lastly, if it is claimed to be antique, then its 
 genuineness is the most important consideration. 
 
 When a carpet is being examined it should first, in order to judge 
 of its general effect, be placed flat on the ground and looked at in turn 
 from all points of view. It is a favourite trick of dealers to throw a carpet 
 which they wish to sell over tables or chairs so that it lies in beautiful 
 folds, for not only does this show off the play of the colours to the best 
 advantage, but it also makes it very easy to overlook faults in the weaving 
 or other blemishes which are detected at once when the carpet is laid flat. 
 Another plan is to cause an assistant to hold up one end and move it 
 gently about ; and still another — perhaps more legitimate — is to place 
 it with a strong electric light falling on it at exactly the most advantageous 
 angle. It may be remarked here in passing that a carpet does not look 
 its best on a wall, and especially high up ; a fact much to be regretted 
 in connexion with museums, where it is generally impossible to exhibit 
 them in any other way. 
 
 The merits of a carpet in respect of design and colouring cannot 
 be usefully discussed here. The purchaser will naturally buy what 
 
 133 
 
HAND-WOVEN CARPETS 
 
 pleases him^ and whether his choice is a good one depends upon the 
 soundness of his artistic judgment* Such help on this point as this book 
 can give will be derived from a general perusal with perhaps special 
 attention to the chapter on design. 
 
 Another aspect of a carpet’s general effect calls^ however, for notice ; 
 that is its accuracy in respect of shape and flatness. Irregularities of 
 shape or a disposition to lie unevenly on the floor are flaws indeed in a 
 carpet, and to some extent have an influence on its value ; but the matter 
 must be considered a little further. Anyone who is buying a big new 
 Smyrna, Feraghan or Tabriz rug, will quite justly demand that it should 
 lie perfectly smoothly, because such products are made in competition 
 with European wares. On the other hand, it would not be fair to judge 
 the unpretentious work of nomadic tribes by the same standard. In 
 fact a perfectly rectangular and evenly lying nomad carpet is a very 
 exceptional rarity. Anyone who makes a great demand in this direction 
 merely betrays his ignorance about the origin and methods of production 
 of such pieces. 
 
 The details of the weaving as described in Part II, Chapter I, should 
 be examined as thoroughly as time will permit ; for it can hardly fail 
 to help towards forming a sound judgment on the rug if such points as 
 the material of warp, weft, and pile, the kind of knot used, the number 
 and nature of the shoots of weft, and the finish of sides and ends are all 
 carefully noted. One other point is of more importance than all these, 
 and should be carefully observed, because it determines to a great extent 
 the market value of the rug. That is the fineness of the knotting. It 
 is by no means necessary to ascertain this very accurately : a measure 
 can be held against the back of the rug parallel first to the warp and then 
 to the weft, and the number of knots to one inch counted for each direction. 
 These numbers multiplied together will give the number of knots to the 
 square inch. So much it is advisable to do, because the eye alone cannot, 
 without much practice, form a reliable idea of the fineness of the 
 knotting. 
 
 It may be mentioned here that it is possible to some extent to remove 
 folds, puckers, and general unevenness from a new carpet. The method 
 is to nail the piece, stretched as tightly as possible, face downwards to a 
 wooden floor, and then wet the back thoroughly with warm water. It 
 is then left until it is quite dry, when it may be found to be considerably 
 improved. In cases of extreme unevenness the treatment may be 
 ineffective, but it is very useful when a carpet, originally fairly flat, has 
 
 134 
 
PRACTICAL CONSIDERATIONS 
 
 been stretched unequally by being tightly folded in transit^ or from some 
 similar cause. 
 
 A disposition to curl up at the edges only is mostly due to an 
 imperfectly executed selvedge, and can often be remedied by fresh over- 
 casting ; but in some closely woven rugs curling of the edges will happen 
 in damp weather though not at other times. 
 
 In a closely woven carpet it is not always easy to count the knots, 
 specially those parallel to the weft, if they are in a patch of uniform colour, 
 but it is always possible to find part of the pattern where the knots are 
 separately visible. 
 
 With regard to the quality of the materials, the only thing that can 
 be usefully observed in a superficial examination is the feel and appearance 
 of the yarn constituting the pile. 
 
 And now comes the very difficult question of the soundness of the 
 dyes. The first thing necessary for anyone who wishes to be able to 
 form an opinion at all useful on this point is to study in soundly dyed 
 carpets the principal colours, especially red, blue and crimson, but as 
 many more as can be visually remembered. 
 
 If in a new or not very old carpet some of the colours appear to 
 resemble those closely in tint and quality, they are probably dyed with 
 the same materials and should be equally sound ; but if there are brilliant 
 hues of a different character they should be regarded as doubtful. Among 
 the most suspicious colours are a bright orange-yellow, a glaring vermilion- 
 red, and any very vivid purples or greens. 
 
 Each colour in the carpet should now be examined separately, in 
 order to see whether there are signs of fading. The back should be 
 compared with the front ; also, after folding the carpet sharply, the 
 pile near the knots, where it is protected from strong light, should be 
 compared with the surface. Even in a new carpet this treatment may 
 show that fading has already begun ; while in a rather older one it may 
 be found that parts of the pile, almost colourless, or of indefinite tint 
 on the surface, are of a brilliant hue nearer to the knots. Such signs 
 indicate that the colours concerned are not permanent, though there is 
 a possibility that the superficial change of tint may be due to chemical 
 washing. 
 
 The next test is to moisten a white cloth (the moisture of the mouth 
 being slightly alkaline is very convenient for the purpose) and rub it 
 vigorously on each colour in turn. Some will produce no effect ; some 
 will tint the cloth slightly ; while it may be that others will stain it deeply. 
 
 135 
 
HAND-WOVEN CARPETS 
 
 Carpets with such colours as the last should be generally avoided. In 
 the next place the whites should be inspected to see whether they have 
 become contaminated with a stain from the adjoining colours. If they 
 are^ it may be due to the use of soluble dyes, or perhaps because the 
 carpet, though soundly dyed, has been exposed to persistent damp. 
 When the dyes have been shown not to be very soluble such a 
 discoloration of the whites, if not unsightly, may be disregarded unless 
 the threads themselves have deteriorated. 
 
 It should be remembered that if the purchaser rejects every carpet 
 that has any suspicious dye in it he will have very few new carpets to 
 choose from. It is so very usual for a carpet in the main quite good to 
 have minute quantities of fugitive dyes — used probably on account of 
 the extreme brilliancy not easily obtained with sound ones — that the 
 purchaser may still wisely consider whether after all the carpet may not 
 be a satisfactory acquisition. If he can picture to himself the carpet 
 with the doubtful colours changed to a nondescript neutral tint, and 
 decides that such a change would entail no real deterioration in the 
 appearance of the carpet, then he need not fear to risk its purchase. After 
 all, bad fading of unimportant colours has occurred in many real old 
 carpets of the highest beauty, and if the effect of a modern carpet is due, 
 as it often is, to large masses of good blue and red, very little harm may 
 be done to the balance and tonality of the design by the weakening or 
 disappearance of the minor colours. If, however, the main colours 
 of a carpet are believed to be unsound, then it should be absolutely 
 rejected without hesitation. 
 
 To judge the dyes of an old carpet is easy. They cannot be synthetic 
 colours, and if they were originally fugitive they must already have faded 
 so much that no further considerable change is likely. In short, the 
 dyes can be judged on their present appearance. 
 
 The condition of a carpet determines in a high measure its value 
 and utility and mmst be studied from every point of view by the purchaser. 
 Luckily it is not very difficult to form an accurate opinion on the point 
 when once it is known what alterations, due to time, wear and accident, 
 are to be looked for. A new and unused carpet is naturally in good 
 condition as a rule, but there is a possibility that it has been kept in the 
 damp or exposed to sea-water in the course of transit, in which cases 
 the foundation threads may have deteriorated badly without there being 
 visible signs of the trouble, although it is more likely that there are stained 
 parts and signs of the colours having run together. The strength of 
 
 136 
 
PRACTICAL CONSIDERATIONS 
 
 the warp- and weft-threads should be tested by absolute force. The 
 carpet should be folded face inwards parallel to the warp and an attempt 
 should be made as it were to pull the weft-threads asunder. There is 
 no danger in applying this method to a sound carpet ; but if the weft- 
 threads actually break or loosen, or if a sound of cracking is heard, the 
 foundation may be known to be rotten in that place. The warp-threads 
 can be tested in a similar way, but the carpet must be very weak indeed 
 for them actually to break. If the carpet is weak — excepting perhaps 
 in a mere isolated spot — it should be rejected, for a new carpet with a 
 rotten foundation is valueless. 
 
 The more obvious signs of wear, such as occur in a second-hand 
 carpet, should now be taken into account. The ends and sides of a carpet, 
 on account of their exposed positions, are the parts which first begin to 
 wear away, and it should be noticed whether they are at all frayed, 
 loosened or weak. Then the pile should be examined. If the carpet 
 has seen service the pile must have worn down to some extent, at least 
 in places, for such wear is seldom uniform. If the examination shows 
 that the pile is worn away down to the knots, even in parts, it is almost 
 certain that the carpet will not stand much more wear, and it will probably 
 prove a bad investment. It may be obvious, but it is important to insist 
 upon the fact that a much-worn carpet, however good it may have been 
 and however beautiful it is in appearance, should not be bought by anyone 
 who intends it for use on the floor. 
 
 More obvious blemishes, which are often very unsightly, such as 
 holes, cuts, or localized stains, may be due to accidental causes and often 
 are of little account, provided suitable allowance is made in the price. 
 A skilful carpet-repairer can do wonders in such cases. Cuts can be 
 sewn together by a process akin to weaving ; holes can be re-knotted 
 on freshly-inserted warps ; and the knots of badly stained parts can 
 be picked out and replaced by new ones. Such work is of course often 
 done before the carpet is offered for sale, but by no means invariably. 
 
 It will aid the foregoing inspection if the carpet is held up with its 
 face to a strong light and then looked at from the back. This will at 
 once make visible small holes, slits, and thin, threadbare places, even 
 if they were overlooked before. Sometimes, too, repairs will be visible 
 by this means, but generally they are best detected by the sense of touch. 
 If the fingers are passed over those parts of the carpet where are seen 
 indefinite and vague dark lines, patches of slightly different colour, 
 or unexpected breaks in the pattern, an extra thickness felt may show 
 
 137 
 
HAND-WOVEN CARPETS 
 
 that the carpet has been joined together or repaired with stitching and 
 new knots. Such repairs are not very detrimental if skilfully done and 
 only of small extent. 
 
 Another defect of condition is that of extreme dirtiness of the surface. 
 A dealer will rarely show such a carpet^ for he knows what a wonderful 
 change may be effected by cleaning ; but it sometimes happens that a 
 piece is met with casually^ or perhaps at an auction sale, whose true 
 quality is entirely obscured by the dirty condition. If in such a case 
 any doubt is felt as to the improvement that cleaning will effect, a damp 
 cloth rubbed on a small part will soon settle the point. 
 
 It is obvious that the whole of the examination outlined above cannot 
 be carried out in some cases, such for instance as when carpets are put 
 up for sale by auction, and may not be convenient even Vv^hen they are 
 seen in a dealer's shop. It is on this account that inexperienced buyers 
 should be very cautious in making purchases at auctions, and should 
 if possible in other cases have the carpet for a few days on approval before 
 coming to a decision. Most dealers will agree readily to this, and indeed 
 it rather tells against them if they are unwilling to do so. Even 
 experienced buyers will be well advised to see a carpet at home before 
 buying, because it may not suit its intended surroundings as well as was 
 anticipated when it was seen in the shop, and also it is by no means 
 uncommon for a carpet which appeals very strongly at first to the eye 
 to seem more ordinary after a few days' familiarity. 
 
 By the genuineness of carpets is meant the correspondence between 
 what they really are and what they appear to be, or are claimed to be 
 by the man who wishes to sell them. Alas, a very great discrepancy 
 is often to be found here. Many dealers describe their goods as being 
 very old, especially if they suppose the purchaser sets store by age. In 
 the East it is a general practice to add about a century to the age of a 
 carpet, and this may be considered a practice of moderation in comparison 
 with what is often said. A dealer who describes a carpet of fifty years 
 old as a genuine antique " must be regarded as being not very untruthful. 
 
 A carpet may be justly called antique if it is over a century old, and 
 all such pieces, or even fragments, have a certain value unless the condition 
 is exceptionally bad. As the age increases the value increases enormously, 
 so that complete carpets of two or three centuries old in good condition 
 command immense sums. It is not to be wondered at that sellers try 
 to persuade customers that their carpets are of this class, and in some cases 
 they no doubt believe the wares to be as they describe them. When, 
 
 138 
 
PRACTICAL CONSIDERATIONS 
 
 to quote an actual case, a modern Multan rug of indifferent quality and 
 no artistic interest is offered as a seventh-century rug for the sum of 
 £50,000, it is only fair to suppose that the owner is under a genuine 
 misapprehension as to its real nature ! 
 
 Mere errors of description are not perhaps very dangerous, because 
 it is well known that the word of a vendor cannot always be relied upon ; 
 and there are not many buyers who will give large prices for goods of 
 whose value and nature they know little* A more insidious, if not quite 
 such a common evil, is for carpets to be treated in various ways with the 
 express intention of deceiving by making them appear to be different 
 from what they really are. It may be useful to describe shortly a few 
 of the operations of this sort which are commonly practised, and to give 
 a few hints by which the deception may be exposed. 
 
 It is well in the first place to ask oneself whenever a carpet is seen 
 for the first time : ** Is this piece genuine i ** Of course, in the majority 
 of cases an affirmative answer can be given immediately ; but the mere 
 asking of the question, and still more an attempt to find evidence of 
 deceit, will obviate the dangerous possibility of the purchaser being taken 
 off his guard. 
 
 The commonest trick is that of chemical washing, which has been 
 mentioned already. This is so widely done and is so well known that 
 it cannot invariably be regarded as a deliberate deception. It has the 
 effect of giving an appearance of age, but apart from that it is used to 
 modify the crude and vivid colouring which is possessed by many modern 
 carpets when new ; and it certainly does result in an immediate, 
 if transitory and undesirable, improvement in the rug's appearance. 
 It is practised to varying extents, from a sponging over with a mild acid, 
 not very injurious, to a severe treatment with strong chemicals which 
 alters the whole colouring and tonality of the design, and seriously impairs 
 the wearing qualities of the wool. It is carried out in the East, probably 
 often by the weavers of the carpets ; and to a still greater extent after 
 they have reached the Western markets. In London there are firms who 
 specialize in this business, and when it is said that the present cost of the 
 process is about gd. a square foot, it will be understood that the dealers 
 are pretty sure that it adds to the attractiveness of their wares. 
 
 It is not easy to be certain whether a rug has been chemically washed. 
 If, however, the whites are pure in colour and rather glaring it is certain 
 that it has not been washed. If, on the other hand, the rug is in good 
 condition and shows no signs of wear, and yet the whites are toned with 
 
 139 
 
HAND-WOVEN CARPETS 
 
 a yellowish-grey, and all the colours are mellow and a little indefinite, 
 then it is very probable that washing has been practised. 
 
 Another method of ageing carpets is more vicious if not more 
 harmful. This is to expose a new carpet to bright sunlight, in a street 
 where all the traffic will pass over it ; a service which the owner's 
 neighbours will cordially assist in. After the rug is cleaned, it will have 
 signs of wear which are not dissimilar to those obtained in genuine service. 
 Thus a rug may look many years old when it is in reality quite new. 
 There is perhaps no particular harm in this, though it is certainly a pity 
 to reduce the life of a rug by artificial means. The chief protection to 
 the buyer lies in the fact that this treatment is rarely accorded to really 
 good rugs, but only to those whose merits are insufficient without the 
 added glamour of age, to attract a purchaser. Anyone who is able to 
 evaluate the quality of the rug apart from the signs of wear, is not likely 
 to make a mistake here. 
 
 A fairly certain sign of genuine age is the unequal wearing of the 
 different colours. It has already been explained that black-dyed wools 
 generally wear away much faster than others, but nearly every dye has 
 some effect upon the quality of the fibre as well as upon its colour. Some 
 dyes actually seem to preserve the wool, but whether that is true or not 
 it will generally be found that the pile of some colours is shorter, less 
 bulky, or otherwise different from the rest. When it is seen that each 
 separate colour is evenly worn throughout and that there are differences 
 between the various colours, then it may be assumed with safety that the 
 carpet is at least several, and probably as much as forty or fifty, years old. 
 
 A process has recently been devised for improving the appearance 
 of the pile, by treating the carpet with glycerine and then passing it 
 between hot rollers. A lustre is certainly given by this means, but it 
 is different from the natural lustre of good wool, and it is almost certain 
 that the effect cannot be permanent, even if it does not seriously injure 
 the wool. 
 
 Other practices, perhaps more legitimate, have for their object the 
 improvement of pieces whose condition is not good. One common 
 artifice is to tint on the surface those parts of the pattern which are so 
 threadbare as to show nothing but the foundation threads. This is 
 permissible to a limited extent, in the case of old pieces kept for inspection 
 only, as for example in Museums or other collections, when the judicious 
 application of a little colour may prevent a few staring patches from 
 entirely spoiling the intended effect of the carpet. For practical service 
 
 140 
 
PRACTICAL CONSIDERATIONS 
 
 such a method is of course useless, as the colour would not last. In 
 such a case the correct remedy is re-knotting the injured parts ; and, 
 if this is not practicable, then the carpet is too far gone to be fit for use. 
 
 Old carpets which are badly worn, or much damaged in places, 
 while the rest is fairly good, are often cut up and joined together again 
 so as to make up a smaller piece of good appearance. Care is taken 
 that the joins do not run across plain parts of the rug ; but they are carried 
 along lines such as the edge of a border or, when necessary, are made to 
 follow more involved courses round the contour of the pattern. Carpets 
 which have been mended cleverly may perfectly well be beautiful and 
 valuable specimens in a collection, but if the purchaser does not detect 
 the repairs, then he may unjustly have to pay the price of a perfectly 
 preserved carpet, because only the best firms base their prices upon 
 the real condition, as ascertained by experts. 
 
 A still more refined art is the deliberate manufacture of bogus antique 
 carpets. A sketch of the process adopted by a certain factory will give 
 an idea of the misapplied skill devoted to this fraud. A carpet with a 
 good old design is made of the finest materials by the best available weavers. 
 It is then taken to the sea and immersed long enough to soften and bleach 
 the colours to the desired extent. Next it is rubbed down back and 
 front with a pumice-stone ; not evenly, but more in those parts where 
 wear would be expected. After this it is, while damp, singed with a 
 spirit-flame, so that the fluffiness and softness of the new wool shall be 
 removed. Lastly, holes are burned in it with red-hot coals, and these 
 holes are repaired so well as to be imperceptible without the closest 
 examination. 
 
 Such productions will not often be dangerous to the ordinary buyer, 
 because the cost of manufacture is so great that they can only be offered 
 for very high sums. Experienced dealers have, however, bought such 
 things under the impression that they were antiques of great rarity, which 
 shows it is not easy to detect the imposture. A microscope, however, 
 will show the burnt and slightly swollen ends of the woollen fibres. 
 
 Luckily, it seems almost impossible to imitate the pattern, colour, 
 and materials of a real old carpet with perfect accuracy ; so that there 
 is nearly always something incongruous in the imitation to the trained 
 eye. Very often the claim that a carpet is of a certain date, can be at 
 once disproved by the fact that the design is one quite unknown until 
 a much later period. 
 
 A close scrutiny of all the details which have been mentioned should 
 
 141 
 
HAND-WOVEN CARPETS 
 
 give considerable help in arriving at a correct appraisement of any carpet 
 whose purchase is contemplated ; but it must not be supposed that it 
 will take the place of the experience which is derived from a long 
 acquaintance and painstaking study of carpets. It is often hard^ even 
 when the facts are observed^ to give them their due weight — neither 
 more nor less — so that experts and dealers of long standing are sometimes 
 mistaken in their conclusions. It is certain, however, that such methods 
 of investigation as have been suggested will lead by degrees to the 
 acquisition of the experience so much to be desired. 
 
 (6) Prices 
 
 The price of an Eastern rug depends upon so many indefinite factors 
 that it is not easy to give much assistance on the point to a would-be 
 purchaser, and nothing that is written can take the place of a practical 
 acquaintance with the market. Although it is the hardest question 
 upon which to give information, it is perhaps that upon which the student 
 is most anxious to receive it, so that an attempt must be made to deal 
 with it as well as may be. 
 
 Although carpets are mostly sold by the piece, it somewhat simplifies 
 the question to consider the price per square foot, and, indeed, it is from 
 this point of view that the dealer fixes his prices. Leaving aside the 
 question of mere area, it is obvious that a piece which entails much work 
 must be sold for a higher price than one v/hich is more quickly made, 
 and accordingly a finely-knotted rug is more costly than a coarse one. 
 In comparison with the labour involved, the materials used do not account 
 for a high proportion of the price, but a very heavy rug is usually more 
 costly than the coarseness of its texture would imply. Other factors 
 which increase the price out of proportion to the labour involved are 
 beauty of design and colouring, rarity and age ; while, on the other hand, 
 unsoundness of workmanship and — to an even greater extent — bad 
 condition detract from the value. 
 
 In order to get a basis upon which to proceed it will be well to fix 
 upon a kind of standard of quality and condition, and this standard should 
 be one which is frequently met with. Let it be assumed then that the 
 standard rug is well-woven, made of good materials and dyed with sound 
 colours, and that it is either new or so little worn as not to have appreciably 
 lost any of its wearing qualities, and that in point of design and beauty 
 of colouring it is neither below nor above the average. When such rugs 
 are examined it will be found that, provided they are of the same fineness, 
 
 142 
 
PRACTICAL CONSIDERATIONS 
 
 it matters little from what country they come or of what type they are, 
 but that the price is pretty constant for them all ; and further it will 
 be found that if two rugs differ in texture then their prices are 
 approximately proportional to the number of knots in a square inch, 
 though there is a tendency, on account of the extra material required, 
 for the coarse rugs, as already stated, to be rather more expensive than 
 this proportion would indicate. Some types of rugs, even when 
 everything otherwise is equal, are rather more and some are rather less 
 expensive than the average, but generally the divergence is not 
 considerable. 
 
