Oriental Rugs Part 3

THE WARP.-The appearance of the back of a rug is partly due to the relative positions of the two threads of warp encircled by a knot. If, for instance, in any Kazak a pin be thrust through the nap wherever a single perpendicular line of one colour appears at the surface, it will be seen that each of the two threads of warp encircled by a single knot lie side by side with equal prominence. This is shown in Plate B, Figs.

7 and 7-a (Page 49), in which the former represents a section of a rug cut transversely to the threads of warp, and the latter the appearance of the rug at the back. The same will be found true of Beluchistans, Feraghans, Yuruks, and many others. If, however, a Kulah, Persian-Kurdish, or Karabagh be similarly examined, it will be seen that one thread of warp to each knot is depressed, so that the back has a slightly corrugated appearance (as in Plate B, Figs. 8 and 8-a). And in the case of a Bijar or Sarouk it will be seen that one thread of warp, included in every knot, has been doubled under so as to be entirely concealed from view; with the result that the foundation of warp has a double thickness, which makes the rug much stronger, as in Plate B, Fig.

9, representing a Ghiordes knot, and Fig. 10 representing a Sehna knot.

To be sure, it occasionally happens that in rugs of a particular cla.s.s some may have each thread of warp included in a knot equally prominent and others may have one slightly depressed; or that in rugs of another cla.s.s some may have one thread of warp depressed and others may have it entirely concealed; but as a rule these tribal features show a remarkable constancy. These relative positions of the two threads of warp encircled by a knot are partly due to the degree of closeness with which the threads of warp are strung, also partly to the method of inserting the threads of weft or "filling" between the rows of knots; but more than all else they are due to the way one end of the knots is pulled when they are tied.

THE WEFT.-In the character and arrangement of weft are technical differences that are more serviceable than any other feature for distinguishing between the rugs of different tribes and districts. So subtle are some of them that they can be learned only by long and painstaking study, and are appreciated by few except native weavers.

Nevertheless, to any one who will carefully examine almost any well-known cla.s.ses, it will be apparent that these differences in the weave are real, and that they are sufficiently constant to differentiate one cla.s.s from another. The fine brown weft of the Bokhara, or equally fine bluish weft of a Sarouk that is almost concealed between firmly tied knots; the fine thread of cotton weft pa.s.sing but once between two rows of knots and covered only by the transverse warp of the Sehna; the coa.r.s.e thread of cotton weft similarly pa.s.sing but once between two rows of knots in the Hamadan; the coa.r.s.e thread of cotton weft that once crossing and recrossing appears irregularly between appressed rows of knots in Kermanshahs; the bead-like appearance of the threads of weft that, as a rule, pa.s.s many times between two rows of knots in Genghas; the crudely spun weft of coa.r.s.e diameter crossing and recrossing once between the rows of knots in modern Mosuls; the very fine reddish brown weft that entirely conceals from view the warp in old Bergamos, -are features peculiar to these separate cla.s.ses with which every rug expert is familiar. The weft of many other cla.s.ses is equally distinctive, though there are exceptions to the types. It should be remembered, however, that the weave of many rugs woven over a hundred and fifty years ago is different from the weave of rugs woven only fifty years ago; and that many modern pieces cheaply made for commercial purposes are more crudely woven than were the same cla.s.ses thirty years ago.

[Ill.u.s.tration: PLATE 8. HAMADAN RUG]

These distinctions in the weft relate to the material of which it is made, its colour, the size of the diameter, the way in which it is spun, to its loose or compressed condition between separate rows of knots, as well as to the number of times it crosses the warp between them, and to whether it is inserted with much or little slack. Most rugs are woven with woollen weft of a natural colour, but occasionally it has a reddish brown, a blue, or a yellow tint. When cotton, jute, or hemp are used, they are almost invariably of natural colour; only in a very few pieces, as some of the Kulahs, are both wool and jute ever used in the same piece. The weft of some cla.s.ses, as Bokharas, Sarouks, and Bergamos, is of a very small diameter, and of others, as the Hamadans and Kurdistans, it is of relatively large diameter. In some cla.s.ses, as the Karajes and Genghas, the weft is tightly spun like twine; while in the Beluchistans, Mosuls, and Kurdistans it is loosely spun, so that the projecting fibres of wool give a rough appearance to its surface.

