Flint was worked to the highest perfection in the prehistoric age, and continued in use till Roman times. Strictly, it is chert rather than flint, as the beds in which it is found are of Eocene limestone. But in general appearance and nature they are closely the equivalent of the chalk with flints in England, only harder. The prehistoric forms are shown in fig. 92. They exceed the flint-work of all other countries in the regularity of the flaking, the thinness of the weapon, and the minute serration of the edges. At present such work is entirely a lost art, and we cannot imagine the methods or the skill required to produce such results. Besides the weapons, flint armlets were made, chipped out of a solid block, yet no thicker than a straw. These were ground with emery finally to smooth them for wearing. Flint was commonly used down to the XIIth dynasty for knives, but all the dynastic working is far inferior to the earlier. In the XVIIIth dynasty, and later, sickle teeth were still made of flint; and flakes were chipped and used in abundance at some centres in the Roman period. [Illustration: FLINT-WORKING 92. Knives and lances of the best prehistoric work] Before leaving the stone-working we may note the accuracy of work, as this is better seen here than in any other subject. The highest pitch of accuracy on a large scale was reached under Khufu in the IVth dynasty; his pyramid had an error of less than ·6 of an inch on its side of 9069 inches, or 1 in 15,000; and its corners were square to 12″. A change of temperature during a day would make larger errors than this in a measuring-rod. The accuracy of levelling, and of finish of the stones, is on a par with this; joints over six feet long are straight to a hundredth of an inch. The pyramid of Khafra has three times this error, varying 1·5 inch on 8475, and 33″ of angle. That of Menkaura is worse, being on an average 3 inches out on 4154, and 1′ 50″ of angle. At Dahshur the errors are 3·7 on 7459 inches base, and 1·1 on 2065, with angular errors of 4′ and 10′. In smaller work, a beautiful example is the granite sarcophagus of Senusert II, which is ground flat on the sides with a matt face like ground glass, and only has about a two-hundredth of an inch error of flatness and parallelism of the sides. The later ages, so far as we know, have left nothing that can be compared with the accuracy of the early dynasties. CHAPTER VIII JEWELLERY