 The following table gives the present average retail price (July, 
 1922) per square foot in London of different types of rugs of the standard 
 quality as defined above. 
 
 
 Knots per 
 
 Price per 
 
 Price for each 
 
 Type. 
 
 square inch. 
 
 square foot. 
 
 50 knots per 
 square inch. 
 
 Smyrna or Turkey . 
 
 16 
 
 4/5 
 
 13/6 
 
 Smyrna or Turkey (“ Standard ”) 20 
 
 5 /- 
 
 12/6 
 
 Sparta 
 
 . 40 
 
 10/- 
 
 12/6 
 
 Chinese (modern) 
 
 • 45 
 
 10/- 
 
 II/I 
 
 Feraghan . 
 
 . 50 
 
 10/- 
 
 10/- 
 
 Shiraz 
 
 . 50 
 
 7/6 
 
 7/6 
 
 Muskabad 
 
 . 50 
 
 7/9 
 
 7/9 
 
 Beluchistan 
 
 • 55 
 
 7/9 
 
 7 /- 
 
 Meshed . 
 
 . 70 
 
 10/- 
 
 7/2 
 
 Caucasian 
 
 . 70 
 
 10/- 
 
 7/2 
 
 Kurdistan 
 
 . 80 
 
 14/5 
 
 9/- 
 
 Kirman . 
 
 . 80 
 
 12/3 
 
 7/8 
 
 Tabriz 
 
 . 80 
 
 12/3 
 
 7/8 
 
 Yomud 
 
 . 150 
 
 14/5 
 
 4/10 
 
 Bokhara . 
 
 . 200 
 
 17/9 
 
 4/5 
 
 Sehna 
 
 . 240 
 
 33/4 
 
 6/8 
 
 Kashan . 
 
 . 250 
 
 33/4 
 
 6/8 
 
 Kashan (silk) . 
 
 . 250 
 
 44/9 
 
 9/- 
 
 
 Warps to 
 
 
 
 
 I inch. 
 
 
 
 Persian Kilims 
 
 10 
 
 3/4 
 
 — 
 
 Other Kilims . 
 
 10 
 
 2/9 
 
 — 
 
 Soumak . 
 
 8 
 
 7/6 
 
 — 
 
 The last column. 
 
 which gives the 
 
 price per square 
 
 foot for each 
 
 50 knots per square inch, shows the relative value of the various types 
 of rugs when the fineness of knotting is taken into account. The first 
 
 143 
 
HAND-WOVEN CARPETS 
 
 four types are heavily-made rugs in which the sheer weight of material 
 increases the proportionate cost. The relative value of the rest^ it will 
 be seen^ does not vary very greatly. This column shows that the most 
 expensive rugs in proportion to their fineness are Feraghans, Kurdistans^ 
 and Kashans with a silk pile ; while the cheapest are Central Asiatic rugs^ 
 such as Bokharas and Yomuds. Other rugs not mentioned in the table 
 range about the average value^ which may be said to be about 75. 6d. 
 per square foot for each 50 knots to the square inch. 
 
 Exceptionally pleasing design in colouring, as already said, will 
 lead to an increased price, though it is quite impossible to suggest any 
 system upon which the increase can be evaluated. It may be said, 
 however, that for standard rugs the prices should rarely be twice as great 
 as those given. On the other hand, an inferior appearance may cause 
 a reduction in price, but it will not on that account fall much below the 
 quoted figures. Bad condition may reduce the price indefinitely, and 
 perhaps a rough guide to the decrease may be afforded by an estimate 
 of how much of the rug's useful life has been destroyed by v/ear or misuse. 
 
 Generally speaking, the value of a rug increases with its age, as soon 
 as the latter becomes considerable. Here again accurate figures cannot 
 be given, but quite roughly it may be suggested that, if the condition 
 is reasonably good for the age, then each century that has passed about 
 doubles the value. 
 
 In this connexion it may be of some interest, even if of little practical 
 use, to work out on these lines the approximate value of the Ardabil carpet, 
 which is probably the best known and most famous rug in the world. 
 It has about 350 knots to the square inch, and contains 604 square ft. 
 Viewed as a standard new rug, its value would be about 604 X 7 X 7/6, 
 or £1,585. As it is nearly four centuries old, this value must be multiplied 
 by 16 ; and as its condition is much above the average for such an age, 
 and its interest is quite exceptional, the resulting price may fairly be 
 at least doubled again. This would lead to the sum of £50,720, a figure 
 not inconsistent with various estimates which have been made. 
 
 Again, to take an antique carpet of fairly common type, it will be 
 found that an early eighteenth-century Ghiordes prayer-rug in fair 
 condition, and containing about 30 square ft. and having 70 knots to 
 the square inch, will cost about £60, which agrees with the formula 
 7/6X30X4X 70/50. A Kula rug of the same kind will usually be 
 rather more coarsely woven and will probably be sold for from £45 to 
 
 £50* 
 
 144 
 
PRACTICAL CONSIDERATIONS 
 
 It will not be supposed from what has been said above that the 
 general level of carpet prices is free from market fluctuations — and indeed 
 such is far from being the case — but in ordinary circumstances the changes 
 are not very great or very rapid. All the time^ however, there is a certain 
 movement going on, sometimes in one direction and sometimes in the 
 other, though the general tendency is for prices to get higher year by 
 year. Up till 1914 they had been slowly but steadily mounting, and then 
 the war, which had such a profound effect upon all prices and values, 
 caused a very rapid increase, and a level was reached such as had never 
 been thought of before. In fact there are few commodities not necessary 
 to life or to the supply of munitions of war which increased in price as 
 much as carpets. The apex was reached about 1918 and 1919, but it 
 is not easy to give a definite figure for the increase, as the various types 
 of carpets were very differently affected. The really old and rare carpets 
 whose prices were already very high suffered the least change, while 
 some more ordinary kinds in special demand at the moment went up 
 to at least five times their pre-war value. Perhaps it would be near the 
 mark to say that the general level, taking one type with another, exceeded 
 by 200 per cent, that existing before the war. Afterwards prices fell 
 again even more rapidly than they had risen and very many carpets had 
 to be sold by dealers at prices much lower than had been given for them. 
 At present (July, 1922) the general level may be taken to be about 20 per 
 cent, higher than it was in 1914, so that it is probable, taking the natural 
 rise into account, that prices are not very different now from what they 
 would have been if there had been no war. 
 
 As to the future, it is dangerous to speak, but it hardly seems possible 
 that the present level will not soon be very much heightened. The 
 tempting prices of the war caused the accumulated stores in private 
 hands to be drawn away to the West ; and the general disturbance in the 
 East, to say nothing of the actual destruction of the weavers and their 
 property, must, it would seem, inevitably reduce fresh production. It 
 also seems likely that labour conditions will change so much, as profoundly 
 to alter the cost of manufacture, so that it is quite possible that in a few 
 years the prices of hand-made Eastern carpets may be such as to make 
 the present ones seem insignificant. 
 
 (c) Treatment 
 
 A few hints may perhaps be usefully given as to how carpets ought 
 to be treated. 
 
 145 
 
 L 
 
HAND-WOVEN CARPETS 
 
 The first problem to be dealt with is that of the best way of cleaning 
 them when dirty. A carpet^ even when bought straight from a dealer, 
 or in any case after it has been in use for a number of years, will generally 
 be full of dust, and dirty on the surface, and it is essential to have it cleaned 
 occasionally if its beauty and strength are to be retained. 
 
 If the carpet is fairly strong in structure it should have the dust 
 removed by beating with a broad carpet-beater, while it rests face 
 downwards on some level and slightly yielding surface ; the use of a 
 stick, whether it is supple or not, can in no circumstances be recommended. 
 A suitable surface for resting the carpet on while it is beaten is a dry 
 lawn, or in winter time on clean snow. It may also be placed on an 
 ordinary floor covered with several layers of newspapers, but it is not 
 advisable to hang a carpet on a line for beating. The use of vacuum 
 cleaners, or of compressed air, is very suitable for new carpets or older 
 ones whose condition is still good ; but if these methods are employed 
 on old carpets which are weak or have been much mended, the powerful 
 suction or pressure of the air causes the loss of a great deal of wool, 
 especially in the repaired places. 
 
 After this process of dry-cleaning, carpets ought to be washed, when, 
 of course, it is essential that the colours are such as will not be spoiled 
 by wetting. Washing, however, must not be left to the mercy of an 
 ordinary laundry, because they would use very hot water and possibly 
 alkalis, the result of which is that the carpet will return with dead and 
 lustreless wool. The best way is to wash it with warm water and carpet 
 or wool soap, which is a compound of ordinary good soap and ox-gall. 
 The lather can be applied freely with a soft brush rubbed the way the 
 pile lies. The soap must be well washed away by repeated soakings, 
 or treatment with a hose-pipe, after which as much water as possible 
 should be pressed out by stroking with a smooth stick in the direction 
 of the pile. 
 
 This process gives softness and even an enhanced lustre to a carpet 
 when once it is thoroughly dry. 
 
 Carpets are best preserved by washing every few years, and only 
 beating occasionally and moderately in the meantime. They must, 
 of course, be swept or brushed frequently. This should be done chiefly 
 in the direction in which the pile lies. Brushing from side to side may 
 be done gently, but never should it be against the pile. In respect of 
 this, domestic servants are very obstinate, for they often allege that more 
 dust is removed by brushing in the wrong direction. They are quite 
 
 146 
 
PRACTICAL CONSIDERATIONS 
 
 right in the statement, but the reason is that the pile is raised and separated, 
 so that far more dust can accumulate in it» When the brush is used 
 in the right direction its bristles gradually coax the dust to the surface, 
 and the pile is left in the condition best fitted for resisting wear. Mechanical 
 carpet-sweepers are inadvisable as they sweep in both directions. 
 
 If the owner of a carpet has a small river handy he cannot do better 
 than immerse it in the flowing water after he has secured it tightly. He 
 may also spread it back uppermost on clean grass or on a network of 
 ropes when heavy rain is falling, or may drag it about with its face on a 
 patch of clean snow. 
 
 It will be seen that some of these processes are not easy to apply 
 to large rugs, and it may be necessary to send them for cleaning to a 
 reliable firm who understand the work. One great difficulty is the final 
 drying of the carpet, which must be done out of doors or in a very large 
 room where the air is circulating freely. As the drying often takes several 
 days, it may happen, especially in big towns, that a lot of fresh dirt may 
 fall on the carpet while it is exposed. 
 
 Other methods of cleaning carpets are sometimes recommended 
 but are of doubtful value. Treatment with wood-shavings soaked in 
 petrol is expensive, dangerous and not effective. Rubbing them while 
 damp with rough sea-salt is liable to injure the colours. 
 
 The worst enemy of the carpet collector is the common moth. 
 Carpets which are walked on are much safer from its attacks than those 
 which are used as door or wall-hangings or turned into upholstery, but 
 the greatest danger is to rugs which are stored away. As moths have 
 a liking for quiet, warm places it is advisable to disturb them by airing 
 floor carpets and shaking hanging specimens as often as possible. Moths 
 have a preference for certain colours, especially red, and for loose textures 
 rather than for tightly-knotted ones. For this reason it is particularly 
 desirable that pieces which have once been affected with moth should be 
 examined with especial care, 
 
 A very useful way of keeping moth away from carpets is to place 
 the wings of birds, such as hens, pigeons and crows, on the top of cupboards. 
 Moths have quite a peculiar passion for feathers and will lay their eggs 
 in them in preference to any other place. By leaving such temptation 
 about and making examination from time to time it is comparatively 
 easy to destroy these enemies. 
 
 If carpets must be put away for a time, it is well to pack them with 
 naphthalene, camphor, or lavender oil, in tin-lined cases, or strong paper 
 
 147 
 
HAND-WOVEN CARPETS 
 
 parcels into which the flying moth cannot penetrate* If^ however^ the 
 eggs of the moth are packed up in the carpet they may hatch out, in spite 
 of any strong-smelling substance, and do much damage* It is, accordingly, 
 wise to make as sure as possible that the carpet is free from infection 
 when put away, and as there is never complete certainty as to this, there 
 is danger unless the carpets are examined from time to time* 
 
 A safe plan is to put the goods into cold storage, but this is expensive, 
 and in many places not practicable* It is claimed that petrol poured 
 over will destroy the eggs of the moth ; but it is not quite certain whether 
 it is a fact* An ingenious method of destroying infection, and one which 
 has been tried with success, is to place the carpets in a warm chamber, 
 saturated with the vapour of chloroform* The warmth hatches out 
 the eggs, but the grubs are immediately destroyed before they can do 
 any damage* 
 
 It must be remembered that good carpets must be taken care of* 
 For instance, heavy pieces of furniture with sharp edges should never 
 be allowed to stand continuously on the same place of a carpet, or marks 
 and holes will result according to the strength of the fabric* If a piece 
 of linoleum, or, better still, a sheet of thick rubber is placed under the 
 legs of tables and chests there will be little danger of damage 
 
 If it is necessary to put away, or pack a carpet, it is better not to fold 
 it in both directions, especially if it is tightly-knotted or weak : folds 
 in one direction only, do little damage unless the carpet is very stiff or 
 very old* The best plan is to roll the carpet* Even this may injure a 
 very stiff carpet, especially if it is subjected to pressure when rolled ; 
 but the difflculty may be surmounted by rolling it round some cylindrical 
 object like a strong tube of cardboard or another softer carpet* A small 
 thin carpet may be protected by a stick in the middle, which will prevent 
 the whole roll being bent* 
 
 It should be noted at times whether any attention is needed in the 
 way of repairs* If the fringe gets ragged it should be trimmed evenly* 
 If the ends or sides begin to unravel, they can be oversewn as a blanket 
 is, or re-overcast, and perhaps the work will be stronger if a piece of 
 webbing is sewn on at the back along the edge* Cuts and tears should 
 be repaired as soon as possible, and holes or small threadbare places 
 should be re-knotted* Old rugs that will not stand much wear but are 
 still beautiful objects can be hung up instead of being left on the floor ; 
 but it is often possible to find a place in the house where they can be 
 placed flat and yet only occasionally be walked on* 
 
 148 
 
Chapter IV 
 
 CARPETS IN THE MARKET— THEIR GROUPING AND 
 
 IDENTIFICATION 
 
 (a) Turkish 
 
 Turkish carpets may be divided into two groups : (i) those produced 
 in factories and (ii) those produced as a home industry, 
 
 (i) Turkish Carpets made in Factories , — The sedentary Turk is not 
 a craftsman ; so that practically all carpet-making in the Turkish dominions 
 is the work of other races^ such as the Greeks and Armenians. The 
 Greek element is strongly predominant along the coast of Asia Minor, 
 and for this reason it has been possible to evolve a factory-system, which 
 is now so highly developed and so ready to cater for the needs of the 
 West that carpets of any design and colour, and of almost any size, may 
 be ordered and finished within six or eight weeks. The factories are 
 well-equipped and not lacking in capital, which accounts for one of the 
 chief differences between their products and those made in the homes. 
 The latter is carried on by the poorer people in their small living-rooms 
 and, consequently, it is not practicable to have a loom measuring more 
 than about 6 ft. wide. The rich factories on the other hand have large 
 premises which permit of the erection of looms up to a width of perhaps 
 45 ft. For this reason Smyrna, and its back country, Kula, Ghiordes 
 and Demirdji, are able to produce large carpets 6 X 9, 8 X ii, 9 X 12, 
 II X 13, 12 X 15 ft. in size and even much larger ones, which are known 
 under the collective name of Smyrna carpets in Europe and America. 
 They are exported in large numbers and are found all over the world, 
 chiefly in use in public rooms, business offices, and dining-rooms. 
 
 These carpets may be regarded as the descendants of the large Ushak 
 carpets which were made in Asia Minor from at least the sixteenth 
 century. Though they are inferior in design and coarser in texture, 
 they yet retain much the same colouring and many of the same technical 
 peculiarities. Such carpets were brought into Europe freely in the 
 eighteenth century, and ever since have been very largely used. So 
 well-known has this type been, that the name ** Turkey Carpet ” was 
 at one time applied to all Oriental rugs, just as it is now common for 
 
 149 
 
HAND-WOVEN CARPETS 
 
 nearly all to be included by the term '' Persian/' The best known types 
 of Smyrna carpets are very thick and durable^ but among the coarsest- 
 woven of all rugs^ there being only from 13 to 20 knots to the square 
 inch. They are made entirely of wool, and while the warp is usually 
 white, the weft is almost invariably dyed red. Plates 93 to 97 show 
 five common types of Smyrna carpets. 
 
 Plate 93. Modern Smyrna carpet — also known as Ushak or 
 Medallion Carpet. Carpets of this kind have from 10 to 16 knots to 
 the square inch. They range in si2;e from about 6 X 8 ft. to 20 X 40 ft., 
 and they are made almost exclusively of three colours, red, green and 
 blue. They are among the cheapest of Eastern carpets. 
 
 Plate 94. Modern Smyrna carpet — also known as Ghiordes carpet. 
 Carpets of this kind have from 13 to 18 knots to the square inch. They 
 range in size from 6 x 9 ft. to 20 X 40 ft., and can be obtained in 
 whatever colours are in favour in Europe at the time. They are also 
 made with more elaborate designs as in the next Plate. They are the 
 cheapest of the Smyrna carpets. 
 
 Plate 95. Modern Smyrna carpet — also called Demirdji carpet. 
 Carpets of this kind have from 16 to 22 knots to the square inch. They 
 range in size from 6 X 9 ft. to 20 x 40 ft., and can be obtained in whatever 
 colours are in favour in Europe at the time. They are also made with 
 a central medallion on a plain ground as in the last Plate. They tend 
 to be rather more expensive than the last type. 
 
 Plate 96. Modern Smyrna carpet — also called after various towns 
 of Asia Minor. Carpets of this kind have from 19 to 58 knots to the 
 square inch. They range in size from 6 x 9 ft. to 20 x 40 ft., and 
 may be purchased in any colours, and in a great variety of modern and 
 old Persian designs. When of the same texture, they are about the same 
 price per square foot as the last type. 
 
 Plate 97. Modern Hereke carpet — also called Smyrna carpet. 
 Carpets of the kind have from 45 to 64 knots to the square inch, and 
 so are the finest-knotted of the Smyrna carpets. They are made in 
 the factory of the Sultan at Hereke (close to the railway from Scutari 
 to Angora). They are of various designs and in all sizes. The Plate 
 shows one of the best types, but cheaper carpets are made with designs 
 as in Plates 51 and 52. The price is higher than the other Smyrna carpets 
 in proportion to the fineness of the texture. 
 
 A kind of carpet made of recent years in Smyrna is sold under the 
 name of Sparta. These are mostly of small or medium size and are 
 
 150 
 
CARPETS IN THE MARKET : GROUPING— IDENTIFICATION 
 
 knotted rather more closely than the ordinary Smyrna carpet. Their 
 designs are very varied but are usually based upon Anatolian or Caucasian 
 models and they may easily be mistaken for products of the latter region. 
 
 (ii) Turkish Carpets of the home Industry . — Carpets are woven in 
 the homes about Smyrna and also in the less accessible parts inland, 
 but especially in Anatolia where the full assistance is obtained of the 
 Armenian element in the population. 
 
 The fact that Anatolia borders in the West on the region of the 
 factories and that in the East it approaches the ^one of Persian influence, 
 has had considerable effect on the kinds of carpet made. Those of western 
 Anatolia are reminiscent in quality, colour and pattern of the products 
 of the coast, whereas those from the Eastern side compete in quality and 
 beauty with the good Persian and Caucasian specimens. 
 
 Generally speaking, Turkish carpets are made entirely of wool, 
 though in some of the finer ones, especially those from Ghiordes, cotton 
 is sometimes found in the warp and weft. Other materials are so rare 
 as to be quite exceptional. The warp is usually of white or at least undyed 
 wool : in the more barbaric types it tends to be coarse and browner 
 than in the others. The weft is mostly of red wool, though white, yellow 
 and even blue are not uncommon colours. The usual plan is to have 
 two shoots of weft after each row of knots, and in the coarser rugs when 
 more shoots are found, the larger number is not constant throughout 
 the carpet, but happens at occasional intervals. 
 
 A peculiarity often seen in the weft, but perhaps never in rugs of 
 any other country, is that it does not invariably go right across from 
 side to side, but goes backwards and forwards over sections of the carpet ; 
 each section being roughly triangular in shape. The diagonal lines, 
 marking the juncture of the sections, are clearly seen at the back and are 
 sufficient to identify the specimen with Anatolia. The convenience 
 of this plan when several weavers are at work on a large rug is obvious, 
 for each weaver can go on for a time independently of the others, and 
 need not wait till all are ready to allow a shoot of weft right across the 
 loom. The habit is still often retained in smaller rugs, and sometimes 
 it is carried to the extent of using weft of different colours. It will be 
 seen that the plain field of some prayer-rugs has a weft of the same colour 
 as the pile, and this no doubt helps to give a superior effect to the colour 
 of the ground. In some cases it may have induced the belief that the 
 border was added recently to a rug of older date. 
 
 The pile consists of lustrous wool of good quality and is mostly of 
 
 151 
 
HAND-WOVEN CARPETS 
 
 medium length. In the Yuruk rugs it is^ however^ rather long, and 
 in the Ghiordes rugs it is notably short — in fact in these the collars of the 
 knots are sometimes almost visible. The Turkish or Ghiordes knot 
 is invariably used. The selvedge usually consists of a narrow web of 
 several cords overcast with wool, and the ends are nearly always fringed. 
 