Also the weavers of some districts invariably compress very firmly the yarn of weft between every two rows of knots, while other weavers compress it only to a slight degree; as, for instance, in the Afghan, Tabriz, and Kirman the rows of knots are pressed down so firmly that the weft is almost concealed at the back and the transverse threads of warp are entirely covered; whilst, on the other hand, in the Karabagh or Kazak between every two rows of knots the weft and part of the transverse threads of warp are exposed to view. According as the rows of knots are pressed down upon the threads of weft or not, one of the two halves of each separate knot, as shown at the back, may extend slightly or very much beyond the other in the direction of the length of the rug, or each of them may lie in a straight line at right angles to the warp.

Comparing Kazak, Kutais, and Tiflis rugs, for example, it will be noticed that as a rule the line thus formed in Tiflis rugs is nearly an even, clear cut line at right angles to the warp, that in the Kutais part of one knot extends beyond the other, while in Kazaks this unevenness is even more conspicuous. Or again, if typical Shirvans, Kabistans, and Daghestans be compared, it will be noticed that in Shirvans the half-knots, or parts of the knot encircling the two adjacent threads of warp, are often inclined at an angle of at least thirty degrees to the line of weft so as to present a serrated appearance, but that the alignment formed by knots of Daghestans is nearly even, and that of Kabistans is intermediate. To be sure, there are exceptions to this rule, but these features are remarkably constant.

The number of times that a thread of weft is inserted between two rows of knots varies with the practice of different localities, but is almost constant in each locality. Weavers of Sehna rugs insert only a single thread of weft between every two rows of knots, which winds in front of and behind alternate threads of warp, with the result that the back of these rugs have a checkered or quincunx appearance, caused by minute portions of exposed warp and weft crossing each other at right angles.

In Hamadans a much heavier thread of weft pa.s.ses only once before and behind alternate threads of warp, so that the appearance of the weave is very similar to that of Sehnas.[9] In almost all other rugs the weft crosses twice, that is, across and back once, between every two rows of knots so as to completely encircle each thread of warp. The weave of a few rugs, as some Anatolians, shows the weft crossing three times, that is, twice in one direction and once in the opposite direction. In the Genghas, Tcherkess, Bergamos, and in many rugs over one hundred and fifty years old, the threads of weft frequently cross many times; and it is not unusual for the number to vary in different parts of the same rug. In Plate B, Figs. 15, 16, and 17 (Page 49), is ill.u.s.trated the appearance of the back of rugs in which a thread of weft crosses once, twice, and three times between two adjacent rows of knots. There are likewise rugs in which the number varies with methodical regularity; for instance, in Khora.s.sans it is usual to find an extra thread of weft inserted at intervals of every few rows of knots; in many Herats the threads of weft cross twice between several successive rows of knots, then three times between the following several rows, and so continue to alternate; and in some of the Kulahs a thread of woollen weft that crosses twice alternates with a single coa.r.s.er thread of jute.

As the shuttle pa.s.ses back and forth, the thread of weft may be allowed considerable slack, so that when it is pressed down by the comb it will fit about the sides of the warp; or it may be drawn tightly across, so that it has a tendency to displace the threads of warp. If, for instance, a Hamadan and one of the Feraghans which, unlike the majority of them, has only one thread of weft crossing between the rows of knots, be examined, it will be seen that in the Hamadan the weft crosses with hardly any slack, so that the warp stands out clearly and well defined at the back; but that in the Feraghan the weft crosses with much slack, so that it folds about the warp, which accordingly seems slightly imbedded in it at the back. In some rugs the weft is pa.s.sed across once with very slight slack, and as it recrosses it is allowed much slack.

These features of the weave, which are followed with remarkable constancy in the same cla.s.s, can be observed to advantage in a fragment of a rug cut transversely to the direction of the warp. If, for instance, the weft which crosses and recrosses between the rows of knots be carefully removed so as not to disturb its mould, it will be noticed in the case of many rugs that each thread of yarn has a similar shape of moderately deep undulations, which show how it conformed to the warp and indicate how slack it was when inserted. In the case of a few rugs, as the Luristans, each thread will likewise have similar undulations, but they will be very prominent. If, on the other hand, the weft be removed from some cla.s.ses, as the Tabriz, Sarouk, and Kashan, one of the two threads will be almost straight or have slight undulations, while the other thread will have deep undulations. The weft of the Herez, Herats, old Khora.s.sans, and Koniehs have the same peculiarities, excepting that the thread with very slight undulations is of three or four ply, while the other is of a single ply.

