Indigo and madder yield a range of most satisfactory and perma nent blues and reds ; kermes or some other form of cochineal will supplement the latter ; and several good though less perma nent yellow dyes are available. The secondary colours can be obtained by double dipping; and the more neutral colours—brown, white, black—are often supplied by undyed wool, but in any case present no technical difficulty.
The strips, called runners or kanara, made in pairs to go along the sides of a Persian room, are very useful in corridors. A very common size for small rugs is about 61 by 42 ft., while large ones may reach 25 ft. in length or occasionally more. The large rugs are mostly now made with an eye to the western market, and there is a general tendency for the smaller rugs to be of better quality of design, workmanship and material.
The natural tendency of workers in a district to follow a fixed system of technique even though wide variations are available, and the conservatism which causes them to copy traditional patterns with only slight modifi cations, allow the origin and approximate date of most or iental carpets to be determined with fair certainty. Difficulties of attribution, of course, do frequently arise, especially in the case of the earlier carpets, but it is still often possible to form groups of like carpets even when their exact origin is not known. With regard to dating, much difference of opinion exists and there is a tendency for the as signed antiquity to err in excess. Positive evidence on the point is the exception. A close study, how ever, of the development of de signs shows that their secular changes proceed along definite lines that are mostly in the direction of simplification and debase ment. Also it is seen that new motives are occasionally introduced.
Upon these observations a scheme of chronological sequence may be founded, and if certain carpets in the series can be definitely fixed in time either by the internal evidence of dated inscriptions, or because they are seen reproduced in pictures or mentioned in historical documents, then the whole scheme of dating may be considered to be established. The name commonly applied to a type is usually that of the town or district where it is made, though sometimes that of the tribe that makes it, or of the place where it is marketed. Persia.—In the 16th and 17th centuries the Sehna knot is always used—at least in those carpets that have come down to us : the warp and weft are of silk or cotton and the pile mostly of wool. The texture is fine and there are usually three lines of weft after the knots. The patterns are based on floral motives and the treatment is often highly naturalistic : cloud-bands and other Chinese details are frequently introduced. The colours are rich but quiet and there are usually from ten to twenty different tints, of which crimson is most commonly used for the ground though blue is also seen. The term "Ispahan" is used in the trade to describe these early carpets, but it is not known cer tainly where the various groups were made, and the following classification is based chiefly upon the patterns.
Garden.—In this well marked but very rare type there is a map-like representation of a garden, with paths, flower-beds, trees and water-channels (see fig. 9).
Floral.—One type of floral car pet has a mass of trees and flowers forming almost a thicket, grow ing upwards from one end. More often however the ornament is arranged in a more ordered man ner on a basis of scrolling stems curving about the field. The scheme may be associated with variously shaped panels as in the very famous "Ardabil" carpet in the Victoria and Albert Museum (see Plate IX.). In the rather later examples the floral forms— mostly palmettes—become larger and fewer and give the effect of a number of definite masses, sometimes almost touching, arranged more or less evenly throughout the field.