 The colours are bright and varied. No attempt is made to get 
 a rich effect by means of deep colours of quiet tone, but prismatic hues 
 are placed together in juxtaposition, rather in the manner of a mosaic. 
 In the newer rugs this somewhat daring method often gives unpleasantly 
 crude effects, but the older ones challenge comparison with those of 
 Persia and other parts, although the colour-scheme is so different. 
 
 In point of design the first thing to notice is the great prevalence 
 of prayer-rugs, which certainly outnumber all other kinds. The niche 
 in such rugs more definitely suggests an archway than those of Persia, 
 Often there are two columns supporting the arch, but even when these 
 are not present, or when they are represented by a mere band running 
 down the side of the field, there is often an indication at the springing 
 of the arch of a flattened surface ready for receiving them. The arch is 
 angular rather than curved in shape, though the straight lines are often 
 softened by having a stepped contour. The exact form varies considerably 
 in different districts, though nearly every district has a type almost peculiar 
 to itself. Cross-panels (in reality short horizontal border-stripes) are 
 mostly found, either above or below the niche or sometimes in both 
 positions. The niche itself is plain more often than not, but it may be 
 ornamented with floral forms projecting from the sides, or sparsely scattered, 
 or hanging from the top. Ornament in the last position evidently takes 
 the place of the lamp, which itself is sometimes present there, or in the 
 spandrels above. The field is never covered with an all-over pattern 
 (apart from a very small spot or diaper) but always, even when most 
 ornate, gives the impression of a plain field with a little superimposed 
 decoration. Prayer- rugs with several niches side by side are found 
 in Anatolian work, and are often falsely supposed to come from Eastern 
 Turkestan, where that form is much more common. The other rugs 
 of Anatolia are of various sizes, but never very large. Runners are very 
 uncommon. 
 
 The details of the design are mostly derived from floral forms highly 
 conventionalized ; and even when geometrical devices are seen it will 
 often appear on closer study that they really have a floral origin. There 
 is, however, a certain amount of purely geometrical design, and the rugs 
 
 152 
 
CARPETS IN THE MARKET : GROUPING— IDENTIFICATION 
 
 in which it occurs have a great resemblance to those of Caucasia and 
 are often mistaken for them. Certain floral devices are almost peculiar 
 to Anatolian patterns — the carnation on short sterns^ which in various 
 stages of conventionalization often projects into the central panel ; little 
 sprigs of three blossoms or three leaves ; sprays of hyacinths ; and tall 
 tulips on upright stems, which last are often described as Rhodian lilies. 
 Human figures and animals are very rarely found. 
 
 The border-stripes, unlike those in Persian and Caucasian rugs, 
 rarely have continuous wavy stems, but are usually filled with detached 
 sprays, leaves, blossoms, or S-forms, though these are often so closely 
 arranged as to give the effect of a continuous pattern. Some of the 
 borders still in frequent use are derived with very little alteration from 
 forms found in carpets of the sixteenth century and onwards. Many 
 of the more elaborate border patterns are almost peculiar to certain 
 types of rugs and, accordingly, assist greatly in the problem of identi- 
 fication. Plate 194 shows devices found almost exclusively in Turkish 
 carpets. The majority of these are freely used throughout the group, 
 but some are almost peculiar to certain types. For examples Figs, i 
 and 14 in the plate are characteristic of Ladiks ; Fig. 3 of Mujurs ; and 
 Figs. 5 and ii of Ghiordes rugs. Fig. 21 shows the lamp so often found 
 in Turkish prayer-rugs, and it is a simplification of this device that may 
 result in the small figure of the diaper referred to on p. 157. 
 
 It must be stated here that the quite modern Anatolian rugs are 
 in every way inferior to the older ones. The patterns have perhaps 
 altered least, but the texture is coarser and looser, and the colours are 
 very crude and often unsound. In fact, apart from the Kilims, the 
 modern Anatolian rugs are, as a whole, the worst of any produced in 
 the East. The importance of the group lies with those of moderate or 
 greater age, and nearly all that follows should be regarded as applying 
 to those only. 
 
 The general characteristics of Anatolian rugs as described above 
 are naturally somewhat varied in certain districts, but these variations 
 when of consequence will be mentioned in the short description which 
 follows of the recognized types of rugs. 
 
 As it is scarcely possible to give any system of classification of these 
 types — apart from the broad distinction between the rough nomadic 
 and the more refined rugs — it will suffice to take them without any 
 methodical order. 
 
 Ghiordes Rugs, which get their name from the town on the site 
 
 153 
 
HAND-WOVEN CARPETS 
 
 of the ancient Gordium^ are perhaps the best known and most pri2;ed 
 of the Anatolian group. They are fairly finely woven and the pile is 
 so short that the wool never looks lustrous. The warp, and weft, and, 
 less often, the white part of the pile are sometimes of cotton. In some 
 pieces there is an added silk fringe at the corners. The colours are 
 chiefly red, blue and white, which apart from the niche are never in large 
 masses. In the later specimens the red inclines towards crimson. The 
 niche usually has a plain blue or red field, and a high pointed arch, often 
 supported by columns. There are mostly two cross-panels. The borders 
 and spandrels are filled with fine floral detail. The other rugs usually 
 have a hexagonal central panel fringed with projecting floral sprays, but 
 are otherwise like the prayer-rugs though mostly larger. A special 
 kind of prayer-carpet — known as the Ghiordes-Marpudjle — has narrow 
 stripes running across the whole carpet. 
 
 Kula Rugs are rather similar to the last, but are usually coarser 
 in texture. The colours are chiefly red, yellow, and blue. In most 
 of the older rugs the red (why, is not known) has faded to a brownish 
 yellow ; so that such rugs have a characteristic blue and yellow appearance. 
 The arch of the prayer-rugs is pointed but flat, and there is only one cross- 
 panel as a rule. The spandrels, and even the field sometimes, are filled 
 with a diaper of repeated floral forms. The borders are peculiar, as 
 they often consist of a large number of narrow stripes, each filled 
 with small detached blossoms or buds. A type, known as ** Cemetery 
 Carpets,'" has one or more pieces of land with a tomb and cypress trees. 
 The other carpets are mostly like double-ended prayer-rugs. 
 
 Ladik Rugs (named from Laodicea) often have the warp and weft 
 dyed the predominant colour of the pile. The ends have a fairly wide 
 red web, often striped with other colours, ending in fringes. The colours 
 are chiefly red and blue, but with fairly large quantities of other hues, 
 and they are used in larger masses than in Ghiordes rugs. They are 
 mostly bright, but sometimes a rich but quieter colour-effect is found. 
 The arch of prayer-rugs has three points, and there is a wide cross-panel 
 above or below the niche, containing tall, upright tulips. A typical 
 border-stripe is one with large rosettes alternating with sprays of three 
 blossoms. The other rugs, which are not common, have various patterns, 
 in which long hexagonal panels and stars are favourite devices. 
 
 Mujur Rugs often have a red or brown weft. They are rather 
 coarsely knotted and have a fairly long pile. The ends have coloured 
 webs and fringes. The colours are bright and numerous — red, green, 
 
 154 
 
CARPETS IN THE MARKET : GROUPING— IDENTIFICATION 
 
 blue^ yellow^ and violet mostly being plentiful. The prayer-rugs have 
 a steep, pointed, stepped arch, which opens out into a small lobe at the 
 point. There is one cross-panel above the niche. The main border- 
 stripe has a characteristic row of rosettes made into a square form by 
 the addition of corner ornaments, and one of the smaller stripes usually 
 has a succession of touching lozenges. 
 
 Kir-Shehr Rugs resemble those of Mujur in weave and colour, 
 though green is more plentiful. The arch of prayer-rugs is also similar 
 to the last, but mostly has a larger number of narrow bands to form its 
 contour. There is one cross-panel, either above or below the niche. 
 A characteristic of the border is the occurrence of several narrow stripes 
 like those of a Kula, but alternate ones are usually bright red in colour. 
 
 Nigde and Tuzla Rugs resemble the last two, but are not important 
 types. 
 
 Bergama Rugs (from the ancient Pergamum) have always a dyed 
 warp and weft and a long silky pile. The ends have a wide web crossed 
 by coloured stripes, and a tasselled fringe. These ends often have small 
 rosettes on them, or shells and other small objects attached to them for 
 the purpose of averting the evil eye. The colours are rich and dark 
 and consist chiefly of red and blue. The rugs are nearly square in shape, 
 and prayer-rugs are exceptional. When they do occur they have arches 
 of various forms, including the peculiar one shown in Plate io8a. The 
 designs are very varied and hard to describe, but they are mostly bold 
 in proportion to the size of the rug and floral forms are highly conven- 
 tionalized. 
 
 Anatolian is the name given in the trade to a nondescript lot of 
 rugs made throughout Anatolia. They are mostly small in size, and are 
 coarsely-woven with long, shaggy, soft wool. The colouring is rather 
 crude and the designs are very varied, but often copy features from the 
 better defined classes. Any small rug that does not conform to one 
 of the well-known types can with justice and convenience be called 
 Anatolian. 
 
 Melas Rugs (named from Melassa) form an ill-defined group char- 
 acterized chiefly by the form of the prayer arch, which is narrow and 
 furnished with a diamond-shaped lobe at the point (Plate 113). 
 
 Konia Rugs (named from Iconium) are similar to the last group, 
 but have a pointed arch with a small triangular projection at its springing 
 (Plate 1 13). 
 
 Makri Rugs are sometimes called Rhodian, but are not made in 
 
 155 
 
HAND-WOVEN CARPETS 
 
 that island. They are divided into a number of arched niches and are 
 often compared to a cathedral window. 
 
 Yuruk Rugs^ unlike the last few classes, form a well-defined group. 
 They are made by the Kurdish mountaineers of the West and have much 
 in common with the Nomadic rugs of the Caucasus. The wool is long 
 and shaggy and the colours are dark and characteristic, consisting of 
 brown, red and blue, with a little deep-toned white. The pattern consists 
 mostly of concentric diamonds, heavily fringed with latch hooks. They 
 are often mistaken for Caucasian products ; but should be distinguished 
 by their colour-scheme. 
 
 The Plates 98 to 114 show specimens of all the principal types 
 described above, and should be sufficient for identifying any fairly 
 characteristic specimen. Pieces which do not follow these types may 
 be difficult or impossible to associate with any definite district, but those 
 who have had many through their hands can often, by paying attention 
 to their subtler features, group them with certain well-known types even 
 if they differ from them in many particulars. 
 
 Plate 98. Ghiordes Prayer-carpet. 
 
 Carpets of this kind have from 65 to no knots to the square inch. 
 The si^e is generally about ft, x ft. They are somewhat rough 
 to the touch, and have a short and lustreless pile. The chief colours 
 are red, blue and white. The ground of the niche is usually blue or 
 red, the former being perhaps the more common ; a white ground is 
 very rare. The main border-stripe of this specimen has a pattern which 
 has come down almost unchanged from the sixteenth century ; in it 
 may be seen sprays of the hyacinth, which was so much used in early 
 Turkish art. Ghiordes prayer-rugs more often have no columns and 
 a steeper arch than this one, and one of the two cross-panels is generally 
 below the niche. Ghiordes rugs command rather higher prices than 
 similar rugs in equal condition, A good specimen is considered to 
 show the best work done in Anatolia during recent centuries. 
 
 Plate 99. Kis (i.e.. Maiden) -Ghiordes Carpet. 
 
 These are given as presents to brides, or according to another account 
 are woven by a bride as a sample of her skill in carpet-making. They 
 have from 60 to 90 knots to the square inch. The si^e is generally about 
 3^ ft. X 6 ft. They are rough to the touch, and have a short and lustreless 
 pile. The chief colours are white, blue and red. The arrangement 
 
 156 
 
CARPETS IN THE MARKET : GROUPING— IDENTIFICATION 
 
 of the fields both with regard to the medallion and the shape of the corners, 
 follows a type of rug common in the sixteenth century. The main 
 border-stripe, which is a corruption of an early pattern, is peculiar to 
 Kis-Ghiordes rugs. The small objects forming the diaper of field and 
 border may possibly be conventionalized lamps. 
 
 Plate ioo. Kula Prayer -carpet. 
 
 Carpets of this kind are known in the trade very generally, but 
 erroneously, as Ghiordes. They have from 6o to 90 knots to the square 
 inch. The size is about 4 ft. X 6 ft. They are fairly hard to the touch. 
 The pile is short and differs from that of most of the Anatolian prayer-rugs 
 in being without any lustre. The chief colours are blue, red and yellow. 
 This example has the typical Kula colouring, the red being fairly well 
 preserved. It has not the usual repeated narrow stripes in the border, 
 but the arrangement of leaves in the principal stripe produces much the 
 same effect. Antique pieces, probably owing to the short pile and rather 
 loose weave, are generally in very bad condition. Kula prayer-carpets 
 with a purple ground are not much prized. 
 
 Plate ioi. Kula Prayer-carpet. 
 
 Carpets of this kind are called ** Cemetery Carpets,^' and in the 
 trade are also, erroneously, known as Ghiordes. They have from 60 
 to 90 knots to the square inch. The usual size is about ft. to 6 ft. 
 They are somewhat rough to the touch and have a short and lustreless 
 pile. The chief colours are blue, red and yellow. The repeated designs 
 in the field represent a tomb with a cypress and another tree ; from 
 which, of course, the name ** Cemetery Carpet ** is derived. 
 
 Plate 102. Ladik Prayer-carpet. 
 
 Carpets of this kind are known in the trade also as Ghiordes or 
 Kula. They have from 65 to no knots to the square inch. The size 
 varies from 3 ft. X 4 ft. to 4 ft. X 6 ft. They have a lustrous pile of 
 medium length. The principal colours are red, blue, green, and yellow. 
 
 Plate 103. Ladik Prayer-carpet (Dated a.h. 1211 = a.d. 1797). 
 
 Carpets of this kind are also known in the trade as Ghiordes or 
 Kula. They have from 65 to 90 knots to the square inch. The size 
 is generally from 3 ft. X 4^ ft. to 4 ^ ft. X 6 ft. They are soft and flexible 
 and have a lustrous pile of medium length. The chief colours are red, 
 blue, green, and yellow. The main border-stripe is almost peculiar 
 to Ladiks. 
 
 157 
 
HAND-WOVEN CARPETS 
 Plate 104, Ladik Carpet* 
 
 Carpets of this kind have 50 to 70 knots to the square inch. The 
 size is about 3 ft. X 4I ft. They are soft and pliable^ with a lustrous 
 pile of medium length. The chief colours are light red^ blue, and green. 
 This design would not often be found ; but the Plate serves as an example 
 of the kind of pattern seen in Ladiks that are not prayer-rugs. 
 
 Plate 105. Ladik Carpet. 
 
 Carpets of this kind have about 70 knots to the square inch. The 
 size is 3I ft. X 7 ft. They are soft and pliable with a lustrous pile of 
 medium length. This specimen is not characteristic of the type, and 
 clearly has been affected by nomadic influences, as can be seen by the 
 small conventional animals and other detached devices. A study of 
 the illustration without an examination of the actual carpet would be 
 insufficient to connect this with certainty with the typical Ladiks. It 
 shows well, however, the kind of variation from type that may be met 
 with. 
 
 Plate 106. Mujur Prayer-carpet. 
 
 Carpets of this kind are known in the trade also as Anatolian or 
 Sultan carpets. They have from 45 to 65 knots to the square inch. The 
 usual size is ft. X 5 ft. They are soft and pliable and have a fairly 
 short and lustrous pile. The chief colours are red, green, blue, yellow 
 and violet. The two largest border-stripes are characteristic. It should 
 be noted that the cross-panel is an abbreviated form of that commonly 
 seen in Ladiks. These rugs have a very sparkling appearance. They 
 are made in the town of Mujur on the river Kisil-Irmak. 
 
 Plate 107A. Kir-Shehr Prayer-carpet. 
 
 Carpets of this kind are known in the trade also as Sultan or Anatolian 
 carpets. They have from 45 to 65 knots to the square inch. The usual 
 size is about 3^ ft X 4I ft. They are soft and pliable and have a lustrous 
 pile of medium length. The chief colours are red, green, white, and 
 blue. The narrow stripes in the border are more clearly marked in 
 most specimens. The multiple bands forming the arch may be contrasted 
 with those in Plate 57. They are made in the town of Kir-Shehr on the 
 river Kisil-Irmak. 
 
 Plate 107B. Anatolian Saph (or Family) Prayer-rug. 
 
 Carpets of this kind have about 50 to 70 knots to the square inch. 
 
 158 
 
CARPETS IN THE MARKET : GROUPING— IDENTIFICATION 
 
 The size is usually about 3^ ft, x 11 ^ ft. They are soft and pliable 
 and have a lustrous pile of medium length. Each carpet is equivalent 
 to several single prayer-rugs joined together and it is from this that their 
 name is derived. The niches are variously coloured in the same carpet. 
 These carpets are sometimes erroneously supposed to come from Chinese 
 Turkestan. Some, however, are found with patterns so close to other 
 Anatolian rugs, such as Mujurs, that no mistake can be made in their 
 case. 
 
 Plate io8a. Jakshibehdir (i.e., charcoal-burner) Prayer-carpet. 
 
 Carpets of this kind constitute a subdivision of Bergama carpets 
 and are often called by that name. They have from 50 to 65 knots to 
 the square inch. The size is generally about 2 ^ ft. X 3^ ft. They 
 are soft and pliable and have a long and often very close pile. The chief 
 colours are dark blue and dark red ; and the whole colour-scheme is 
 sombre, whence is derived the name. 
 
 Plate io8b. Bergama Carpet. 
 
 Carpets of this kind have from 50 to 80 knots to the square inch. 
 The size ranges from 2| ft. X 3J ft. to 4 ft. X 5 ft. They are soft and 
 pliable, and have a fairly long and lustrous pile. The chief colours are 
 dark cherry-red, dark blue, dark brown, and that of natural camel-hair. 
 The pattern is often picked out with yellow and white. 
 
 Plate 109. Anatolian Kilim Prayer-carpet. 
 
 Carpets of this kind are tapestry-woven and have about twelve warp- 
 threads to the linear inch. They are soft and thin. The size is about 
 ft, X 5I ft. The colours are very bright but the effect, even in the 
 modern specimens, is rarely displeasing. Large numbers are made by 
 the mountain-tribes from the region of Erzerum ; but the manufacture 
 is fairly widespread. It is quite common to find metal thread woven in 
 with the wool. Kilims, other than prayer-rugs, are made in still greater 
 numbers. Many of the largest are in two strips sewn together down 
 the middle. The patterns are often reminiscent of those of the well- 
 known types of pile-rugs, though the technique does not allow a very 
 close adherence. They are extremely useful for hangings and covers 
 of all sorts, and are comparatively inexpensive. 
 
 Plate no. Anatolian Carpet. 
 
 Carpets of this kind are also known in the trade as Mosul, Yuruk, 
 Kurd, or Montagnard. They have from 40 to 60 knots to the square 
 
 159 
 
HAND-WOVEN CARPETS 
 
 inch. The size is usually about 3I ft. x 7I ft. They are generally made 
 in the mountainous regions and are pliable and have a long^ rough and 
 lustrous pile. The shape is almost invariably irregular. The colours 
 are chiefly cinnamon-red^ dark blue, and yellow. They are closely 
 akin to the Yuruk carpets. 
 
 Plate iiia. Makri Carpets. 
 
 Carpets of this kind are also called Rhodian Carpets. They have 
 from 45 to 65 knots to the square inch. The size is about 3I ft. X ft. 
 They are rather stiff, and have a lustrous pile of medium length. The 
 chief colours are blue, yellow and red. The sparkling patches of colour, 
 together with the arrangement of the field, give an effect which has 
 been aptly compared to that of a stained-glass window. 
 
 Plate iiib. Anatolian Carpet. 
 
 Carpets of this kind are also known in the trade as Sultan carpets. 
 They have from 45 to 65 knots to the square inch and range in size from 
 2 ^ ft. X 3 ft. to 3|- ft. X 4 ^ ft. They are soft and pliable, and have 
 a long and lustrous pile. The chief colours are red, yellow, blue and 
 brown. 
 
 Plate ii2A. Anatolian Carpet. 
 
 Carpets of this kind are also known in the trade, on account of their 
 light background, as Akterai (i.e.. White Castle). They have from 40 
 to 60 knots to the square inch, and are usually small and almost square 
 in shape. They are soft and pliable and have a fairly short and lustrous 
 pile. The colour of the ground is white or yellow, and the pattern is 
 dark in tone, often with brown-black predominating. The likeness 
 to Caucasian design, which is so commonly noticed in Anatolian carpets, 
 is very marked here. 
 
 Plate ii2B. Mel as Carpet. 
 
 Carpets of this kind have from 60 to 80 knots to the square inch. 
 They are soft and pliable, and have a very lustrous pile of medium length. 
 Older pieces, which are now getting very scarce, are generally striped as 
 in the illustration and are very bright in colour. 
 
 Plate 113. Melas Prayer-carpet. 
 
 Carpets of this kind are known in the trade also as Anatolian. They 
 have from 60 to 80 knots to the square inch. The size is generally about 
 2 \ ft. X 4 ft. They are made in the town of Melassa and its environs 
 
 160 
 
CARPETS IN THE MARKET : GROUPING— IDENTIFICATION 
 
 near to the south-west coast of Asia Minor, and are soft and pliable and 
 have a fairly short but very lustrous pile. The chief colour is strawberry- 
 red, but there is often a good deal of yellow in the border. This example 
 has the characteristic Melas prayer-arch. 
 
 Plate 114. Yuruk Carpet. 
 
 Carpets of this kind are sometimes erroneously known in the trade 
 as Daghestan or Kasak. They have from 50 to 70 knots to the square 
 inch. The usual si2;es are 2^ ft. X ft., 3i ft. X 5I ft. and 3 ft. X 7 ft. 
 They are soft and pliable and have a long, lustrous pile. The chief 
 colours are dark red, brown, dark blue, and white. The border of this 
 specimen is characteristic of the type. 
 