The only instruments employed in weaving are the knife, comb, and scissors. The first is used, after each row of knots has been tied, to cut the ends of yarn to nearly the proper length; the second, to press firmly each row of knots; and the last, to trim the nap with care, so that the finished product may present an even and compact appearance.

FINISH OF SIDES.-As the sides are constantly exposed to wear weavers of different districts strengthen them in different ways, which may be designated as Weft Overcasting, Double Overcasting, Weft Selvage, Double Selvage, and Added Selvage. Each of these terms, though not euphonious, suggests the method employed. In Weft Overcasting (Plate B, Fig. 11, Page 49) the thread of weft, after encircling the threads of warp to which the knots are tied, is wound about a much heavier cord that is strung at the side of the rug for a s.p.a.ce equal to the thickness of the knots. It then partly encircles the threads of warp between the next two rows of knots as it pa.s.ses to the other side, when it is wound about the heavy cord there. As this process continues, the cords at the sides are completely overcast with the thread of weft. When the sides have a Double Overcasting (Plate B, Fig. 12), heavy yarn is wound about the cord that has previously been encircled by the weft. Frequently several threads of yarn take the place of a single heavy cord. Weft Selvage (Plate B, Fig. 13) is made by placing two or more heavy cords instead of a single one at the side of the warp, and encircling them by the weft in figure-eight fashion. As they extend beyond the rows of knots they form a plain flat selvage. The Double Selvage (Plate B, Fig. 14) is used among nomadic tribes such as the Beluches and Afghans, whose rugs receive an unusual amount of hard wear, so that an extra selvage is necessary. In adding this extra selvage the threads of weft are carried about the heavy cords, as in weft selvage, and then the extra yarn is wound over it in figure-eight fashion so as again to encircle the heavy cords. In Added Selvage the heavy cords are not encircled by the weft, but are attached to the side of the rug by the extra yarn that winds about them figure-eight fashion, and also encircles two or more adjoining threads of warp. Sometimes also the selvage is "Mixed," or made by the weft encircling only one or two of the heavy cords, and then an extra yarn is wound about these and the remaining cord or cords.

Moreover, the Double Overcasting and Double Selvage may be "attached"

more firmly to the sides of the rug by the yarn pa.s.sing in figure-eight fashion about the adjacent thread or threads of warp between the two rows of knots. When a Double Overcasting is thus "attached," it is somewhat similar to a two-cord Double Selvage.

These are the princ.i.p.al methods of finishing the sides, though they are sometimes modified by tribal customs. Simple as they seem, skill is required in both overcasting and making the selvage; for if carelessly done the sides are frequently made to curl. This is occasionally found to be the defect of old rugs, the worn sides of which have been recently overcast by inexperienced weavers. When such is the case, removing the st.i.tches and overcasting with more care will remove the defect.

[Ill.u.s.tration: PLATE 9. SAROUK RUG]

FINISH OF ENDS.-The ends, which receive more attention than the sides, are treated in several different ways; and in many rugs a distinction exists between the treatment of each end. The simplest finish is where the warp and weft are woven like the threads of a kilim, and extend beyond the pile as a web, which may be exceedingly short, or, as in Beluches and Afghans, several inches long. Frequently the ends are finished by a selvage formed by cords heavier than the weft braided into the warp; or the upper end may be doubled back and hemmed. It is not unusual to find both web and selvage; but though the finish be web, selvage, or web and selvage, the warp of the end finished last generally forms a fringe, and often each end will have a fringe. Sometimes each separate thread of warp hangs loose; sometimes a number a foot or more in length are twisted together in cords; and again they are knotted or are tied to one another diagonally so as to form a network from which hang the loose ends. Very frequently the loops formed by the warp that encircled the rod extend beyond the web at the lower end of the rug, or else are braided about the ultimate thread of weft in the web. Though the warp and weft are generally undyed in the body of the rug, the web of the ends is very frequently coloured. Some of the webs, particularly those of the Beluches, are embroidered; and through others, as the Kurdistans, a parti-coloured cord runs transversely; still others, as Bergamos, are adorned with sh.e.l.ls, beads, or other articles to avert the evil eye.

In the study of rugs it should be remembered that the effect of rough usage is so considerable that in old pieces the webs of the ends have frequently disappeared, leaving short fringes composed of the ends of warp from which some rows of knots have been removed, and that recent overcasting of the sides may take the place of former selvage.