 (6) Caucasian 
 
 Although the production of carpets in the Caucasus does not, as 
 far as is known, go back much beyond the eighteenth century, and although 
 the proximity of both Turkey and Persia — two such great carpet- weaving 
 countries — has naturally had much influence, yet the Caucasian carpets 
 are a more independent and well-marked group than might be expected. 
 On the whole the influence of Turkey has been greater than that of Persia, 
 especially with regard to technique. 
 
 Caucasian carpets are made almost entirely of wool, though, as will 
 be seen later, a little cotton occurs in the foundation of some of the rugs. 
 The warp is mostly undyed and ranges in colour from pure white to a 
 deep brown. Wools of different colour are often mixed before spinning, 
 but even when the colour of each strand is uniform, it is often found 
 that differently- coloured yarns are twisted together to make a thread. 
 The weft also is commonly undyed. The normal plan is to have two 
 shoots of weft after each row of knots, but a feature in some of the nomadic 
 rugs is the occurrence of three or more shoots at irregular intervals. The 
 Ghiordes knot is exclusively employed in the pile-carpets. The pile is 
 invariably of wool which is mostly of good quality and lustrous in 
 appearance. In the nomadic rugs the pile is long and in the others of 
 medium length, or short, but never very short. The sides are usually 
 finished with an overcast woollen selvedge of several cords, though the 
 Persian method of employing a single-cord selvedge is not unknown. 
 The ends are very simply treated, there being a short plain web and a 
 simple fringe, after perhaps a few rows of knots formed by tying small 
 groups of warp -threads together. 
 
 161 
 
 M 
 
HAND-WOVEN CARPETS 
 
 In the design of Caucasian rugs, naturalistic drawing is not found, 
 and floral forms, though perhaps not more conventionalized than in 
 Turkish rugs, are not used to so great an extent. Geometrical design, 
 which does not differ in its general aspect from that in Turkish rugs, 
 is very much used, but though the general aspect is similar, in the details 
 of the drawing it is easy to notice a difference between Caucasian and 
 Turkish designs. The same borders and field devices are found over 
 and over again in the two countries, but the treatment of the Caucasian 
 version is always distinct from that of the Turkish (cf. Plate 196, No. 8, 
 with Plate 194, No. 22), though it is not easy to give a general description 
 of the kind of change that is found. Perhaps it may be said that the 
 Caucasian rendering shows greater angularity and more quaintness. 
 
 Many patterns have also been borrowed from Persia. Borders 
 with wavy floral stems, so rarely found in Turkey, are quite common, 
 though the treatment is decidedly more angular and the blossoms have 
 usually degenerated to almost geometrical rosettes. Lattice field-diapers, 
 which are met with in a few types of rugs, have also a Persian origin. 
 Caucasian carpets have, in addition, numerous field and border patterns 
 quite peculiar to themselves. Generally the masses of colour are greater 
 in Caucasian than in Turkish or Persian rugs, and fine detail is not common. 
 The popularity of the latch-hook is very probably due to its utility in 
 softening the outline of a large mass of colour. Central medallions, 
 which are so common in Persian rugs, are almost unknown ; while 
 scattered devices such as stars, octagons, rosettes, quite unrelated to 
 each other, are much employed. It might be said that the rugs of Persia 
 are designed from the centre and those of the Caucasus from the 
 ends. 
 
 Among other features commonly seen in the field may be mentioned 
 rectangular panels ; large irregular forms of radiating contour, sometimes 
 known as sunbursts ** ; and small conventional renderings of animals, 
 birds, and human figures. 
 
 Plates 195 to 199 contain motives for the most part peculiar to 
 Caucasian carpets. Counterchanged border patterns are shown in Plate 
 196 (i, 2 and 6), and Plate 199 (17); and various renderings of S-forms 
 are seen in Plate 196 (5, 8 and 9) ; Plate 198 (6, ii and 15), and Plate 
 199 (7)* It is probable that Plate 196 (3, 4 and 10), and Plate 199 (12 
 and 16) are all debased forms of early borders with Kufic lettering. Plate 
 196 (19) is the ** Crab ** pattern, almost peculiar to Kazaks, and Plate 196 
 (22) is the ** Leaf and Wine-cup ** border very often seen in Shirvans. 
 
 162 
 
CARPETS IN THE MARKET : GROUPING— IDENTIFICATION 
 
 Plate 198 (18) is the so-called ** Georgian ** pattern ; and Plate 198 (19) 
 is the border characteristic of Chichi rugs. 
 
 The colouring is usually bright and vigorous but rarely unpleasing ; 
 even the modern rugs being as a rule satisfactory in this respect. The 
 dyes of the latest rugs are not on the whole inferior to those of other 
 countries^ though a certain — and too large — amount of crude and unsound 
 colours is met with. 
 
 The classification of Caucasian carpets into definite types, depending 
 upon the place of their manufacture, is a very difficult problem even 
 for the experienced connoisseur. The attempt of many writers to do 
 more than is possible in this respect has led to much confusion, and to 
 the bewilderment of amateurs, who cannot understand the distinction 
 between the types as defined in the text-books, and despair of reconciling 
 the description given by dealers with these definitions. Only a very 
 simple classification will be attempted here, though the various trade 
 names will be mentioned with as much, or as little, information as can 
 be given about their significance. 
 
 The various Caucasian carpets may be divided into two main groups, 
 according as they are produced north or south of the great mountain 
 range which crosses the country from north-west to south-east. To 
 the north of the mountains lies the province of Daghestan, which has 
 given its name to some of the best Caucasian carpets produced. They 
 are fine in texture and have frequently a cotton warp and weft. Lattice- 
 diapers in the field, and a diagonal arrangement of small devices, are 
 common. The drawing is refined and the detail finer than in most of 
 the Caucasian products. There is a tendency towards a mosaic effect 
 in the design. The weaving, as seen from the back, is very regular, 
 Kabistan is a name given to some Daghestan rugs. It is not claimed 
 that the difference is great but they are said to be slightly inferior. As 
 the word Kabistan is a corruption of Kubistan, which is derived from the 
 town of Kuba, and as the name Kuba is assigned to another alleged type, 
 it will be seen at once that the geographical distinction is negligible, 
 Derbend, or Derbent, the main town of Daghestan and the 
 distributing centre of its rugs, is the name given to a somewhat coarser 
 variety of rug with a woollen foundation. 
 
 If a Daghestan rug has a rather floral pattern, or exhibits large 
 sunbursts,'^ it is called a Kuba, from an important town towards the 
 south, 
 
 Lesghian is a name applied to nomadic rugs of the province, but 
 
 163 
 
HAND-WOVEN CARPETS 
 
 no serious attempt has been made to distinguish them from Trans- 
 Caucasian nomadic rugs, though they are said to have a selvedge of several 
 cords rather than of one. 
 
 To the south of the mountains lies the district of Shirvan, which 
 produces many rugs, and the name is applied to all Caucasian carpets 
 that do not appear to belong to one of the other types. They are similar 
 to the products of Daghestan, but tend to be coarser and to have larger 
 and less refined patterns and colouring. Cotton is more rarely used. 
 So great is the variety of design found in Shirvan rugs that it is almost 
 impossible to give an adequate idea of it, A few types of field design, 
 however, are so common that they must be mentioned. One of these 
 is a succession of diamond-shaped panels, each the full width of the 
 field, and touching the next one at the angle. The diamonds are mostly 
 fringed with latch-hooks. Often the diamonds are so deeply stepped 
 as to appear rather octagonal or cross-shaped in form. Another plan 
 is to have a succession of large rectangular panels, separated only by a 
 narrow space, A modification of this arrangement is to have the field 
 itself contracted at intervals, so that virtually a succession of rectangles 
 connected by wide necks is formed. This last design is characteristic 
 of Shirvans, Nearly all the border-stripes found in Shirvan rugs are 
 also used in other Caucasian types, but one much favoured is known as 
 the ** leaf and wine-cup ** border, A wavy line of serrated leaves has 
 in the interspaces devices resembling wine-cups on stems, but probably 
 really conventionalized floral buds, 
 
 A group of Shirvans with a border peculiar to itself (see Plate 198 (19) ) 
 are known as Chichi, Tchetcen, or Tzitzi from the name of the tribe 
 that makes them. 
 
 The city of Baku gives its name to rugs with dull, faded-looking 
 colours and a pattern of large, elaborately- drawn cones of peculiar form 
 in the field. 
 
 An important group of rugs made by wandering tribes and having 
 the usual characteristics of nomadic weavings, is called by the name 
 Kazak, which is a form of the word Cossack, The pile is long and 
 lustrous and the weft often crosses three or more times at intervals. The 
 sides are finished with a selvedge of several cords, overcast with wool of 
 different colours, which extends for various distances along the side. 
 These changes of colour in the selvedge seem to be quite capricious but 
 are highly characteristic of the type. The ends have fairly wide coloured 
 webs and knotted fringes. The patterns of Kazak rugs are very varied 
 
 164 
 
CARPETS IN THE MARKET : GROUPING— IDENTIFICATION 
 
 and are not dissimilar to the more geometrical of the Shirvans. There 
 is usually only one wide border-stripe, which often has patterns peculiar 
 to the type (e.g* Plate 196 (i9))» The dyes are mostly sound and the 
 colouring, though bold and bright, is almost invariably very pleasing. 
 Though not of very fine texture, these rugs are among the most satisfactory 
 to be had nowadays. 
 
 The towns Tiflis and Kutais give their names to unimportant 
 or indefinite types of rugs, resembling Kazaks, but differing from them 
 in having cone-patterns and tree-forms in the design. 
 
 Tcherkess (or Circassian) rugs are Kazaks of a tawny colour, and 
 with heavy, braided ends. 
 
 There is a flourishing carpet industry in the southern province of 
 Karabagh. The ordinary specimens bear the name of the district and 
 the better ones are called after the town of Shusha. The latter tend 
 to be broader in proportion. Both kinds are rather loosely woven and 
 have a pile of medium length. The weft is often dyed red, and the 
 selvedge is usually in lengths of different colours like in the Kazaks. 
 The designs have a tendency to resemble Persian models. The colouring 
 of Karabagh carpets is characteristic, as it nearly always includes an 
 amount of bright purple and light blue. Shusha carpets have a rich 
 colour scheme, consisting chiefly of dark blue, dark red and brown. 
 
 Rugs called after the town of Gengha in the same province are 
 very much like the Kazaks. They have, however, an unusually large 
 number — up to six or eight — shoots of weft at frequent intervals ; and 
 white is a prominent colour in the design. 
 
 Kilims are woven throughout the Caucasian region. They are 
 sometimes not easy to distinguish from the similar rugs of Turkey, but 
 often there is so much resemblance in their patterns to exclusively Caucasian 
 designs as to put their origin beyond doubt. 
 
 Thin and soft Kilims made in the neighbourhood of Shusha are 
 known in the trade by the name of Verne. 
 
 Another variety of Kilim, called Sile, is distinguished by its type 
 of design. 
 
 The woven carpets known as Soumaks are said to get their name 
 from the Caucasian town of Shemakha. The chief centre of their 
 production is, however, Derbend. Soumaks as usually met with range 
 in size from about 5 ft. X 7 ft. to 7 ft. X 10 ft. ; though there is in the 
 Victoria and Albert Museum an exceptionally finely-woven one measuring 
 only 3 ft. 6 in. X 7 ft. A very usual field pattern consists of three or 
 
 165 
 
HAND-WOVEN CARPETS 
 
 more flattened octagons ; but the variety in design is considerable* The 
 older Soumaks have a deep and rich colouring, but the modern ones 
 are cruder and often have unsound dyes. 
 
 On the whole the Caucasian region may be said to produce most 
 desirable carpets. Their prices are not high, and it is at least as easy 
 to find among them, as anywhere else, soundly-woven rugs of good colour 
 and design. 
 
 Plates 1 15 to 146 show most of the commoner types of these carpets. 
 Plate 115. Daghestan Carpet. 
 
 Carpets of this kind are known in the trade also as Karabagh, Ka^ak, 
 Kenguerlu, or simply as Persian. They have from 50 to 80 knots to 
 the square inch. The usual sizes are about 2| ft. X 3|- ft. and 4 ft. X 
 6^ ft. They have a soft and lustrous pile of medium length. The chief 
 colours are red, blue and white ; the first two alternating with the third 
 in the field. 
 
 Plate 116. Daghestan Prayer- carpet (Victoria and Albert Museum). 
 Dated a.h. 1287 (= a.d. 1870). 3 ft. 10 in. x 3 ft. 2 in. 
 Warp : Three-ply white wool. On one level. 21 to i inch. 
 
 Weft : White cotton. Two shoots after each row of knots. 
 Knots : Wool. Ghiordes. 10 to i inch. 105 to the square inch. 
 Colours : Nine. White (field and spandrels) : red (2nd border) : 
 dark blue (3rd border) : blue : black (ist border) : ochre : 
 light green : brown : dark red. 
 
 Plates 117 and ii8a. Daghestan Carpets. 
 
 Carpets of this kind are known in the trade also as Kazak, Gengha, 
 Karabagh, or simply as Persian. They have from 50 to 80 knots to the 
 square inch. The usual sizes are about 2^ ft. X 3|- ft., 4 ft. x 9 ft. 
 and ft. X ft. They have a soft and lustrous pile of medium length. 
 
 The chief colours are red and blue, with some white and grey ; the main 
 border-stripe usually being white. 
 
 Plate ii8b. Caucasian Kilim. 9 ft. 4 in. x 5 ft. 6 in. 
 
 Warp : Brown and white wool. 15 to i inch. 
 
 Weft : Wool. About 55 shoots to i inch. 
 
 Colours: Ten. Crimson (field) : white: green-blue: black (border) : 
 blue : dark red : red : yellow : ochre : light red. The pattern 
 of this carpet is a rendering, as close as a somewhat coarse 
 tapestry texture will allow, of the well-known ** sunburst ** 
 device. 
 
 166 
 
CARPETS IN THE MARKET : GROUPING— IDENTIFICATION 
 
 Plate 119. Derbend Carpet. 
 
 Carpets of this kind are known in the trade also as Persian. They 
 have from 45 to 60 knots to the square inch. The usual si2;e is 5 ft. X 
 10^ ft. They are soft and pliable and they have a long and very lustrous 
 pile. The chief colours are blue and red, with some white. The pattern 
 of the illustrated specimen with its curious bracket-like forms is typical 
 of Derbend rugs. The main border-stripe is adapted from a common 
 Ghiordes pattern. Old rugs of this type are rare, and modern pieces 
 fetch rather higher prices than other Caucasian carpets of similar quality. 
 
 Plate 120A. Daghestan Carpet. 
 
 Carpets of this kind are known in the trade also as Karabagh, Gengha, 
 Mosul, or simply as Persian. They have from 50 to 100 knots to the 
 square inch. The usual sizes are 3 ft. X 4 ^ ft. and 4 ft. X 9 ft. They 
 have a lustrous pile of medium length. The chief colours are blue, red 
 and white ; the darker stripes in the field being often alternately blue and 
 red. Such a diagonal arrangement of the field is common in Daghestans. 
 
 Plate 120B. Shirvan Carpet. 
 
 Carpets of this kind are known in the trade also as Moghan, Talish, 
 Daghestan, and also simply as Persian. They have from 60 to 85 knots 
 to the square inch. The size is about 3^^ ft. X ft. They are fairly 
 hard to the touch and have a short and not very lustrous pile. The 
 chief colours are dark blue, red and white. The large- tailed birds in 
 the field should be noticed. 
 
 Plate 121. Shirvan Carpet. 
 
 Carpets of this kind are known in the trade as Kuba, Kabistan, 
 Talish, Gengha, and also simply as Persian. They have from 60 to 100 
 knots to the square inch. The size is about 3^ ft. X 9 ft. They are 
 fairly hard to the touch and have a short and not very lustrous pile. The 
 colours are chiefly red and blue with lesser quantities of white and yellow. 
 This specimen has the field contracted at intervals, which arrangement 
 is peculiar to Shirvans ; and in the border are two stripes with a small 
 repeating pattern of carnations, which is found in various types of Caucasian 
 carpets. 
 
 Plate 122A. Shirvan Carpet. 
 
 Carpets of this kind are known in the trade also as Kuba, Kabistan, 
 Talish, Gengha, and also simply as Persian. They have from 60 to 100 
 
 167 
 
HAND-V/OVEN CARPETS 
 
 knots to the square inch. The si2;e is about si ft. X 9 ft. They are 
 fairly hard to the touch, and have a short and very lustrous pile. The 
 chief colours are red and blue with a smaller quantity of white. This 
 specimen shows the cross-panels and the '' leaf and wine-cup border 
 which are so common in Shirvans. 
 
 Plate i22B. Shirvan Carpet. 
 
 Carpets of this kind are also known in the trade as Kuba, Kabistan, 
 Talish, Gengha, and also simply as Persian. They have from 60 to 85 
 knots to the square inch. The size is about si ft. x 9 ft. They are 
 fairly hard to the touch and have a short and not very lustrous pile. The 
 chief colours are red, blue, and lesser quantities of white and yellow. 
 This carpet with its rayed medallions is of the kind mostly called Kuba. 
 The peculiar pattern in the widest border-stripe is usually known as 
 the ** Georgian ** border, after the district of Georgia. Its origin is not 
 clear, though there is little doubt that it is derived from some floral form. 
 It is quite peculiar to Caucasian carpets. 
 
 Plate 123. Shirvan Carpet. 
 
 Carpets of this kind are known in the trade also as Daghestan or 
 simply Persian carpets. They have from 60 to 85 knots to the square 
 inch. The size is about si ft* X 5 ft. They are fairly hard to the touch 
 and have a short and not very lustrous pile. The chief colours are dark 
 blue, red and white. The main border-stripe has the leaf and wine- 
 cup pattern. 
 
 Plate 124. Shirvan Carpets. 
 
 Carpets of this kind are knov/n in the trade also as Daghestan, Kasak, 
 or simply as Persian. They have from 50 to 75 knots to the square inch. 
 The size is about ft. X 5 ft. They are rather hard to the touch and 
 have a short and not very lustrous pile. The chief colours are dark blue 
 and red, with little or no white. 
 
 Plate 125. Shirvan Carpet. 
 
 Carpets of this kind are known in the trade also as Daghestan, Mecca, 
 or simply as Persian. They have from 65 to 100 knots to the square inch. 
 The size is about si ft* X ft. They are rather hard to the touch, 
 and have a short and not very lustrous pile. The chief colours are dark 
 blue with lesser quantities of red and white. The pattern of the field 
 in this specimen is a comparatively rare one. 
 
 168 
 
CARPETS IN THE MARKET : GROUPING— IDENTIFICATION 
 
 Plate 126. Shirvan Carpet, 
 
 Carpets of this kind are also known in the trade as Kuba^ Kabistan, 
 Talish^ Gengha^ or simply as Persian. They have from 60 to 85 knots 
 to the square inch. The size is about 3^ ft. X 9 ft. They are fairly 
 hard to the touch and have a short and not very lustrous pile. The chief 
 colours are red and blue with lesser quantities of v/hite and yellow. The 
 main border-stripe with two kinds of angular S -forms is based on a 
 common Turkish pattern (see Plate 194^ No. 22)/ though this rendering 
 is typically Caucasian. 
 
 Plate 127. Shirvan or Chichi Carpet. 
 
 Carpets of this kind are known in the trade also as Daghestan^ Mecca, 
 or simply as Persian. They have from 50 to 85 knots to the square inch. 
 The size is about 3|- ft. X 6| ft. They are fairly hard to the touch and 
 have a short and not very lustrous pile. The chief colours are dark 
 blue, with lesser quantities of red and white. The main border-pattern, 
 consisting of rosettes separated by diagonal ribbon-like bands, is that 
 associated with the name ** Chichi."' 
 
 Plate 128. Shirvan or Baku Carpet. 
 
 Carpets of this kind are known in the trade also as Mecca, Daghestan, 
 Koltuk, Kuba, or simply as Persian. They have from 50 to 85 knots 
 to the square inch. The size is about ft. X 5 ft. or 4I ft. X 9 ft. 
 They are rough to the touch and have a short and not very lustrous pile. 
 The chief colours are blue, red and purple, often rather faded. 
 
 Plate 129. Shirvan or Baku Carpet. 
 
 Carpets of this kind are known in the trade also as Daghestan, Mecca, 
 and simply as Persian. They have from 60 to 90 knots to the square 
 inch. The usual sizes are 3I ft. X 5^^ ft. and 5 ft. X 9 ft. They are 
 rough to the touch and have a short and not very lustrous pile. The chief 
 colours are blue and red with a lesser amount of white. The conventional 
 birds seen in this specimen are very common in Baku rugs. 
 
 Plate 130. Shirvan Carpet. 
 
 Carpets of this kind are known in the trade also as Derbend, Kuba, 
 or simply as Persian. They have from 60 to 90 knots to the square inch. 
 The usual sizes are 3^ ft. X 5 ft. and 5 ft. X 9 ft. They are rough to the 
 touch and have a short and not very lustrous pile. The chief colours 
 
 169 
 
HAND-WOVEN CARPETS 
 
 are blue and red, with a lesser amount of white. The pattern of this 
 specimen is not so common as those of the other Shirvans illustrated. 
 
 Plate 131. Shirvan Carpet. 
 
 Carpets of this kind are known in the trade also as Mecca^ or simply 
 as Persian. They have from 60 to 90 knots to the square inch. The 
 usual si^e is about 4 ft. X 9 ft. They are rough to the touch and have 
 a short and not very lustrous pile. The chief colours are blue and red, 
 with a lesser quantity of white. The dotted stripes of several colours 
 in the border and the angular S -forms are typically Caucasian, though 
 a border somewhat like the former is found in the Persian rugs of Shiraz. 
 It may be mentioned here that Shirvan and Shiraz rugs, though coming 
 from such distant places, are often confused, and the name Mecca ** 
 is erroneously given to both classes. A close study of the technique 
 will soon show the difference between the two, even when they are 
 superficially alike in pattern. 
 
 Plates 132A and 133. Kazak Carpets. 
 
 Carpets of this kind are known in the trade also as Daghestan, 
 Gengha, or simply as Persian. They have from 50 to 80 knots to the 
 square inch. The usual sizes are about ft. X 4 ft., 4 ft. X 9 ft. and 
 4I ft. X 6-| ft. They have a thick, long and lustrous pile. The chief 
 colours are red and blue, with a lesser amount of white, the ground 
 of the main border-stripe usually being white. The border-pattern, 
 which is seen in two degrees of conventionalization, is derived from the 
 motive of four leaves springing from a rosette ; but from the resulting 
 form is often known as the Crab ** pattern. 
 
 Plate 132B. Shirvan or Baku Carpet. 
 
 Carpets of this kind are known in the trade also as Daghestan, or 
 simply as Persian. They have from 50 to 85 knots to the square inch. 
 The size is about 3 ft. X 5 ft. or 4|- ft. X 9 ft. They are rather hard 
 to the touch and have a short and not very lustrous pile. The chief 
 colours are blue, red, and some purple, but they are often faded to a 
 dull brownish appearance. This specimen has the typical pattern of 
 elaborately drawn cones, which is associated with the name '' Baku.'' 
 
 Plate 134. Kazak Carpet. 
 
 Carpets of this kind are known in the trade also as Gengha, Karabagh, 
 or simply as Persian. They have from 50 to 80 knots to the square inch. 
 
 170 
 
CARPETS IN THE MARKET : GROUPING— IDENTIFICATION 
 
 The usual siz:es are 2,\ ft* X 4 ft*, 4 ft* x 9 ft* and 4|- ft* x 6 ^ ft* They 
 have a long, thick and very lustrous pile* The chief colours are red, 
 blue and white* 
 
 Plate 135* Kazak Carpet* 
 
 Carpets of this kind are known in the trade also as Daghestan, Gengha, 
 or simply as Persian* They have from 50 to 80 knots to the square inch* 
 The usual size is about 5 ft* X 6| ft* They have a thick, long and very 
 lustrous pile* The chief colours are red and blue, with some white and 
 green, the ground of the main border-stripe being commonly white* 
 
 Plates 136 and 137* Kazak Carpets* 
 
 Carpets of this kind are known in the trade also as Daghestan, Gengha, 
 Derbend, or simply as Persian* They have from 50 to 80 knots to the 
 square inch* The usual sizes are ft* X 4 ft*, 4 ft* x 9 ft* and 4|- ft* x 
 ft* They have a thick, long and very lustrous pile* The chief colours 
 are red and blue, with some white and green, the border usually having 
 a fair amount of white* 
 
 Plate 138* Kazak Prayer-carpet* 
 
 Carpets of this kind are known also in the trade as Daghestan, Gengha, 
 Derbend, or simply as Persian* They have from 50 to 80 knots to the 
 square inch* The usual sizes are si ft* X 4 ft*, 4 ft* X 9 ft* and 4^ ft* X 
 6i ft* They have a thick, long and very lustrous pile* The chief colours 
 are red and blue, with some white and green, the border usually having 
 a fair amount of white* The small angular arch at the top of the field, 
 and the oblong recess at the bottom, constitute the usual arrangement 
 of Kazak prayer-rugs* 
 
 Plates 139, 140 and 141* Kazak Carpets* 
 
 Carpets of this kind are known in the trade also as Daghestan, 
 Shirvan, Gengha, or simply as Persian* They have from 50 to 80 knots 
 to the square inch* The usual sizes are 2| ft* X 4 ft*, 4 ft* X 9 ft* and 
 4i ft* X 6i ft* They have a thick, long and very lustrous pile* The 
 chief colours are red and blue, with some white and green, the border 
 usually having a fair amount of white* 
 
 Plate 142* Karabagh Carpet (Victoria and Albert Museum)* 
 8 ft* 9 in* X 4 ft* 10 in* 
 
 Warp : White wool* On one level* 16 to i inch* 
 
 Weft : White wool* Two shoots after each row of knots* 
 
 171 
 
HAND-WOVEN CARPETS 
 
 Knots : Wool. Ghiordes. 6 to i inch. 50 to the square inch. 
 
 Colours : Ten. Purple (field and 2nd border) : light blue (ist 
 border) : dark blue (panels in field) : white : yellow : black : 
 light red : brown : red : green. 
 
 Plates 143 and 144. Verne Carpets. 
 
 These are made by the ordinary Kilim process, and are thin, soft 
 and almost alike on both sides. The usual si^e is about 4 ft. X 5 ift. 
 The chief colours are Indian red, white and yellow. They are mostly 
 used for covers. 
 
 Plate 145. SiLi Carpet. 
 
 Carpets of this type are woven by the ordinary Kilim process. They 
 often consist of two pieces sewn together and are thin, soft, and almost 
 alike on both sides. Their principal use is for covering tables and for 
 similar purposes. The usual ske is 4 ft. x 5|- ft. The chief colours 
 are Indian red, white and yellow. 
 
 Plate 146. Soumak Carpet. 
 
 Carpets of this kind are known in the trade also as Persian. They 
 are woven by the special Soumak method, and have from 8 to 25 warp- 
 threads to the inch. The usual sizes are about 5^ ft. x ft., 6| ft. x 
 10 ft. and, more rarely, 8 ft. X ii| ft. The chief colour is dark Indian 
 red, with the pattern picked out in yellow. The flattened octagons in 
 this specimen are typical of the group. 
 
 (c) Persian 
 
 Of all the great classes into which the carpets of the East are divided, 
 the carpets of Persia have the highest reputation, and on the whole this 
 reputation is deserved. It must not be supposed for a moment that 
 Persian carpets of medium or of good quality are better than the best 
 from other parts, but the finest Persian specimens, either of the past 
 or the present, do certainly surpass all others in refinement of drawing, 
 in subtlety and harmony of colour, and in perfection of weaving. There 
 is so much variety to be found in the carpets that come from different 
 parts of this vast region that it is somewhat difficult to give a general 
 description of them, and in order to make the subject manageable it is 
 almost necessary to divide them into geographical groups before discussing 
 the separate types ; but as far as it is possible to treat the whole class 
 -together, it will be advisable to do so. 
 
 172 
 
CARPETS IN THE MARKET : GROUPING— IDENTIFICATION 
 
 Even in Persia wool is by far the commonest material used ; but 
 cotton, which is rare in Turkey and Caucasia, is employed to a very large 
 extent in the foundation of carpets ; and silk, which is quite exceptional 
 in those countries, is here found fairly often. For the warp, wool and 
 cotton are used about equally, and both are as a rule undyed. The 
 arrangement of the warp on two levels, one set being right behind the 
 other, is very common in Persia, but scarcely ever seen in other countries. 
 The weft is usually of the same material as the warp. A woollen weft 
 is commonly undyed, but in a few types is coloured red or brown. When 
 the weft is of cotton, it is frequently dyed, blue and red being the favourite 
 colours. Two shoots of weft is the usual number, but in a few types 
 only one shoot is employed, and in many of the older carpets, and a few 
 of the recent ones, three shoots are found. 
 
 For the pile, both the Sehna and the Ghiordes knot are largely used, 
 whereas the former is unknown in modern Turkish and Caucasian carpets. 
 The texture, on the average, tends to be fine, for while the coarser carpets 
 are about the same as those of Turkey and Caucasia the finer ones are 
 two, three or several times more closely knotted than any from those 
 countries. With certain exceptions the ends of Persian carpets are 
 finished in a very simple way, there being a short web either terminating 
 at the loops of the warp, or furnished with a plain fringe. The sides 
 nearly always have a selvedge of one cord only, overcast with fine wool 
 or silk. 
 
 There is a tendency for Persian carpets to fall into two groups 
 according to the general nature of their technique, and these groups 
 may be said quite roughly to have separate geographical ranges. Towards 
 the north-west there is mostly found a woollen warp on one level, associated 
 with a coarse texture, a long pile and the Ghiordes knot. Towards the 
 south-east the warp is usually of cotton arranged on two levels ; the 
 texture is fine, the pile is short and the Sehna knot is used. There are 
 exceptions to this rule, but the general tendency cannot be overlooked. 
 
 Persian carpets have a soft, rich and deep colouring rather than a 
 brilliant and decided one. The softness is to a great extent due to the 
 use of several shades of each colour and to the fact that the various colours 
 rarely occur in large masses. 
 
 In point of design, Persian carpets are far more naturalistic than 
 any other kind, though even with them the naturalism is not really carried 
 far. Nearly all the ornament is based on floral forms and, if a 
 few unimportant border-stripes be excepted, purely geometrical patterns 
 
 173 
 
HAND-WOVEN CARPETS 
 
 are very rare. The general character of Persian design can be best 
 understood by the study of such examples as are illustrated in the 
 second volume of this work ; but a few features often met with may 
 be briefly mentioned. 
 
 Firsts attention may be drawn to the prevalence of central medallions 
 of all sorts often associated with corner-pieces which soften the 
 rectangularity of the field. Secondly^ the fields or that part of them not 
 occupied with medallion or corner-pieces, is very often covered with 
 comparatively small repeating patterns, such as diapers or lattices ; or 
 with an intricate arrangement of scrolling stems interspersed with leaves 
 and blossoms. Thirdly, a great majority of the borders are filled with 
 wavy floral stems, rather than a succession of detached devices which is 
 so common in Turkish and Caucasian carpets. 
 
 Devices and patterns peculiar to Persian rugs are illustrated on Plate 
 
 202 . 
 
 For the purpose of discussing in more detail the various recognised 
 types of Persian carpets, it is convenient to divide them into six large 
 groups, according to the provinces in which they are made. 
 
 (i) Azerbaijan, In Karadagh, a district bordering in the north 
 on Caucasia, carpets are produced which resemble closely in quality and 
 pattern the carpets of Karabagh. They have a woollen warp and weft 
 and a long, lustrous and fairly coarse pile, tied with the Ghiordes knot. 
 A peculiar and characteristic red colour is much used, and the ground 
 is often of undyed camel-hair. The field is mostly covered with a floral 
 diaper, or the Herati or Mina Khani pattern, and the border is more 
 geometrical than that of most Persian rugs. 
 
 Near the town of Tabriz, a type of carpet has been made lately 
 which is considered one of the best-knotted and most beautiful of Persian 
 rugs. The warp is of cotton on two levels, and the weft is also ot cotton 
 frequently dyed pink. The Ghiordes knot is used ; the pile is short 
 and harsh to the touch, and the texture is fine. There is mostly a two-cord 
 selvedge, and the rugs are stiff and tend to curl at the edges. The designs 
 are of the elaborate floral type, and are perhaps somewhat too formally 
 and accurately balanced, which is due no doubt to the strong European 
 influence under which they are manufactured. 
 
 In the district of Gorevan (or Yoraghan), not very far from Tabriz, 
 new carpets are made also under European influence. The warp and 
 weft are of cotton, which in the case of the latter is mostly dyed blue or 
 brown. They have the Ghiordes knot and are coarsely woven. The 
 
 174 
 
CARPETS IN THE MARKET : GROUPING— IDENTIFICATION 
 
 colour-scheme is unusual^ consisting of copper-red^ dark blue and buff, 
 distributed in rather conspicuous masses. The design nearly always 
 includes concentric hexagonal medallions ; and the drawing is peculiarly 
 bold and angular, 
 
 Herez, Serapi and Bakshi are names given to carpets which do not 
 differ in important particulars from Gorevans, Suj-bulak, a village in 
 the province, gives its name to carpets resembling those of Kurdistan, 
 
 (ii) Irak-Ajemi* One of the most famous types of Persian carpets 
 is made in this province and is called after the elevated plain of Feraghan, 
 These carpets have both warp and weft made of cotton, the weft sometimes 
 being dyed blue or pink. The warp is on one level. In the older pieces 
 the Sehna knot is used, but in later specimens the Ghiordes knot is also 
 common. The pile is fairly short and of good wool, A large number 
 of different colours generally occur in the rugs, and as usual red and 
 blue predominate, though there is also a notable amount of green. The 
 colours are not used in large masses unless for a plain ground, and generally 
 when viewed from a distance no part of the rug differs much in tone from 
 the rest. The field is mostly covered with a repeating pattern such as 
 the Herati, Guli Hinnai or some other floral diaper, the first-named 
 being the most common, A feature almost peculiar to the type is the 
 cutting off of the corners of the field by straight serrated lines. 
 Occasionally a pole-medallion on a light-coloured ground is seen, but 
 this is not very common. The border is narrow in proportion to the 
 size of the rug, and a favourite pattern in the main stripe, which often 
 has a green ground, is that sometimes known as the ** turtle ” (Plate 155), 
 Feraghans were largely made at one time, but new ones of good quality 
 are rather rare. They are very satisfactory for their wearing qualities. 
 Another type of rug similar in technique to the Feraghan is known 
 as the Saraband (from Sarawan, a mountainous district). The weft is 
 nearly always blue, and the knot is the Sehna, though sometimes a small 
 proportion of Ghiordes knots are mixed with them — a very unusual thing 
 in other types. It is in their designs that these rugs chiefly differ from 
 the Feraghans, They have neither central medallions nor corner-pieces, 
 but the field is covered with a repeating pattern of detached cone-devices. 
 The main border-stripe usually has a white ground, and contains a wavy 
 stem with a cone depending from each bend. They are often of the 
 Kanara shape, and are very strong and serviceable, 
 
 A well-marked group of carpets are called Hamadan, after the town 
 of that name on the site of the ancient Ecbatana, They have a white 
 
 175 
 
HAND-WOVEN CARPETS 
 
 cotton warp, and a weft of cotton, wool or camel-hair* The weft only 
 crosses once after each row of knots. The Ghiordes knot is used and 
 the pile is of medium length. The most striking thing about the colouring 
 is the large amount of natural camel-hair that is used. The other colours 
 are soft and harmonious. There is usually a pole-medallion in the field, 
 or more than one if the rug is a long one ; and the corners of the field 
 are frequently shaped. The ground is much lighter in tone than the 
 medallions and corner-pieces, and is sometimes covered with a delicate 
 diaper. The border-stripes have no characteristic patterns, but a 
 peculiarity connected with the border is that there is a wide, plain band 
 of camel-hair all round the edge of the rug. This band, which often 
 has small devices upon it arranged quite capriciously, is very characteristic 
 and enables an unmutilated specimen to be recognized at a glance. A 
 good many Kanara are found among Hamadans, but they are often 
 skilfully cut down so as to appear like rugs of the commoner type. 
 
 The town and district of Jushaghan has been noted for its carpets 
 for a long time (see p. 37). The warp and weft are of cotton, or sometimes 
 of wool, which in the case of the weft is mostly dyed red or brown. There 
 are sometimes three shoots of weft after the knots. The Ghiordes knot 
 is used and the pile is thick, soft and lustrous. The colouring is very 
 rich and mellow. The field does not have a medallion but is covered 
 all over with an irregular floral diaper on rather a large scale, and there 
 is often an impression of an incomplete lattice running through it. 
 The border has no characteristic patterns. 
 
 The city of Kashan gives its name to a type of rugs which are 
 considered to be the best that Persia produces nowadays. The warp 
 is of cotton and lies on two levels ; the weft is of fine blue cotton. The 
 Sehna knot is used and the pile, which is very closely knotted, is fairly 
 short but lustrous and very soft and velvety. The colours are deep, 
 rich and mellow ; and, unfortunately, to get this effect the modern rugs 
 are nearly all slightly washed, though not to an injurious extent. The 
 design is rather like that of Tabriz rugs but is less formal. There are 
 usually medallions in the field, and the whole of the rug including the 
 border is filled with gracefully drawn floral patterns tending towards 
 the naturalistic. It is not probable that many rugs of this group are 
 actually made in Kashan, but the type is a fairly definite one, though 
 on account of its high reputation it has become customary to call any good 
 rug a Kashan. 
 
 Most of the silk rugs made in Persia during the last half-century 
 
 176 
 
CARPETS IN THE MARKET : GROUPING— IDENTIFICATION 
 
 can be associated with the Kashan group. They are mostly very fine 
 in texture and rarely have fewer than 200 knots to the square inch. The 
 warp is either of silk or cotton^ and the weft usually of cotton. The 
 designs are sometimes based upon the early Persian traditions, but more 
 often follow, though not very closely, the modern woollen rugs of Persia. 
 In the East they are not usually placed upon the floor, for which position 
 they are indeed unsuited ; but are used for covering seats and for similar 
 purposes by the wealthier people. None of the later silk rugs appear 
 to go back farther than the middle of the nineteenth century, and there 
 is no evidence that such rugs were made in Persia during the century 
 or more preceding that date. 
 
 The name Saruk (not to be confused with Sarakhs) is given to rugs 
 similar to Kashans but not of quite such good quality. They sometimes 
 have a rendering of the early palmette pattern, which is known as the 
 Shah Abbas.'^ 
 
 In the neighbourhood of Muskabad are made rugs rather like the 
 Kashans and Saruks but considerably coarser. The warp lies on two 
 levels not widely separated, and the Ghiordes knot is sometimes used 
 instead of the Sehna. The colouring is quiet and harmonious, and the 
 designs are similar to those of the Feraghans. 
 
 The town of Sultan abad is a centre of the European- controlled 
 carpet industry, and as Muskabads are marketed there, they are often 
 called by its name. 
 
 Mahal and Savalan are other trade names given to some Muskabads. 
 
 (iii) Ardelan, In this province is the town of Sehna, which has 
 given its name to the knot largely used in Persia, and also to one of the 
 most esteemed types of Persian carpets. The Sehna carpet has a warp 
 and weft of white cotton, the former being on one level. The weft 
 generally crosses only once after each row of knots, but in some pieces 
 it crosses twice. The Sehna knot is used and the texture is as fine as 
 in any type of carpet, there being occasionally more than four hundred 
 knots to the square inch. The pile is very short and stands so nearly 
 upright that the surface feels rough and like a file in whichever direction 
 the hand is passed over it. The wool is good and lustrous, though the 
 lustre is not very apparent owing to the shortness of the pile. The 
 colours are numerous and rather bright ; but they are only used in small 
 masses, so that the general effect is soft rather than vigorous. The 
 designs resemble those of Feraghans and Sarabands. There are often 
 concentric pole-medallions and corner-pieces, and the rest of the ground 
 
 177 N 
 
HAND-WOVEN CARPETS 
 
 is either plain or covered with the Herati pattern or some other floral 
 diaper, or else filled with repeated cone-devices. The whole field 
 may be covered with such patterns to the exclusion of medallions and 
 corner-pieces. The border is rather narrow and has no patterns peculiar 
 to the type. The rugs are thin, yet rather stiff, and are not suited for 
 hard wear. The edges, as often in the case with tightly-woven rugs, 
 have a tendency to curl up. 
 
 From the town of Bijar come rugs very different from the Sehnas. 
 The warp is of thick wool, and lies on two levels. The weft is also of 
 wool, which is often dyed red. The pile is of heavy lustrous wool of 
 medium length, tied with the Ghiordes knot. The colours are 
 rich and bright — red, blue, white and natural camel-hair colour 
 predominating. The designs are very various, but generally combine 
 the features found in such rugs as Sehnas with those of the truly Kurdish 
 rugs made in the neighbourhood. A peculiarity often noticed is for 
 the pattern to be quite irregular and suddenly change in the middle or 
 at one end of the rug. Bijar rugs are sometimes called by the names 
 Sarakhs and Lule. 
 
 Throughout the province the Kurds make rugs which are given 
 specifically the name Kurdistan. They have a coarse woollen warp 
 lying on two levels only slightly separated, and a woollen weft. 
 The Ghiordes knot is used and the pile is very long and thick. The 
 rugs are so stoutly woven that frequently they can hardly be bent. The 
 ends are often treated very elaborately, and have coloured threads running 
 through or woven patterns on the webs. The loose ends of the warp 
 are looped and braided, or interwoven with each other in a great number 
 of intricate ways. The colours are rich and fairly numerous, with red, 
 blue, and white predominating. Quite a feature is the want of uniformity 
 in each colour, which often consists of several different shades. As 
 a consequence of this, the grounds are frequently streaked in a very 
 noticeable way. Patches of colour varying from blue to green are 
 very characteristic and very pleasing. The designs are various, but 
 lattice patterns are much favoured, and the floral diaper called the Mina 
 Khani is very common. The border is not conspicuous. Kurdistan rugs, 
 though coarse, are mostly soundly made and harmonious in appearance, 
 and are consequently a very useful kind. 
 
 The city of Mosul gives its name to a large number of rugs which 
 are marketed there. They are practically Kurdistans from the country 
 both West and East of Mosul, and they cannot be distinguished by any 
 
 178 
 
CARPETS IN THE MARKET : GROUPING— IDENTIFICATION 
 
 rigid criterion from the Kurdistan rugs as described above. In specimens 
 called by the trade Mosul, it will usually be found that the warp lies 
 on one level, and that the rugs are softer and more flexible than other 
 Kurdistans. The more definite patterns are not as a rule found, but 
 the design tends towards small diapers. 
 
 Another name, restricted to a group of Kurdistans made near Hamadan, 
 is Karaje. These are distinguished chiefly by the fact that the weft 
 only crosses once after each row of knots, though the one shoot may 
 consist at irregular intervals of two or more untwisted strands. 
 
 The city of Kermanshah in this province gives its name to a group 
 marketed there but made near Tabri2; or Sultanabad. They have the 
 general characteristics of the rugs of the districts where they are made, 
 but have a tendency to resemble in colour and design the rugs of Kirman. 
 