The many characteristics of knot, warp, weft, sides, and ends, with all the variations made by innumerable tribes, remarkably constant in their methods, are technical peculiarities that are uninteresting to those who feel only an aesthetic interest in rugs, yet they demand the most careful consideration of whoever would learn to differentiate accurately between the many cla.s.ses. Though admitting of exceptions, these peculiarities are real and definite, yet their a.n.a.lyses often require the subtlest perception of small though exact distinctions, without which expert understanding would be impossible.

CHAPTER VI

DESIGNS AND SYMBOLS

HOWEVER well woven, however resplendent in rich modulations of colour, Oriental rugs would quickly lose their fascination if in patterns and designs there were not at least some partial expression of the simple lives of the people, of their religious feelings, and of that veiled mysticism which pervades the thought of every Eastern race.

In all nomadic rugs as well as in many others are innumerable reminders of common life. It may be only crude outlines of the goat or camel, or realistically drawn rose and lily; but even these are suggestive of a.s.sociations.

Religion, too, exercised on the character of rugs an important influence, which is expressed in the symbolism of both designs and colour. Without a doubt, some of these well-known designs have been transmitted from the earliest times, and were once a.s.sociated with different forms of idolatry. Thus, among the old Babylonians the sun and moon, which are sometimes seen in the old pieces, represented particular deities; and very many of the oldest Chinese rugs that remain also contain symbols of their deities. In the early religion of Iran, which over 1000 years B.C. was reduced to a system by Zoroaster, the elements were worshipped, so that designs representing these elements would likewise represent the divine forces they personified. This ancient fire worship of the Pa.r.s.ees, which even to-day has a few devotees in parts of Persia, and the kindred sun worship have added much to the symbolism of Oriental rugs. Buddhism also has contributed its share; and with the spread of Mohammedanism appeared a Saracenic influence that is frequently recognised. Even the two great sects of Mohammedan followers, the Shiites and Sunnites, have had distinct effects in the use of designs, as the former employed animal figures and the latter prohibited them.

[Ill.u.s.tration: PLATE 10. SARABEND RUG]

There was, moreover, a symbolism that in a manner expressed the vague philosophic teachings of the ancient races. It was but natural that the early weaver engaged in tedious sedentary work, often requiring many months of constant application to complete, should endeavour to express therein not only artistic taste, but also the spirit of his innermost thoughts. So as he wove he sometimes left the result, though poorly defined and little understood to-day, of that struggle to interpret the great mysteries of the visible and unseen universe, from which arose the crude beginnings of philosophy.[10]

The patterns, however, of many Oriental rugs are chiefly decorative.

Even a casual examination shows that in all of them the coloured knots of the surface represent a border surrounding a central field. The former serves much the same function to the latter as a frame to a picture; yet its office is in no wise subordinate. Nor is there any part of the pattern more useful in determining the place of its origin. For this reason it is well to clearly define the special names which in this work are applied to its different parts. The lineal divisions are designated "Stripes," though they are frequently spoken of as separate borders. At or near the centre of the border is the main stripe, which is generally accompanied by a much narrower pair, one on each side, known as "Guard stripes." Very narrow stripes are sometimes called "Ribbon stripes," and those of only one or two knots in width are called "Lines." The latter are of solid colour or have the simplest geometric device. The decoration of the ribbon stripes is also necessarily simple; but in the main stripe of artistic rugs the patterns are often exceedingly elaborate, of exquisite colours, and co-ordinate in character with those of the field.

The fields display even greater diversity of pattern than the borders.

Frequently they are covered with a heterogeneous ma.s.s of detached and unrelated figures, as in many of the nomadic rugs; or, on the other hand, are entirely covered with repet.i.tive patterns, as in the Turkomans; or with intricate and correlated designs, as in the diaper and floral patterns of so many of the Persian and Indian rugs. Others consist of a background of solid colour on which appear isolated formal designs, as in some of the Khora.s.sans, or medallions on which are represented smaller figures, as in some of the old Sehnas and Feraghans.

Occasionally the medallions are so large as to extend to the sides and ends of the fields, and thus separate the corners into nearly triangular shaped sections, such as are characteristic of a number of Persian and one or two Caucasian rugs. In a few of the latter, also, the fields are covered with large geometric figures suggesting the medallions. It is only within comparatively modern times that weavers have used solid colours for large portions of the fields. The intent no doubt was a saving in labour and pains, but the effect is frequently most gratifying; as when, for instance, the s.p.a.ce beneath the arch of Asia Minor prayer rugs is of a uniform red, blue, or cream, relieved only by a gracefully suspended lamp, the tree of life, or some other emblem of immortality.