 (iv) Khorassan* Of the various kinds of carpets made in this province, 
 one kind is called specifically Khorassan. These have a cotton warp 
 on two levels ; and a cotton weft which is usually dyed blue. The 
 Sehna knot is used and the pile is long, lustrous and often unevenly cut. 
 A peculiarity of the knotting that is frequently seen is for the knots to 
 be tied on four warp threads instead of two, at intervals throughout 
 the rug. This causes a line across the back, which appears at first sight 
 to be due to a difference in the manner of inserting the weft, rather than, 
 as it actually is, to a change in the knotting. The colours are 
 numerous and bright. Rose-red, dark blue, white and purple are mostly 
 included, and the general effect is rather purple in character. A deep 
 vermilion and a red-purple which fades easily are often found and are 
 rather characteristic. In the design a very common motive peculiar 
 to the type is a compound cone-device consisting of two or more small 
 cones springing from a larger one (see Plate 202 (13) ) ; these devices 
 are found in both field and border, detached, or forming a diaper. Central 
 medallions and corner-pieces are common ; or the Herati pattern is 
 used. Sometimes the field is divided into a number of narrow vertical 
 bands, each resembling a border. A common border-pattern is shown 
 on Plate 202 (2). Khorassan rugs are usually large in si^e, and one 
 peculiarity is that there are often short ends of thread hanging loose 
 at the back. 
 
 A group of Khorassan carpets is called Meshed, after the chief 
 town of the province. They are generally finer and more firmly woven, 
 with a shorter and more even pile. The colouring tends to be more 
 delicate, with less of the purple character. The designs are mostly 
 
 179 
 
HAND-WOVEN CARPETS 
 
 of the central medallion type, and do not include the peculiar Khorassan 
 border, and not so often the compound cone-device, Mesheds and 
 Khorassans, however, may be regarded as the two halves of a graduated 
 series of rugs, so that some specimens cannot be certainly classified. 
 
 A quite different type of rug, and one much more closely resembling 
 the rugs of North-West Persia, is called after the town of Herat. The 
 warp and weft may be either of wool or cotton, and the warp lies on two 
 levels, though they are not always very widely separated. The Ghiordes 
 knot is mostly used and the pile is soft and of medium length. The 
 chief colours are dark blue, red, white, green and yellow ; but there is 
 mostly a faint tinge of purple in the colouring. The field is covered 
 all over with a floral diaper, mostly the Herati (which of course gets 
 its name from the same city), but sometimes the Mina Khani or some 
 other pattern. The border is mostly conspicuous ; a favourite pattern 
 being that shown on Plate 202 (i). 
 
 (v) Kir man. The carpets made in this province have been practically 
 free from Western influence, and though considerable variation of pattern 
 is seen in them, they are all called Kirmans. The warp is of cotton 
 and lies on two levels ; the weft is of blue cotton or sometimes of fine 
 wool. The Sehna knot is used, and the pile is soft and silky, of medium 
 length and very closely woven. The colours are light in tone and soft 
 and delicate ; those most frequently occurring being white, grey, rose, 
 pale brown and pale blue. The design is generally more naturalistic 
 than in any other kind of carpet. Many sorts of flowers, especially roses, 
 are nearly always included in the design. Besides these, there often are 
 trees such as the cypress, vases, animals, birds and even human figures. 
 Inscriptions, stating the name of the maker, are perhaps more common 
 in these rugs than in any others. There may be a medallion or other 
 arrangement of the ornament, but it is the naturalistic detail that is so 
 characteristic. 
 
 In quite recent years it has become a custom to make rugs of an 
 extraordinarily fine texture, with as many as a thousand knots to the 
 square inch. Though, of course, these are very costly to produce, the 
 result unfortunately is not equivalent to the expenditure, for the knotting 
 is too fine to give either a pleasing or a durable fabric. The 
 designs too of these very fine rugs are rarely satisfactory. They depart 
 entirely from tradition and exhibit perhaps symbolical or narrative 
 subjects with lengthy inscriptions ; or there may be crowds of 
 historical personages, or a spiritless copy of a European picture. In 
 
 180 
 
CARPETS IN THE MARKET : GROUPING— IDENTIFICATION 
 
 fact^ the designs are not suitable for carpets^ and the subjects would 
 be much better rendered in some other medium. The originality of 
 the Kirman weaver is shown in other ways too^ among which may be 
 mentioned the production of carpets with a plain-woven, unknotted 
 ground, and a pattern knotted in the usual way. In some of these rugs 
 a gold-coloured or grey weft is used, probably with the idea of giving 
 the effect of the gold and silver backgrounds of some of the antique Persian 
 rugs. 
 
 (vi) Farsistan or KashkaL All the carpets made in this province 
 have marked characteristics of technique which make them easy to recognise, 
 though there is considerable variety of design. Different types are 
 described by writers, but the difference between them is so slight that 
 they cannot always be distinguished. The most important type is known 
 as Shiraz, after the town of that name. These have a white woollen 
 warp on one level, and a red woollen weft that crosses twice after each 
 row of knots. The Sehna knot is the most common, but the Ghiordes 
 is also used. The pile is fairly long and is very soft and lustrous, and 
 the wool gives, perhaps, richer and more transparent colours than any 
 other grown in Persia. There is a single- cord selvedge, which is overcast 
 in short lengths of different colours, mostly red and blue, and there are 
 often tassels depending from the selvedge at intervals. There are broad 
 webs at each end woven in horizontal coloured stripes, again mostly red 
 and blue ; and these webs are often embroidered slightly with coloured 
 wool. The colours are deep and rich. Blue and red predominate, 
 but there are usually small areas of white which stand out very 
 conspicuously. Another favourite colour is a peculiar grey-green, which 
 is not often seen in other rugs. The design is more geometrical than 
 that of most other Persian carpets, and as free use is made of latch-hooks 
 and small conventional animals and birds, Shiraz rugs are often mistaken 
 for Caucasian products. The most common field-design is a long 
 hexagonal panel, or a succession of hexagonal panels linked together. 
 The field is often shaped in various forms outlined with straight lines. 
 The ground outside the panel is often filled with parallel bands, each 
 of which has a small pattern. The whole field is frequently covered 
 with a succession of small angular cones, which may have different 
 colourings but are often arranged so that those in diagonal lines are coloured 
 alike. When none of these schemes are used, the field is usually filled 
 with a variety of detached geometrical or conventional floral devices, 
 among which will mostly be found small birds and animals. The border 
 
 i8i 
 
HAND-WOVEN CARPETS 
 
 often consists of a large number of stripes^ some of which are filled with 
 the common Persian motive of blossoms depending from a wavy stem, 
 Another favourite border-stripe is the barber's-pole, the successive 
 bands of which usually show all the colours of the rug. Shiran rugs 
 are, as a rule, very pleasing and desirable pieces, but they nearly always 
 have the slight disadvantage of not lying quite flat on the floor. A good 
 many saddle-bags come from Shira:?, and some of them are exquisitely 
 woven. In the Victoria and Albert Museum there is a small saddle-bag 
 with the cone pattern on a yellow ground, so finely knotted that there 
 are over six hundred knots to the square inch. 
 
 A very similar group of rugs coming from the east of Shira2; is called 
 Niris (the name of a salt-lake in the district) or alternatively Laristan. 
 They have the Ghiordes knot and a warp on two levels more or less distinct. 
 They favour the all-over cone pattern rather than the central panel, and 
 the cones are often elaborately drawn and large rather than small and 
 angular. Sometimes the field has a small lattice pattern, or a diaper 
 of small devices arranged so as to have a pronounced striped effect. 
 
 Shiraz and Niris rugs are frequently called Mecca, for the reason, 
 so it is said, that large numbers of them are taken to and left at Mecca 
 by pilgrims. 
 
 Illustrations of the principal types of Persian rugs will be found 
 on Plates 147 to 175. 
 
 Plates 147 and 148. Karadagh Carpets. 
 
 Carpets of this kind are known in the trade also as Feraghan, Kara- 
 bagh, ShiraZt and simply as Persian. They have from 50 to 100 knots 
 to the square inch. The usual sizes are 2 ^ ft. X 5 ft., 4J ft. X 6|- ft. 
 
 and 5 ft. X 10 ft. They are pliable and soft to the touch, and have a 
 
 short and not very lustrous pile. The chief colours are wine-red and 
 black. 
 
 Plate 149. Karadagh Carpet. 
 
 Carpets of this kind are known in the trade also as Feraghan, Karabagh, 
 ShiraZ/ or simply as Persian. They have from 50 to 100 knots to the 
 square inch. The usual sizes are 2I ft. X 5 ft., 4 ft. X 6| ft. and 5 ft. X 
 
 10 ft. They are pliable and soft to the touch, and have a short but 
 
 lustrous pile. The chief colours are blue, red and white. 
 
 Plate 150. Karadagh Carpet. 
 
 Carpets of this kind are known in the trade also as Feraghan, Karabagh, 
 Shiraz, and simply as Persian. They have from 80 to 100 knots to the 
 
 182 
 
CARPETS IN THE MARKET : GROUPING— IDENTIFICATION 
 
 square inch. The usual sizes are ft. X 5 ft.;^ 4J ft. X 6^ ft. and 5 ft. 
 X 10 ft. They are pliable and soft to the touchy and have a short 
 and not very lustrous pile. The chief colours are blue, red and green. 
 
 Plate 151. Tabriz Carpet. 
 
 Carpets of this kind are known in the trade also simply as Persian. 
 They have from 200 to 260 knots to the square inch. The usual sizes 
 are 8 ft. X ii| ft., 10 ft. X 13 ft. and ft. x 15 ft. The chief colours 
 are red, blue and green, and the general scheme of colour is soft and 
 harmonious. They are among the most perfectly made of modern 
 carpets. The complicated mass of floral scrolls is typical. 
 
 Plate 152. Gorevan or Yoraghan Carpet. 
 
 Carpets of this kind are also known in the trade simply as Persian. 
 They have from 30 to 40 knots to the square inch. The usual size is 
 about io| ft. X 15 ft. They are very harsh to the touch and have a 
 thick, long and dull-looking pile. The chief colours are chocolate-red, 
 blue and white. The characteristic angular drawing of the detail, as 
 well as the heavy massing of the colour, are well shown in the illustration. 
 
 Plate 153. Hahadan Carpet. 
 
 Carpets of this kind are known in the trade also as Camel-hair, or 
 simply as Persian carpets. They have from 30 to 40 knots to the square 
 inch. The usual size is 4I ft. x 15 ft. They are harsh to the touch and 
 have a very thick and long pile. The chief colours are the brown of 
 camel-hair, red and blue. 
 
 Plate 154. Hamadan Carpet. 
 
 Carpets of this kind are also known in the trade as Kirman, or simply 
 as Persian. They have from 30 to 40 knots to the square inch. The 
 usual sizes are 4 ft. X iij ft. and 6| ft. x i6| ft. They are somewhat 
 harsh to the touch and have a fairly long and lustrous pile. The chief 
 colours are the brown of camel-hair, red and blue. The ground of 
 this specimen has a more elaborate pattern than usual, and there is 
 little doubt that the rug had once an outer edging of camel-hair. 
 
 Plate 155. Feraghan Carpet. 
 
 Carpets of this kind are known in the trade also simply as Persian. 
 They have from 65 to 130 knots to the square inch. The usual sizes 
 are 8 ft. X ii|- ft., 10 ft. X 13 ft. and ii| ft. X 15 ft. Antique pieces 
 are generally about 4 ft. x 10 ft. The more recent pieces have a fairly 
 
 183 
 
HAND-WOVEN CARPETS 
 
 shorty but the older ones a long pile. The illustration shows a typical 
 Feraghan with the usual Herati pattern. 
 
 Plates 156 and 157. Saraband Carpets. 
 
 Carpets of this kind are known in the trade also as Feraghan, Shiraz, 
 Mecca, or simply as Persian. They have from 100 to 160 knots to 
 the square inch. The usual sizes are 2^ ft. X 4 ft., 4 ft. X 8| ft. and 
 5^ ft. X 6| ft. They have a fairly short and rather lustrous pile. The 
 chief colours are dark blue, red and green. The first specimen illustrated 
 has a pattern much like a Feraghan with the cone-diaper substituted 
 for the Herati or similar pattern. Neither shows the most typical Saraband 
 border-stripe of cone-devices depending from every bend of a wavy 
 stem. 
 
 Plate 158. Jushaghan Carpet. 
 
 Carpets of this kind are also known in the trade as Ispahan, Shah- 
 Abbas, or simply as Persian. They have from 100 to 160 knots to the 
 square inch. The usual size is about 6 1 ft. x 13 ft. They have a soft 
 and lustrous pile of medium length. The chief colours are dark blue, 
 red and yellow. The pattern, which is typical of Jushaghans, is obviously 
 descended from the well-known design of palmettes, blossoms and cloud- 
 bands which is seen so often in Persian rugs of the sixteenth and seven- 
 teenth centuries. 
 
 Plate 159. Saruk or Kashan Carpet. 
 
 Carpets of this kind are also known in the trade simply as Persian. 
 They have from 160 to 225 knots to the square inch. The usual sizes 
 are 6 ft. x 10 ft., 10 ft. X 13 ft. and ft. x 18 ft. They have a 
 short, close and very lustrous pile. The chief colours are copper-red, 
 blue and green. 
 
 Plate 160. Persian Silk Carpet (Victoria and Albert Museum). 
 
 6 ft. 5 in. X 4 ft. 5 in. 
 
 Warp : White silk. On two levels. 29 to i inch. 
 
 Weft : Blue cotton. Two shoots after each row of knots. 
 
 Knots : Silk. Sehna. 21 to i inch. 300 to the square inch. 
 Colours : Eight. White (field) : light green (border) : purple : 
 black : blue : yellow : red : magenta. 
 
 Plates 161 and 162. Muskabad Carpets. 
 
 Carpets of this kind are also known in the trade simply as Persian. 
 They have from 40 to 45 knots to the square inch. The usual sizes are 
 
 184 
 
CARPETS IN THE MARKET : GROUPING— IDENTIFICATION 
 
 lo ft. X 13 ft. and iij ft. x 15 ft. They have a rather lustrous pile 
 of medium length. The chief colours are dark blue, red and white. 
 
 Plates 163, 164A, b, 165, 167B. Sehna Carpets. 
 
 Carpets of this kind are also known in the trade as Kirman, or simply 
 as Persian. They have about 200 knots to the square inch. The size 
 is nearly always about 5I ft. X 8 ft. They are very thin and have a 
 close pile, rather rough to the touch. The chief colours are blue, red and 
 white. There are no better carpets than these in point of technique and 
 delicacy of drawing, and their price is high, but not when allowance is 
 made for the very close knotting. 
 
 Plate 166. Kurdistan Carpet. 
 
 Carpets of this kind are known in the trade also as Iran, Mosul, 
 Hamadan, and simply as Persian. They have from 40 to 65 knots to 
 the square inch. The usual sizes are 3I ft. x 8 ft. and 5^ ft. x 13 ft. 
 The chief colours are red, blue and white. They are very thick and 
 have a coarse pile, which is lustrous in the older specimens but less so in 
 the modern ones. 
 
 Plates 167A and 168. Khorassan Carpets. 
 
 Carpets of this kind are known in the trade also simply as Persian. 
 They have from 90 to no knots to the square inch. The usual sizes are 
 10 ft. X 13 ft., II ft. X 18 ft. and 14 ft. X 20 ft. Smaller carpets are 
 rare. They are soft to the touch, and have a lustrous pile of medium 
 length. The chief colours are rose-red, blue, white and green. 
 
 Plates i6g, 170, 171, 172. Kirman Carpets. 
 
 Carpets of this kind are also known in the trade as Tabriz, or simply 
 as Persian. They have from 200 to 260 knots to the square inch. The 
 usual sizes are 8 ft. x ii| ft., 10 ft. x 13 ft., ii^ft. x 15 ft., ii ft. x 21 ft. 
 and even larger. They are thin and have a short and not very lustrous 
 pile. The chief colours are red, blue-green and white. 
 
 Plates 173, 174, 175B. Shiraz Carpets. 
 
 Carpets of this kind are also known in the trade as Mecca, or simply 
 as Persian. They have from 45 to 100 knots to the square inch. The 
 usual sizes are 4^ ft. X 6| ft. and 5^ ft. X 11^ ft. They have a rather 
 loose texture and are soft to the touch. The pile is very lustrous and 
 of medium length. The chief colours are dark blue, red and white. No. 
 174 has the triple medallion and shaped field that is so common in Shirazis, 
 
 185 
 
HAND-WOVEN CARPETS 
 
 and No, 175B has the other characteristic pattern of small angularly drawn 
 cones. 
 
 Plate 176. Persian Kilim (Victoria and Albert Museum). 6 ft. x 
 4 ft. 6 in. 
 
 Warp : Three-ply white wool. 17 to i inch. 
 
 Weft : Two-ply wool. About 96 shoots to i inch. 
 
 Colours : Eight. Dark blue (field) : white (central medallion) : 
 red : light blue (ist and 3rd borders) : light red : green : 
 yellow (2nd border) : brown. 
 
 (d) Central Asiatic 
 
 The vast and wild region which lies between Persia and China is 
 sparsely inhabited by fierce and uncultured nomadic tribes, and yet it 
 produces carpets which are beautifully made and of great artistic 
 interest. A great part of this region falls within the Russian political 
 sphere, and as a consequence most of the first rugs to leave were exported 
 by way of Russia. The carpets made in the eastern part of the region 
 under consideration show strong Chinese influence, but all the others 
 have a resemblance to each other that is very remarkable when it is 
 considered how many different and antagonistic tribes are engaged in 
 their production and what great distances separate the boundaries of 
 the region. 
 
 The carpets of Western Turkestan are made almost entirely of wool 
 which is of very good quality. Both warp and weft are commonly white, 
 or at least not dyed, and the weft crosses twice after each row of knots. 
 The Sehna knot is generally used, though there are a few exceptions. 
 The knotting is never very coarse and often is remarkably fine, and the 
 counting of the knots in making the pattern is done at least as carefully 
 as anywhere else. It is true that the style of pattern adopted almost 
 necessitates careful weaving, for any serious miscalculation would be 
 very apparent; but even then it is remarkable that nomadic people on 
 the most primitive looms should produce such very accurate work. 
 
 Another peculiarity of the texture is that the knots are considerably 
 smaller in the direction of the warp than in the direction of the weft. 
 In fact the small rectangle occupied by each knot is about twice as wide 
 as it is high. 
 
 The most characteristic feature of these rugs is their colouring, 
 which is quite unlike that of any other class. The colours are almost 
 
 186 
 
CARPETS IN THE MARKET : GROUPING— IDENTIFICATION 
 
 invariably deep and sombre, and when bright colours are used they 
 occur only in very minute quantities. Red is by far the commonest 
 colour, but it is never brighter than a deep brick tint, and is more often 
 a deep wine-red or a brown- or purple-red. Dark blue is the next most 
 common colour, and there is a fair amount of white, which is often rather 
 conspicuous. Black, green, blue and brown are sometimes used in 
 smaller quantities, while very little of other colours is seen. 
 
 The designs are on the whole more geometrical even than those 
 of Caucasian rugs. Floral devices do occur, but, with a few exceptions, 
 they are so highly conventionali2;ed as to be almost unrecognizable. 
 Two types of design are so widely spread through the different groups 
 of rugs that they must be mentioned here. In one type the field is covered 
 with a repeating pattern of octagons, regularly arranged in rows and 
 columns ; the octagons are considerably flattened, on account of the 
 unusual proportions of each knot in the texture. In the other type the 
 field is divided by two crossing bands into four panels approximately 
 equal and square. These latter pieces are really portieres. 
 
 There are many borders and smaller details of the designs which 
 are exclusively found in Central Asiatic carpets, and several of these 
 will be seen illustrated on Plates 200 and 201, 
 
 Central Asiatic carpets are very rarely of large dimensions, and 
 only a small proportion of them are rugs. The majority of pieces are 
 saddle-bags, tent-bags, mats or tent-bands, and even some of the larger 
 pieces, which look like rugs, were really intended for hanging across 
 the doorway of a tent. As all carpets of this class were meant for use 
 rather than show, and, what is more, for the use of nomads, it is not 
 remarkable that very old pieces are quite unknown, A certain number 
 may be found going back to the eighteenth century, but it is very doubtful 
 whether any pieces of an earlier date still exist. Carpets of a 
 moderate age are invariably sound in colour and weaving, and even 
 modern ones are nearly always well woven, though they have suffered 
 in common with most new carpets from unsound dyes. Almost without 
 exception they wear well, and if the dyes are sound they are very desirable 
 carpets. 
 
 Most of the rugs made in Western Turkestan are called Bokhara., 
 though very few are made there. It is best as far as practicable to classify 
 the rugs according to the tribe that made them, though there are great 
 difficulties in the way of a strict nomenclature and much confusion on 
 the subject. The rugs made by the Salor tribes (known frequently 
 
 187 
 
HAND-WOVEN CARPETS 
 
 as Royal Bokhara) have a pattern of repeated octagons of the kind 
 illustrated on Plate 175A. In the interspaces between the octagons there 
 are various figures which are roughly diamond-shaped in contour* Salor 
 rugs are not woven now and have become very rare^ but their patterns 
 have been adopted by other tribes. 
 
 The majority of Turcoman rugs are made by the Tekke tribes whose 
 pieces are generally coarser than those already mentioned. The warp^ 
 which is of medium wool or goat-hair^ lies on two levels slightly separ- 
 ated. There is mostly a three-cord selvedge and the ends have a long, 
 coloured web and a long fringe. The border shown on Plate 201 (27) 
 is almost invariably found. The chief colour is red, which ranges from 
 a brick-red to a purple-red. 
 
 The rugs of the Pinde tribes are darker in tone than the last. The 
 ground is of a deep brown-red or purple-red, sometimes approaching a 
 purple-black, and the pattern stands out rather strongly in white lines. 
 
 All Katchli pattern rugs are often erroneously called prayer-rugs. 
 
 In the neighbourhood of Khiva rugs are made which much resemble 
 those called Bokhara, but tend to be coarser and generally more barbaric 
 in appearance. Animals and geometrical forms rather like the Caucasian 
 are introduced. 
 
 The rugs woven by the Yomud tribes form a group more easy to 
 recognize than many of the others. They have a rather coarse warp 
 which tends to lie on two levels. The Ghiordes knot is commoner 
 than the Sehna, whereas the latter is almost invariably found in other 
 Central Asiatic rugs. They have not the flattened octagons, but mostly 
 have patterns constructed with diamond-shaped latch-hooked figures 
 often of two types or at least coloured in two ways. Some of their border- 
 stripes are characteristic (Plate 201 (i, 7, 13, 18, 22, 23) ). The Yomud 
 tent-bags are not as a rule rectangular, but are pointed at the top and 
 are mostly furnished with long fringes and cords at the corners. The 
 colour-scheme mostly includes a purple-brown and a rich red as well 
 as a fair amount of rather conspicuous white. A great peculiarity of 
 knotting is seen in some Yomud carpets, but perhaps in no other kind. 
 Its nature and the reason for it will require a little explanation. The 
 simplest diagonal contour that can occur in a knotted pattern is formed 
 by carrying one of the two colours concerned one knot farther in each 
 successive row, and if the knots are approximately square this makes 
 an angle of about 45 degrees. In Central Asiatic carpets the knots are 
 much wider than they are high, so a diagonal line formed as above will 
 
 188 
 
CARPETS IN THE MARKET : GROUPING— IDENTIFICATION 
 
 be much less steep. If an angle of 45 degrees is wanted it can be obtained 
 by carrying one colour one knot further in every second row, but the line 
 will not be a smooth one. The Yomuds get out of the difficulty by 
 breaking up the pairs of warps on which most of the knots are tied 
 and by tying in the new row midway between those of the previous row. 
 The knots are thus arranged like the bricks in a half- brick wall, and a 
 fairly smooth diagonal line of about 45 degrees will result. This is only 
 done where necessary in the pattern, and the usual arrangement is restored 
 by tying a knot on three warps at each end of a row of abnormally-placed 
 knots. 
 
 Rugs called Afghan are made by wandering tribes partly in Afghan- 
 istan but also outside its boundaries. They are coarsely-woven and 
 have broad webs at each end divided into several coloured bands, and 
 usually long fringes. The pattern mostly consists of large repeated 
 octagons which almost touch each other in the vertical direction, and 
 these octagons mostly contain trifoliate forms (Plate 201 (20) ). The 
 colours are chiefly dark red and brown, and the rugs have a reddish-tawny 
 appearance. 
 
 The rugs called Beshire, from a district to the west of Khiva, 
 have a coarse or medium texture, and usually have wide striped webs 
 at the ends, and fringes. The colours are mostly red and brown with 
 a fair amount of blue. The patterns are more floral than those of other 
 Central Asiatic rugs, and often are reminiscent of Persian designs. Trellis 
 patterns are not uncommon, and sometimes a curious pattern consisting 
 of repeated scrolls having a vermiculated appearance are seen (Plate 
 189). The prayer-rugs have a simple pointed arch occupying nearly 
 the whole of the rug (Plate 190). Beshire rugs have only been exported 
 in fairly recent times. 
 
 A well-marked type of rugs akin to the Turcoman is called 
 Beluchistan or Beluchi, and some may actually be made there by 
 wandering tribes ; but probably the majority are made by similar tribes 
 further to the north. The rugs are very dark and sombre, the prevalent 
 colours being dark blue, dark red and black ; invariably, however, there 
 are small areas of white which are very conspicuous, and sometimes 
 there is a good deal of natural camel-hair. On account of the blue, 
 which is unusual in Central Asiatic rugs, they are sometimes called Blue 
 Bokhara. The designs are different from the Turcoman rugs. They 
 are to some extent geometrical, but they also have conventional tree-forms 
 of a peculiar kind, and also copy some of the simpler floral patterns of 
 
 189 
 
HAND-WOVEN CARPETS 
 
 Persia. The prayer-rugs of this group are fairly numerous and have 
 a simple rectangular arch. The rugs are thick and well woven, and 
 the pile is rather long and mostly very lustrous. They wear well, and 
 in spite of their dark colour are very attractive. 
 
 Quite different from the Central Asiatic rugs already described are 
 those from Eastern or Chinese Turkestan. They have usually a cotton 
 warp on two levels and a cotton weft. The Sehna knot is used, and the 
 texture is coarse and rather loose with from about 40 to 65 knots to the 
 square inch. These rugs usually have a rich yellow and brown appearance, 
 and are so closely connected with the Chinese carpets in point of design 
 that they are most conveniently classed with them A further description 
 will be found in Part I, Chapter VI (6). 
 
 Plates 175A, and 177 to 193 illustrate the principal types of Central 
 Asiatic rugs apart from those of Eastern Turkestan. 
 
 Plate 175A. Turcoman Carpet. 
 
 Carpets of this kind are also known in the trade as Bokhara. They 
 have from 130 to 200 knots to the square inch. The usual skes are 
 5I ft. X 10 ft., 5 ft. X 8 ft., as well as i X 2 ft., and 2X4 ft., in 
 which case they are sometimes known as Yastik. They are closely knotted 
 and have a short and lustrous pile. The chief colours are red-brown 
 with smaller quantities of dark blue and white. 
 
 Plates 177. Turcoman Strips or Tent-Bands. 
 
 Carpets of this kind are also known in the trade as Bokhara or Khiva. 
 These border-like pieces are used as decorations, chiefly round the walls 
 of tents where the sides join the roof. The ground is of plain weaving 
 and the pattern is knotted in the usual way. The usual si2;e is 6 ins. X 
 10 ft., though some are much larger. The ground is white, or the natural 
 colour of the wool, and the pattern is chiefly in dark purple, blue and 
 white. 
 
 Plates 178 and 179. Turcoman Carpets. 
 
 Carpets of this kind are known in the trade also as Bokhara. They 
 have from 130 to 200 knots to the square inch. The usual size is 
 2i ft. X 5 ft. They are closely knotted and have a short and lustrous 
 pile. The chief colours are red-brown, dark blue, and white. 
 
 Plates 180, 181, and 182. Turcoman Carpets. 
 
 Carpets of this kind are known in the trade also as Bokhara, Khiva, 
 Tekke, and Kisiliyak. They have from 105 to 106 knots to the square 
 
 190 
 
CARPETS IN THE MARKET : GROUPING— IDENTIFICATION 
 
 inch. The usual size is ft. X ft. They are closely knotted and 
 have a short and lustrous pile. The chief colours are dark red-brown, 
 dark blue, white and yellow ; the white sometimes being of cotton. 
 
 Plates 183, 184, 185. Turcoman Yomud Carpets. 
 
 Carpets of this kind are also known in the trade as Bokhara. They 
 have from 130 to 200 knots to the square inch. The usual si2;e is 5 ft. X 
 10 ft. They are closely knotted and have a short but frequently very 
 lustrous pile. The chief colours are dark purple, blue and white. 
 
 Plates i86a, and 187. Afghan Carpets. 
 
 Carpets of this kind are known in the trade also as Bokhara. They . 
 have from 45 to 70 knots to the square inch. The usual sizes are 
 4 k ft* X 7 ft. and 6 ft. X 9 ft. They usually have a long pile which is 
 very lustrous in the better pieces. The chief colours are dark red- 
 brown, dark blue, white and yellow. 
 
 Plate i86b. Turcoman Camel Bag (Victoria and Albert Museum). 
 
 2 ft. X 4 ft. 4 ins. 
 
 Warp : White wool. On one level. 19 to i inch. 
 
 Weft : Red wool. Two shoots after each row of knots. 
 
 Knots : Wool. Sehna. 20 to i inch. 200 to the square inch. 
 Colours : Seven. Purple-red (field, 3rd border and edging) : red 
 (2nd and 4th borders) : white : dark blue : green-blue : brown : 
 yellow. 
 
 Plates 188 and 189. Beshire Carpets. 
 
 Carpets of this kind have from 50 to 100 knots to the square inch. 
 The usual sizes are 5 ft. X ft. and 6| ft. x i6| ft. The pile is 
 very long and very lustrous. The chief colours are red-brown and dark 
 blue, with a little white and yellow. These carpets have only been exported 
 for twenty years or so, and are still not well known. 
 
 Plate 190. Beshire Prayer-carpet. 
 
 Carpets of this kind have from 50 to 100 knots to the square inch. 
 The size is about 3|- ft. X 6 ft. They have a long and lustrous pile. 
 The chief colours are red, brown, white and dark blue. 
 
 Plates 191, 192 a, b, 193. Beluchistan Carpets. 
 
 Carpets of this kind are also known in the trade as Afghan and Bokhara. 
 They have from 50 to 65 knots to the square inch. The usual size is 
 about 5 ft. X 8 ft. ; large pieces being very rare. They have a long and 
 
 191 
 
HAND-WOVEN CARPETS 
 
 very lustrous pile. The chief colours are dark blue^ brown, and an 
 intense white, 
 
 Plate 192c, Beluchistan Prayer- carpet, 5 ft, 3 ins, x 2 ft, 7 ins. 
 Warp : White wool. On one level, 15 to i inch. 
 
 Weft : Black wool. Two shoots after each row of knots. 
 
 Knots : Wool, Sehna, 1 1 to i inch, 80 to the square inch. 
 Colours : Six, Camel-hair (field) : crimson-red : black (border) : 
 red : dark blue : white. 
 
 192 
 
BIBLIOGRAPHY 
 
 Bode, Wilhelm von. — ^Vorderasiastische Knupfteppiche aus alterer Zeit. 92 
 {i chromo) 'process Ulus. (10 X 7.) Leipzig : H. Seemann Nachfolger. 1901. 
 
 Bogoloubov, Andreya Andreevich. — Tapis de TAsie centrale, faisant partie de la 
 collection reunie par A.B. (Text also in Russian.) 52 pp. 36 chromo-lithogr., 
 
 7 phototype plates and 2 maps. (28 X 20.) St. Petersbourg. 1908. 
 
 DUMONTHIER, Ernest.- — Recueil de dessins de tapis et de tapisseries d ’ameublement 
 du mobilier de la Couronne. 8 pp. 48 {some col.) phototype plates. (25 X 18.) 
 Paris : Librairie generale de 1 ’ Architecture et des Arts decoratifs. 1912. 
 
 GrOTE-Hasenbalg, Werner. — Der Orientteppich, seine Geschichte und seine Kultur. 
 Process Ulus, incl, 137 chromo, and map. 3 vols. (10 X 10.) Berlin : Scara- 
 baeus-Verlag. 1922. 
 
 Hackmack, Adolf. Der Chinesische Teppich. 44 pp. Process Ulus. incl. 25 
 (i chromo) plates, and map. (9 X 6.) L. Friederichsen and Co. Hamburg. 
 1921. 
 
 Hawley, Walter A. — Oriental rugs, antique and modern. Process Ulus., some 
 chromo and maps, (ii X 8.) New York : John Lane Co. 1913. 
 
 Hendley, Thomas Holbein, Col. — Asian carpets : i6th and 17th century designs 
 
 from the Jaipur palaces. 22 pp. 146 (138 chromo) lithogr., incl. frontispiece 
 and 1 phototype. (20 X 14.) London : W. Griggs. 1905. 
 
 Holt, Rosa Belle. — Rugs, Oriental and Occidental, antique and modern. A hand- 
 book for ready reference. 31 (12 chromo) process Ulus, (ii X 8.) Chicago : 
 A. C. McClurg and Co. 1901. 
 
 Krygowski, T. — Polenteppiche (Polnische Knupfteppiche). 23 pp. ii process 
 Ulus. (In “ Orientalisches Archiv,” II, 70, 106.) 1911-12. 
 
 Lessing, J. — Orientalische Teppiche. 12 pp. 13 (12 chromo) plates and i text 
 Ulus. (19 X 13.) Berlin : E. Wasmuth. 1891. 
 
 Lessing, Julius. — Alt-Orientalische Teppichmuster, nach Bildern und Originalen 
 des XV-XVI Jahrhunderts. Herausgegeben mit Unterstiitzung des Kdnigl. 
 Preussischen Ministeriums fiir Handel, Gewerbe und offentliche Arbeiten. 30 
 chromo-lithogr. (20 X 14.) Berlin : E. Wasmuth. 1877. 
 
 English translation. Ancient Oriental carpets, etc. 28 pp. 30 chromo- 
 lithogr. (20 X 14.) London : H. Sotheran. 1879. 
 
 Lewis, G. Griffin. — The practical book of Oriental rugs. 2nd edition. Process 
 Ulus., some chromo and map (9 X 7.) Philadelphia : J. B. Lippincott Co. 
 
 1913- 
 
 193 
 
 o 
 
BIBLIOGRAPHY 
 
 Victoria and Albert Museum. Guide to the Collection of Carpets. 49 process 
 Ulus. (lo X 6.) London : H.M. Stationery Office. 1920. 
 
 Victoria and Albert Museum. Notes on Carpet Knotting and Weaving. 13 
 process Ulus. (8 X 5 >) London : H.M. Stationery Office. 1920. 
 
 Martin, F. R. — A history of Oriental carpets before 1800. Plates and illustrations 
 in the text, many chromo. (26 X 20.) Vienna : I. R. Court and State Printing 
 Office. 1908. 
 
 Mumford, John Kimberly. — Oriental rugs. 33 (i6 chromo) plates, 2 maps, and 
 illustrations in the text, (ii X 8.) London : Sampson Low, Marston, and Co. 
 1901. 
 
 Mumford, John Kimberly. — The Yerkes collection of Oriental carpets. 62 pp. 
 27 chromo process Ulus. (22 X 14.) London : B. T. Batsford. 1910. 
 
 NeugebauER, Rudolf, and Orendi, Julius. — Handbuch der orientalischen Teppich- 
 kunde. Mit einer Einfiihrung von Richard Graul. Process Ulus., many chromo. 
 (9 X 6.) Leipzig : K. W. Hiersemann. 1909. 
 
 INVENTAIRE G£n£ral des Richesses d’Art DE LA FRANCE. Vol. IV. Tapis en 
 Savonnerie. P. 173 foil. Paris. 1913. 
 
 Riegl, Alois. — Altorientalische Teppiche. 36 process Ulus. (9 X 6.) Leipzig : 
 T. 0 . Weigel Nachfolger. 1891. 
 
 Robinson, Vincent J.— Eastern Carpets : twelve early examples. With descriptive 
 notices by V. J. R., and a preface by Sir G. Birdwood. 44 pp. 12 chromo- 
 photo-lithogr. (20 X 15.) London : H. Sotheran. 1882. 
 
 Sarre, Friedrich, and MARTIN, F. R. — Die Ausstellung von Meisterwerken muham- 
 medanischer Kunst in Miinchen 1910. Vol. I. Miniaturen und Buchkunst : 
 die Teppiche. i chromo-photo-lithogr., and 46 phototype plates. (10 X 15.) 
 Miinchen : F. Bruckmann. 1912. 
 
 Stebbing, Edward. — The Holy Carpet of the Mosque at Ardebil. 18 pp. 7 (6 col.) 
 photo-lithogr. (25 X 21.) London : Robson. 1893. 
 
 Van DE Put, Albert. — Some 15th century Spanish carpets. 8 pp. Process Ulus, 
 inch I chromo. (In Burlington Magazine, XIX, 344 ; XX, 124.) 1911. 
 
 Vienna. — K. K. Oesterr. Handelsmuseum. Katalog der Ausstellung orientalischer 
 Teppiche im . . . Museum, 1891. (Preface by A. von Scala.) Cuts and 2 
 plans. (10 X 7.) Wien : Verlag des K. K. Oesterr. Handelsmuseum. 1891. 
 
 Vienna. — K. K. Oesterr. Handelsmuseum. Oriental carpets. Published by Order 
 of the Imperial and Royal Ministries of Commerce, Worship and Education. 
 English edition edited by C. Purdon Clarke. Chromo-photo-lithogr. and photo- 
 types. (26 X 20.) Vienna. 1892-96. 
 
 Vienna.— K. K. Oesterr. Museum fiir Kunst und Industrie. Ancient Oriental 
 carpets. . . . With preface by A. von Scala ; introduction by W. Bode ; 
 text by F. Sarre. 16 pp. 2^chromo-lithogr. (27 X 20.) Leipzig ; K. W. Hierse- 
 mann. 1906-08. 
 
 194 
 
INDEX 
 
 Abul Fazl on carpet-making in India, 37 
 
 Adam, Robert ; encouragement by, of carpet-making, 
 81 
 
 Afghan carpets, 189, 191 
 
 Akbar, Emperor of India : encouragement of carpet- 
 making, 37-40 
 
 Anatolian carpets, 1 51-153, 155, 158, 159, 160 
 Saph prayer-carpets, 158 
 Kilim prayer-carpets, 159 
 
 Ardabil carpet : description of, etc., 17-19, 30, 43, 104 
 
 Ardelan carpets, 177. See also the following names : 
 Bijar 
 Karaje 
 Kermanshah 
 Kurdistan 
 Lule 
 Mosul 
 Sarakhs 
 Sehna 
 
 Armenia : examples of Armenian carpets in Italy, 15 
 
 Arts, Society of : premiums offered by, for carpet- 
 making in 1756, 80 
 
 Axminster ; carpet-making in eighteenth century, 82 
 
 Azerbaijan carpets, 174-175. See also the following 
 names : 
 
 Bakshi 
 
 Gorevan (or Yoraghan) 
 
 Herez 
 
 Karadagh 
 
 Serapi 
 
 Tabriz 
 
 Bakshi carpets, 175 
 Baku carpets, 164, 169, 170 
 
 Bardini, Signor Stefano : Persian carpet owned by, 8, 
 27 
 
 Bell, Robert : Master of Girdlers’ Company, 40 
 Beluchistan or Beluchi carpets, 189, 191, 192 ; price 
 of, 143 
 
 Bergama rugs, 155, 159 
 Beshire carpets, 189, 191 
 Bijar carpets, 178 
 Lule, 178 
 Sarakhs, 178 
 
 Birdwood, Sir George ; on origin of pile-carpets, 6 
 Bode, Dr. von : Comments on Turkish carpet, 46 
 Bokhara, 187, 190 ; price of, 143, 144 
 Royal Bokhara, 188 
 Border designs, 125, 126-7 
 
 Brighton Pavilion : Axminster carpets supplied to, in 
 1823, 82 
 
 Buccleuch, Duke of : Persian and Indian carpets 
 
 owned by, 27, 32, 41 
 
 Buerkel, Dr. L. von : Spanish carpet owned by, 70] 
 
 Camel-hair carpets, 183 
 
 Carlisle, Earl of : Morris carpet made for, 83 
 
 Caucasus, The : 
 
 Baku rugs, 164, 169, 170 
 Daghestan carpets. See that name 
 [Designs of carpets, 162 
 
 Caucasus, The {contd.) : 
 
 Historical records of carpet-making, 58-60, 161 
 
 Kilim rugs. See that name 
 
 Kutais rugs, 165 
 
 Manufacturing methods, 161 
 
 Price of carpets, 143 
 
 Shirvan. See that name 
 
 Shusha carpets, 165 
 
 Soumak carpets, 165, 172 
 
 Tcherkess (or Circassian) rugs, 165 
 
 Tiflis rugs, 165 
 
 Central Asiatic carpets, 186-192 ; price of, 144. See 
 also the following names : 
 
 Afghan 
 
 Beluchistan or Beluchi 
 
 Beshire 
 
 Bokhara 
 
 Khiva 
 
 Kisiliyak 
 
 Pinde 
 
 Tekke 
 
 Turcoman Saddle Bag 
 Yomud 
 
 Chichi rugs, 164, 169 
 China : 
 
 Chinese design in Persia and Western Asia, 13 
 Croft-Lyons carpet, 66, 110 
 Cunliffe carpet, 109 
 Price of carpets, 143 
 Records of carpet-making, 62 
 Stockholm Museum carpet, 63 
 Chinese Turkestan : Carpet types, etc., 65-67, 190 
 Tientsin carpet, 66 
 
 Chosroes I : description of carpet made for, 10, ii, 12, 
 14, 29 
 
 Circassian (Tcherkess) rugs, 165 
 
 Clam-Gallas, Count : Persian carpet owned by, 25 
 
 Clarke, Sir C. Purdon : on Indian carpets, 38 
 
 Cleansing of carpets, 146 
 
 Cone (or crown-jewel) motive, 131 
 
 Convent of Sta Clara, Valencia : carpets owned by, 69 
 
 Cracow Cathedral : Persian carpet in, 8, 23 
 
 Croft-Lyons, Colonel : Chinese carpet owned by, 66, no 
 
 Cunliffe, Lady : Chinese carpet owned by, 109 
 
 Czartoryski, Prince : carpet owned by, 30 
 
 Daghestan carpets, 163, 166. See also the following 
 names ; 
 
 Derbend 
 
 Genghis 
 
 Kabistan 
 
 Karabagh 
 
 Kazak 
 
 Kenguerlu 
 
 Kuba 
 
 Mosul 
 
 Demirdji carpets, 150 
 Denmark : Coronation carpet, 32 
 Derbend carpets, 163, 167, 169 
 
 Design of carpets, 123-132, 152, 187. See also names 
 of countries 
 Animals, 46 
 
 195 
 
INDEX 
 
 Designs of Carpets {eontd.) : 
 
 Figures, 26 
 
 Flowers and plants, 18, 24, 34, 46, 152 
 
 Heraldry, 45 {3), 68, 72 
 
 Hunting, 16, 26, 32, 41 
 
 Love stories, 28 
 
 Nautical, 28, 29 
 
 Timur device, 50 
 
 Dilke, Sir Charles : Spanish carpet presented to the 
 Victoria and Albert Museum, 71 
 Dupont, Pierre : originator of carpet-making process, 
 73 
 
 Dye-stuffs and dyeing, 118-122, 135-136, 140 
 
 England : 
 
 Hulse carpet, 79, 114 
 Inscription on early English carpet, 79 
 Parisot’s factory : the first in England, 80 
 “ Turkey work,” 79 
 