[Ill.u.s.tration: PLATE C.-PRAYER ARCHES OF PERSIAN, CAUCASIAN, AND CENTRAL ASIATIC RUGS

No. 1, Antique silk. No. 2, Kermanshah. No. 3, Khora.s.san. No. 4, Sarouk.

No. 5, Persian, XVI Century. No. 6, Kashan. No. 7, Feraghan. No. 8, Shiraz. No. 9, Daghestan, Kabistan, Shirvan, Chichi, Kazak, Karabagh.

No. 10, Kazak. No. 11, Shirvan. No. 12, Karabagh (not usual). No. 13, Beluchistan. No. 14, Bokhara, Tekke. No. 15, Khiva. No. 16, Beshire.]

The patterns of prayer rugs are not only pleasing, but have a peculiar importance, as weavers of certain sections of the country adhere so strictly to time-honoured traditions that the shape of the arch, or mihrab, which is the princ.i.p.al feature, often denotes the cla.s.s, as well as the group, to which they belong. Some of these arches are ill.u.s.trated in Plates C and D (Pages 61 and 63), from which it will be seen that in Persian rugs they are formed by gracefully curving lines, but that in rugs of other groups, with the exception of a very few old Ghiordes pieces, they are geometric. The peculiarities of the arches of the several cla.s.ses, also, are observable; as those of the Beluchistans, which are rectilinear and relatively high, and those of the Bokharas, which are tent-shaped, flat, and small. In the Caucasian group they have a marked resemblance to one another and also to those of the Turkoman rugs, but are larger than the latter. Again, the arch of almost all Asia Minor rugs rises higher than those of any others, excepting the Persian, and extends from one side of the field to the other. In many of them a panel is placed above the spandrel, and occasionally a second panel is placed beneath the field. Above the niche of some Asia Minor and Caucasian prayer rugs is woven a small rhomboidal figure, where the suppliant plants the pebble or bit of earth that he has brought from Mecca; and at the sides of a few arches are crude figures, where are placed the hands during the act of worship. More than one arch is the exception; but now and then are seen two and even four, one above the other, or several parallel to one another. These and other special features a.s.sociated with prayer rugs will be considered more fully in subsequent chapters.

The smaller designs that appear in rugs and compose the general pattern are distinguished as geometric and floral ornamentation. The former is adopted in those countries where the population is princ.i.p.ally nomadic; and the latter is the accepted style in countries where exist numerous towns and cities in which the arts have been cultivated and where a large percentage of the population have enjoyed an advanced state of society. Thus in Caucasia, Turkestan, Afghanistan, and Beluchistan geometric designs are characteristic of the rugs; but in China, India, Persia, and part of Asia Minor floral designs prevail. Sir George Birdwood, an eminent authority on Oriental rugs, has made the statement that the geometric designs are found among the lower Turanian and the floral among the higher Aryan. But it seems most probable that the adoption of the geometric or floral style of ornamentation is due not so much to racial distinctions as to the state to which the textile art had advanced among the different peoples and to the waves of influence that at times spread over the countries. Thus the early rugs of Asia Minor had patterns that were more geometric than those of later times, and during the period when the Mongols ruled in Persia geometric patterns were more frequently employed in the rugs of that country than subsequently.

[Ill.u.s.tration: PLATE D.-PRAYER ARCHES OF ASIA MINOR RUGS

Nos. 1, 2, 3, 4, Ghiordes. Nos. 5, 6, 7, Kulah. No. 8, Ladik. Nos. 9, 10, Bergamo. Nos. 11, 12, Melez. No. 13, Kir-shehr, Mudjar. Nos. 14, 15, Konieh. No. 16, Anatolian.]

In all rugs, however, some trace of the floral design appears. Even in the Turkoman weavings, the pattern of which is strictly geometric, some vestige of the tree of life is manifest. In the fields of Caucasian rugs, in which are represented squares, octagons, triangles, diagonals, lozenges, stars, etc., the weavers have depicted designs that are almost as geometric as those of the Turkoman rugs by which they have been influenced; but, on the other hand, large numbers of the border designs are distinctly floral. In the rugs of China and Asia Minor are found both geometric and floral ornamentation, the latter predominating in pieces woven during the last two centuries, and the former in those of earlier date. Among the woven fabrics of India and Persia, however, few traces of the geometric pattern remain; but vines, leaves, and flowers form the favourite theme for decoration.