 Use of carpets and carpet-making in England, 75-84 
 
 Farsistan (or Kashkai) carpets, 181 
 Niris (or Laristan), 182 
 Mecca, 182 
 
 Feraghan carpets, 175, 182, 183, 184 ; price of, 143, 144 
 Field designs, 127-132 
 
 Figdor, Dr. A. : early Persian carpet owned by, 1 1 
 Finland : 
 
 Carpet-making in, 72 
 
 Victoria and Albert Museum carpet, 73, 113 
 Fortier, Jehan : originator of carpet-making process, 
 73 
 
 France : 
 
 Dupont, Pierre : originator of carpet-making pro- 
 cess, 73 
 
 Early records of carpet-making, 73 
 Fortier, Jehan : originator of carpet-making process, 
 73 
 
 Savonnerie factory : carpets made for Louis XIV, 73 ; 
 union of, with the Gobelins factory, etc., 73-74 
 Fraudulent carpets, 138-142 
 Fremlin carpet in India, 38 
 
 Garland, Mr. C. T. : Persian carpet owned by, 27 
 Genghis carpets, 165, 166, 167, 168, 169, 170, 171 
 Ghiordes carpets, 150, 153 ; prayer-carpets, 54, 156 ; 
 
 value of, 144 
 Ghiordes knot, 93-95 
 
 Girdlers’ Company : Indian carpet owned by, 38, 39-40, 
 80, 106 
 
 Gobelins factory, 74 
 
 Gorevan (or Yoraghan) carpets, 174, 183 
 Grouping and identification of carpets in the market, 
 149-192 * 
 
 Guli Hinnai pattern, 128 
 
 Hamadan carpets, 175, 183, 185 
 
 Harris, Mr. Lionel : Indian and Spanish carpets owned 
 by, 69, 69 (i ), 106 
 
 Hendley, Colonel T. H., C.I.E. : authority on Indian 
 carpets, 38 
 
 Herat carpets, 35, 39, 180 
 Herati pattern, 128 
 Hereke carpets, 150 
 Herez carpets, 175 
 
 Hulse, Lady : early English carpet owned by, 79, 114 
 
 Ilchester, Earl of : Persian carpet owned by, 20 
 India : 
 
 Buccleuch carpets, 41 
 Carpet-making in, 37 
 Fremlin carpet, 38 
 
 Girdlers’ Company carpet, 38, 39-40, 80, 106 
 
 India {conid.) : 
 
 Hunting designs, 41 
 
 Jeuniette carpet, 40 
 
 Lionel Harris carpet, 106 
 
 Nineteenth century and ‘‘ jail carpets,” 41 
 
 Persian carpets and weavers supplied to India, ^7, 29 
 
 Roden carpet, 40 
 
 Salting Collection carpet, 39 
 
 Sarre carpet, 40 
 
 Irak-Ajemi carpets, 175-177. See also the following 
 names : 
 
 Feraghan 
 Hamadan 
 Jushaghan 
 Kashan 
 Mahal 
 Muskabad 
 Saraband 
 Saruk 
 Savalan 
 Sultanabad 
 Iran carpets, 185 
 Ireland : carpet-making in, 83 
 
 Ispahan : Hall of the Forty Columns carpet, 25 ; car- 
 pets made at Ispahan, 36, 184 
 Italy : Armenian carpets in, 15 
 
 “ Jail carpets ” in India, 41 
 Jakshibehdir prayer-carpets, 159 
 Jesser, William ; carpet factory in 1759, 81 
 Jeuniette, M. : Indian carpet in collection of, 40 
 Jushaghan carpets, 176, 184 
 
 Kabistan rugs, 163, 167, 168, 169 
 
 Karabeck, Dr. v. : Description of early Persian carpet, 
 10 
 
 Karabagh carpets, 165, 166, 167, 170, 171, 182 
 Karadagh carpets, 174, 182 
 Karaje carpets, 179 
 
 Kashan carpets, 29, 176 ; price of, 143, 144 
 Kashkai carpets. See Farsistan 
 Kazak carpets, 164, 166, 168, 170, 171 
 Kenguerlu rugs, 166 
 Kermanshah carpets, 179 
 Khiva rugs, 188, 190 
 Khorassan carpets, 179, 185 
 Herat, 180 
 Meshed, 179 
 
 Kilim rugs and prayer-carpets, 159, 165, 166, 186 ; 
 
 price of, 143. See also Sile and Verne carpets 
 Kir-Shehr rugs and prayer-carpets, 155, 158 
 Kirman carpets, 25, 35, 180, 183, 185 ; price of, 143 
 Kis carpets, 156 
 Kisiliyak carpets, 190 
 Koltuk carpets, 169 
 
 Konia (Iconium) ; Oriental carpets discovered in, by 
 Dr. F. R. Martin, 6, 45 
 Konia rugs, 155 
 Koula rugs. See Kula 
 Kuba rugs, 163, 167, 168, 169 
 Kula rugs and prayer-carpets, 55, 154, 157 
 Kunstgewerbe Museum, Leipzig ; Persian, Spanish, and 
 Turkish carpets owned by, 35, 47, 69 
 Kurdistan carpets, 178, 185 ; price of, 143, 144 
 Kutais rugs, 165 
 
 Ladik rugs and prayer- carpets, 56, 154, 157, 158 
 Lamm, Mr. Carl Robert ; carpets owned by, 12, 14, 24, 
 
 29 
 
 Laristan carpets. See Niris 
 
 Lemberg Museum : Polish carpets in, 72 
 
 Lesghian rugs, 163 
 
 Lessing, Dr. Julius : carpets represented in paintings, 44 
 
 196 
 
INDEX 
 
 Liechtenstein, Prince : Turkish prayer-carpet owned by, 
 47 
 
 Local origin of early carpets : inaccuracy of records, 
 38 (2) 
 
 Looms for weaving carpets, 88-90 
 Louis XIV : carpets made for, at the Savonnerie fac- 
 tory, 73, 1 13 
 
 Love stories depicted on Persian carpets, 28 
 Lule carpets, 178 
 
 McLaren, Hon. H. D. : Persian and Spanish carpets 
 owned by, 1 1, 71, 1 12 
 
 Madziarski, John ; weaving establishments of, 30, 31 
 
 Mahal carpets, 177 
 
 Makri carpets, 155, 160 
 
 Manufacture of carpets. See also Technique 
 
 Caucasian methods, 161 ; Persian, 173 ; Turkish, 
 46, 151 ; Western Turkestan, 186 
 Materials used, types, etc., 31-36, 161, 173 
 Superiority of Persian carpets, 43, 172 
 Tapestry process, 33-34 
 
 Marling, Sir Charles : Kirman carpet owned by, 26 (i) 
 Martin, Dr. F. R. : Oriental carpets discovered by, in 
 Konia, 6, 45 ; Asia Minor carpet, opinion as to, 
 13 ; Ispahan carpet, 26 
 Materials used in carpets, 114-116 
 Mecca carpets, 168, 169, 170, 182, 184, 185 
 Medallion or Ushak carpet, 150 
 Melas carpets, 155, 160 
 Meshed carpets, 179 ; price of, 143 
 Mina Khani pattern, 128 
 Moghan carpets, 167 
 
 Moore, Thomas : carpets made by, in the eighteenth 
 century, 81 
 
 Morris, William : carpets made by, 83 
 Mosul carpets, 167, 178, 185 
 Moth, to combat, 147 
 
 Mounsey, Mr. G., carpets owned by : Persian, 36, 
 
 106 ; Turkish, 48, 107, 108, 109 ; prayer-carpet, 
 55, 109 ; Oriental, 108, 109 
 Mujur rugs and prayer-carpets, 154, 158 
 Muskabad carpets, 177, 184 ; price of, 143 
 
 Nigde rugs, 155 
 
 Niris (or Laristan) carpets, 182 
 
 Northumberland, Duke of : early English carpet owned 
 by, 1 14 
 
 Oushak carpets. See Ushak 
 
 Paintings : carpets represented in, 23, 44, 48, 68 
 Parisot, Pierre : first carpet-maker in England, 80 
 Passavant, of Exeter : carpets made by, in 1758, 81 
 Persia : 
 
 Ardabil carpet, 17-19, 30, 43, 104 
 
 Ardelan carpets. See that name 
 
 Azerbaijan carpets. See that name 
 
 Bardini carpet, 8, 27 
 
 Buccleuch carpet, 27, 32 
 
 Camel-hair carpets, 183 
 
 Chinese design in, 13 
 
 Chosroes carpet, 10, ii, 12, 14, 29 
 
 Clam-Gallas carpet, 25 
 
 Cracow Cathedral carpet, 8, 23 
 
 Czartoryski carpet, 30 
 
 Design of carpets, 173 
 
 Farsistan carpets. See that name 
 
 Figdor carpet, 11 
 
 Figure subjects, 26 
 
 Floral designs, 18, 24, 34 
 
 Garland carpet, 27 
 
 Gobelins Museum, Paris : Persian carpet owned by, 
 20 
 
 Herat : carpet-making in, 35, 39 
 
 Persia (contd.) : 
 
 Historical records of carpet-making, 8 et seq. 
 
 Hunting designs, 16, 26 
 Ilchester carpet, 20, 104 
 
 India : carpets and weavers supplied to, 37, 39 
 
 Irak-Ajemi carpets. See that name 
 
 Ispahan. See that name 
 
 Kashan : carpets made at, 29 
 
 Khorassan carpets. See that name 
 
 Kirman carpets, 25, 35, 180 
 
 Kunstgewerbe Museum carpet, 35 
 
 Lamm carpets, 12, 14, 24, 29 
 
 Love stories designs, 28 
 
 McLaren carpets, ii 
 
 Marling carpet, 26 (1) 
 
 Materials used in carpets, etc., 29, 173 
 Miniature paintings and carpet designs, 23 
 Mounsey carpet, 36, 106 
 Nautical scenes, 28, 29 
 Plant forms, 24 
 
 Poland : silk and gold carpets erroneously attri- 
 buted to, 30, 72 
 Poldi-Pezzoli carpet, 28 
 Rothschild carpet, 21 
 Salting carpet in silk and gold, 9, 21, 22 
 SchSnbrunn carpet, 16, 27, 31 
 Schwarzenberg carpet, 23 
 Stieglitz Museum carpet, 22 
 Superiority of Persian carpets, 43, 172 
 “ Turkey ” carpets sometimes made in Persia, 44 
 Verses inscribed on carpets, 21 
 Victoria and Albert Museum carpet, 19 
 White grounds, use of, 50 
 Yezd : carpet industry, 29 
 Pile-carpets : origin of, etc., 3-6, 29 
 Pinde carpets, 188 
 Poland : 
 
 Lemberg Museum carpets, 72 
 
 Persian silk and gold carpets erroneously attributed 
 to, 30, 72 
 
 Poldi-Pezzoli Museum : Persian carpet in, 28 
 Portsmouth, Earl of : Morris carpet made for, 84 
 Practical considerations : 
 
 Purchase and prices, 133-145 
 Treatment, 145-148 
 
 Prayer-carpets : various types, etc., 47, 53-56, 62, 130, 
 132, 152, 154, 157, 158, 159, 160, 166, 171, 191, 192 
 Prices of carpets, 143-145 
 Purchase and prices of carpets, 133-145 
 
 Quedlinburg : panel made by nuns of, 5 
 
 Repair of carpets, 148 
 Robinson, Vincent : on Indian carpets, 38 
 Roden, Dr., of Frankfurt : Indian carpet owned by, 40 
 Roe, Sir Thomas : on Indian carpets, 38 
 Rothschild, Baron Nathaniel : verses inscribed on 
 
 carpet owned by, 21 
 Royal Bokhara carpets, 188 
 Rush matting, 76 
 
 Salting collection : Persian and Indian carpets, 9, 21, 
 22, 39 
 
 Saph prayer- rug, 158 
 Saraband carpets, 175, 184 
 Sarakhs carpets, 178 
 
 Sarre, Dr. F. : Indian and Polish carpets owned by, 40, 
 72 
 
 Saruk carpets, 177, 184 
 Savalan carpets, 177 
 
 Savonnerie factory : carpets made for Louis XIV, 73 ; 
 
 union of, with the Gobelins factory, etc., 73-74, 113 
 Schonbrunn carpet, 16, 27, 31 
 
 Schwarzenberg, Prince : Persian carpet owned by, 23 
 Scotland : carpet-making in, 82 
 
 197 
 
INDEX 
 
 Sehna carpets, 177, 185 ; price of, 143 
 Sehna knot, 93-95 
 Serapi carpets, 175 
 
 Shah-Abbas carpets, 184; pattern, 128 
 Shiraz carpets, 182, 184, 185 ; price of, 143 
 Shirvan carpete, 164. ^See also the following names : 
 Baku 
 Chichi 
 Derbend 
 Genghis 
 Kabistan 
 Kazak 
 Koltuk 
 Kuba 
 Mecca 
 Moghan 
 Talish 
 Tchetcen 
 Tzitzi 
 
 Shusha carpets, 165 
 Sile rugs, 165, 172 
 Silk carpets, 115 
 Single-warp knot, 95-96 
 
 Smyrna carpets, 149, 150 ; price of, 143. See also the 
 following names : 
 
 Demirdji 
 
 Ghiordes 
 
 Hereke 
 
 Sparta 
 
 Ushak or Medallion 
 
 Soane Museum : collection of carpet designs of the 
 eighteenth century, 81 
 Soumak carpets, 165, 172 ; price of, 143 
 Spain : 
 
 Buerkel carpet, 70 
 
 Convent of Sta Clara, Valencia : carpets in, 69 
 Dilke carpet, 71 
 
 Early records of carpet-making, 68 
 Kunstgewerbe Museum : Spanish carpets in, 69 
 Lionel Harris carpets, 69, 69 (i) 
 
 McLaren carpet, 71, 112 
 Spanish and Eastern designs, 71 
 
 Victoria and Albert Museum collection of Spanish 
 carpets, 71, 112, 113 
 Sparta carpets, 150 ; price of, 143 
 Stieglitz Museum : Persian verse-carpet owned by, 22 
 Stockholm : Royal Palace carpet, 32 
 Stockholm Museum : Chinese carpet in, 63 
 Sultanabad carpets, 177 
 Symbolism in carpet-design, 131-132 
 
 Tabriz carpets, 174, 183, 185 ; price of, 143 
 Talish carpets, 167, 168, 169 
 Tapestry process (Persian), 33-34 
 Tcherkess (or Circassian) rugs, 165 
 Tchetcen rugs, 164 
 Technique : 
 
 Colour and dyeing, 118-122 
 
 Dimensions and shape, 116— 118 
 
 Group characteristics of early carpets, 99-114 
 
 Knotting and weaving, 87-98 
 
 Materials, 114-116 
 
 Tekke rugs, 188, 190 
 Tientsin carpet, 66 
 Tiflis rugs, 165 
 Transylvanian carpets, 51 
 Treatment of carpets, 145-148 
 Turcoman saddle-bag carpets, 191 
 Turcoman tribes ; carpet-making by, 61 
 Turkestan, Western : carpet-making in, designs, etc., 
 186, 187. See also Chinese Turkestan 
 Turkey : 
 
 Anatolian carpets. See that name 
 Bergama rugs, 155, 159 
 Bode comments on carpet, 46 
 Factory carpet work, 149-151 
 Historical records of carpet-making, 43 
 Home Industry carpets, 151-161 
 Jakshibehdir prayer-carpet, 159 
 Kir-Shehr rugs, 155, 158 
 Kis carpets, 156 
 
 Konia : Oriental carpets discovered in, by Dr. F. R. 
 
 Martin, 6, 45 
 Konia rugs, 155 
 
 Kula rugs and prayer-carpets, 55, 154, 157 
 
 Ladik rugs and prayer-carpets, 56, 154, 157, 158 
 
 Makri rugs and carpets, 155, 160 
 
 Melas rugs and carpets, 155, 160 
 
 Mounsey carpets, 48, 107, 108, 109 
 
 Mujur rugs, 154, 158 
 
 Nigde rugs, 155 
 
 Prayer-carpets, 47, 53-56, 109, 152 
 Smyrna carpets. See that name 
 Turcoman tribes : carpet-making by, 61 
 “ Turkey-carpet ” : origin of name, etc., 9, 44, 149 
 Tuzla rugs, 155 
 
 Ushak carpets, 48, 100, 149, 150 
 Yuruk carpets, 156, 161 
 
 “ Turkey-carpet ” : origin of name, etc., 9, 44, 149 
 “ Turkey work ” : knotting process, 79 
 Tuzla rugs, 155 
 Tzitzi rugs, 164 
 
 Ushak Turkish carpets, 48, 100, 149, 150 
 
 Verne rugs, 165, 172 
 Verses inscribed on carpets, 21 
 
 Verulam, Lord : early English carpet owned by, 78 
 
 Washing carpets, 146 
 Washing carpets chemically, 139 
 Whitty, Thomas : carpets made by at Axminster, 
 in 1757, 81 
 
 Wolsey, Cardinal : early use of carpets by, 76~j8 
 
 Yezd : carpet industry, 29 
 Yomud rugs, 188, 191 ; price of, 143, 144 
 Yoraghan (or Gorevan) carpets, 174, 183 
 Yuruk rugs, 156, 161 
 
 Printed in Great Britain by Butler & Tanner, Frame and London 
 
 198 
 

— . I 
 
NEt/ON & NEUON 
 
 BOOK/ 
 
 • NfWVO/K 
 
 l^tnv CENTER LIBRARV 
 
 NK2795 K33 
 
 HanV'‘o"e‘^ -rjets, 
 
 M/tIN 
 
 3 3125 00163 9877 
 
SESS\SiSSi 5 iS|{E 
 
 
 
 c’r.i*; i‘iWnlfh^-^ 
 
 f J| • f ^* * f 'T^t* t-^T* i-* ; ' 
 
 ■t'--i' ,r^ * h'rh' r.i’v 
 
 Wti: 
 
 
 JimUlHi' 
 
 
 i» » "t ‘ r t ' t .^v 
 
 H clu': 
 
 : ' ih 
 
 
 «ii 
 
 :i‘t,i;frf;;.!L. 
 
 
 
 nail 
 
 
 t »>W .-»1 '1 ►♦» *»■• >-♦ '" ' 
 
 ; ivivl 14.1 hfxivW tx! r. ' l: IWS^- 
 
 
 •i..:;;V; i ;V.^rw^ v^:s'--.'i :'^uv-rx 
 
 ' * jt'f 
 
 
 
 SHl'‘’'r|^ 
 
 ivT/.te 5 '' ^'V-'V 
 - 1 '; b 
 
 ;• i)Mti t Ui 
 
 u K'*\ 'i: 
 
 kvj^ 
 
 : j-rth* ib ! v V' T 
 
 I'ThVttHtJiJrv 
 
 j; 4 t;;;:.riT!t;;’ 
 
 T?vhL“;;l^i:r; 
 
 i^‘- 
 
 
 Vi 1 i ;.'< K!l> 4 'p. ii'-^ 
 
 jV;r;iVi]s'#T-'#^^ 
 
 ti i V! r ! t r& SV’J ’ 
 
 
 
 •h jn:-.lih 
 
 titTibhirh •.C’% T, • 
 
 rlTJOihh ^’^;b’P•*r^ 
 
 ! l; V,'- h * I h V V? iiV,' i' ; 
 
 r-t^i j/ 1 ? ‘Ti! -..’;J :M : 
 
 i r] 
 
 
 i] V cl V Iti t jcJ t; ijv44 i I ru{ 
 
 ? rpV r‘rrb;ni;^n, ;n j-t< ; 
 
 .jltrdiiKfi tiidi-iT il ;rb,.v 
 
 ■rJlVV;! 
 
 !|»h^!;iI.jtiK.l 
 
 ''i‘; '.-* c 4 \\\j, 
 
 I di ; u > ; I : u' X ' 1 4 U 5 T ^ 4 h b ’ t ; *4 v^v.i ' } h ] 
 
 1 4 t ^‘; I V . •, t, , ; 1 . , K ‘ . i A ^ ^ r'J'* 4 J M - J b x ^ 
 
 fe 5 Mvt|gig!*p|iiii;i'^ 
 
 ;!x^tbb 
 ; *’ 'K 4 *i 
 
 i«?!Hii''i§iflHi 
 
 t*'j i t;;:[ ^p Ww 
 
 . tp., Tblwj, ;:r.trub 5.1 
 
 ’rP}\ Th ', rbh tb/jib 
 
 . ' . > , It. t v^» * ! '■*■*'»'■¥ f'ldt 
 
 
 •i J i Vtii&TJT VJt' t-i'* 
 
 
 ViWiW 4 tt.V i{;V I rt’i 
 ib ! V V]i! t V ' 1. • 1 ; t V h I ; ',; 
 
 i; i 4 'iii|'i '1 -T-t'-V ' 
 
 ^ * t* f rr‘'frl-tlT i^.‘LZ:l ^I’.l *.•»! r. 
 
 c*T‘ihnb i;iiir*f^r.l jrh^ Lito'.r.ijv 
 
 r'r'f-'N'ifLV'l 
 
 U f if' H i * • ' ' f- n h * ? * ' ^'! b 
 
 StirLTaiStjnrirvtei'iiii rJ >; Vf I I-.V- ' 
 
 rWi? jdu tl-cd ;'{ .rh.; 
 
 ih ' 
 
 
 nl f;r^ j 
 
 iiin^ 
 
 
 !V'.'.f<.'';i;.^»-|.'i 2 'V'.i 
 
 
 ' 1 nr' I .' : • 
 
 [.t^bilbirbfTCt,-' 
 
 .' a- ’r '?'fe 5 fi 
 
 
 iiiii^ 
 
 ,sS^:' i.fi 'j b-' . b 
 
 i 7 iVT'r»a< 'ibN-'L^'i '1 
 
 iilsilii 
 
 [gf > 4 ; J ai