The floral patterns are the result of many centuries of growth, that reached its highest development in the Persian carpets of the XVI and XVII Centuries; and since then till the present time they have continued as the most characteristic features of the rugs of that country and India. They represent the highest technique of the weaver. In the borders are generally represented vines from which are pendant rosettes, palmettes, or flowers; and in the fields, particularly those of the fine antique rugs, are a profusion of floral forms realistically portrayed.

On long, gracefully twining and intertwining stems is often the rose, pink, violet, lotus, crocus, narcissus, or daisy. But if the rug is more modern, in its field of uniform colour may be represented a central medallion covered with delicate tracings enriched by bright-coloured conventionalised flowers.

In the general pattern of all rugs are interwoven particular designs or motives that give them a distinctive character and render the greatest a.s.sistance in distinguishing the groups and cla.s.ses. Thus the serrated leaf and wine cup (Plate I, Fig. 1, opp. Page 226) is found only in Caucasian and old Armenian rugs; and the design represented in Plate H, Fig. 10 (opp. Page 194), suggestive of some Chinese character, is found almost exclusively in Kulahs. Not only are all designs important as aids in cla.s.sification, but they have a special interest, as it is maintained by writers of the highest authority that when employed by the earliest weavers each had a symbolic meaning. To be sure, the origin of many has been lost in the remote past and is unknown even to those who now employ them; but others still represent definite ideas, as they did centuries ago, and portray to some extent the thought of the weaver. They therefore deserve the most careful study.

[Ill.u.s.tration: PLATE 11. CARPET FROM NORTHWESTERN PERSIA

Loaned by C. F. Williams, Esq., to the Metropolitan Museum of Art, New York]

Few of these designs have been transmitted from a more remote past or have been more universally employed than has been the figure [Symbol: S]

or [Symbol: horizontal S]. It appears in each of the groups of Oriental rugs excepting the Chinese; and two of the forms it a.s.sumes are exceedingly like the arms of the swastika and parts of the fret as occasionally seen in Chinese designs. In Indian rugs it is rarely found except in the borders, which may have been copied from those of other countries. It is very commonly seen in the Beluchistans, Tekkes, and Beshires, of the Central Asiatic group; and in the rugs of the Persian group that show nomadic influences. There is probably not a single cla.s.s of the Caucasian group, nor any of the Asia Minor group, with the exception of the old Ghiordes, in which it is not sometimes represented.

It may be seen near the corner of the Asia Minor "Dragon and Phnix"

carpet of the XIV Century, ill.u.s.trated in Plate 20 (opp. Page 88); and appears in some of the old Armenian carpets, which are believed to be even older. That it was a.s.sociated with sun worship and regarded as an emblem of light and the deity is the accepted belief. It is possible, however, that it was intended by some weavers to represent the serpent, which among many different races is emblematic of superhuman knowledge.

Probably no other design has been more universally employed than the swastika, which appears in the textile fabrics of North American Indians, on the Maya ruins of Yucatan, among the monuments of the Nile, and on the temples of India. Widely as is its distribution, its most usual form of intersecting right angles is found in each of these countries. It is not improbable that it originated in China, where it is a most common decorative motive, and was almost invariably represented in the borders and in many of the medallions of rugs woven before the beginning of the XVIII Century. It is also very frequently seen in the rugs of Samarkand, and occasionally in those of Caucasia. It seldom appears in the rugs of Persia, Asia Minor, or Turkestan. The universality of the design indicates its great antiquity, yet its primitive symbolic meaning of abundance, fertility, and prosperity has never been lost. Some of its different forms are shown in Plate O, Figs.

5-a, 5-b, 5-c, 5-d, 5-e, and 5-f (Page 291).

The reciprocal trefoil (Plate F, Fig. 17, opp. Page 158) is a very usual design in Caucasian and Persian rugs; it is often found in some of the rugs of India and in Beluchistans, but is very rarely seen in other cla.s.ses of the Central Asiatic or in the Chinese and Asia Minor groups.

Its origin is uncertain, but since it appears in the "Polish Carpets"

and other antique Persian carpets of strictly floral pattern, where its drawing is more elaborate than in modern rugs, it is not improbable that it is the conventionalised form of the lily or a spray bearing three leaves, and that it has the emblematic significance of the tree form.