of Taidu, and represents it. After the expulsion of the Mongols (1368) the new native Dynasty of Ming established their capital at Nanking. But this was found so inconvenient that the second sovereign of the dynasty re-occupied Taidu or Cambaluc, the repairs of which began in 1409. He reduced it in size by cutting off nearly a third part of the city at the north end. The remains of this abandoned portion of wall are, however, still in existence, approaching 30 feet in height all round. This old wall is called by the Chinese The Wall of the Yuen (ie. the Mongol Dynasty), and it is laid down in the Russian Survey. The existing walls were built, or restored rather (the north wall being in any case, of course, entirely new), in 1437. There seems to be no doubt that the present south front of the Tartar city was the south front of Taidu. The whole outline of Taidu is therefore still extant, and easily measurable. If the scale on the War Office edition of the Russian Survey be correct, the long sides measure close upon 5 miles and 500 yards; the short sides, 3 miles and 1200 yards. Hence the whole perimeter was just about 18 English miles, or less than 16 Italian miles. If, however, a pair of compasses be run round Taidu and Yenking (as we have laid the latter down from such data as could be had) together, the circuit will be something like 24 Italian miles, and this may have to do with Polo's error.

Polo, again, says that there were 12 gates-3 to every side. Both Gaubil and Martini also say that there were 12 gates. But I believe that both are trusting to Marco. There are 9 gates in the present Tartar city-viz., 3 on the south side and 2 on each of the other sides. The old Chinese accounts say there were 11 gates in Taidu (see Amyot, Mem. II. 553). I have in my plan, therefore, assumed that one gate on the east and one on the west were obliterated in the reduction of the enceinte by the Ming. But I must observe that Mr. Lockhart tells me he did not find the traces of gates in those positions, whilst the two gates on the north side of the old Mongol rampart are quite distinct, with the barbicans in front, and the old Mongol bridge over the ditch still serving for the public thoroughfare.*

When the Ming established themselves on the old Mongol site, population seems to have gathered close about the southern wall, probably using material from the remains of Yenking. This excrescence was enclosed by a new wall in 1554, and was called the "Outer Town." It is what is called by Europeans the Chinese City. Its western wall exhibits in the base sculptured stones, which seem to have belonged to the old palace of Yenking. Some traces of Yenking still existed in

* Mr. Wylie confirms my assumption: "Whilst in Peking I traced the old mud wall, . . . and found it quite in accordance with the outline in your map. Mr. Gilmour (a missionary to the Mongols) and I rode round it, he taking the outside and I the inside.. Neither of us observed the arch that Dr. Lockhart speaks of. . . . . There are gate-openings about the middle of the east and west sides, but no barbicans." (4th December 1873.)

Gaubil's time; the only relic of it now pointed out is a pagoda outside of the Kwang-An-Măn, or western gate of the Outer City, marked in the War Office edition of the Russian Map as "Tower." (Information from Dr. Lockhart.)

The "Great Palaces" over the gates and at the corner bastions are no doubt well illustrated by the buildings which still occupy those positions. There are two such lofty buildings at each of the gates of the modern city, the outer one (here shown) forming an elevated redoubt.

"Elle a douze portes, et sor chascune porte a une grandisme palais et biaus."

NOTE 4.-The French writer cited under Note 2 says of the city as it stands: "La ville est de la sorte coupée en échiquier à peu près régulier dont les quadres circonscrits par des larges avenues sont percés eux-mêmes d'une multitude de rues et ruelles. . . qui toutes à peu près sont orientées N. et S., E. et O. Une seule volonté a evidemment présidé à ce plan, et jamais édilité n'a eu à exécuter d'un seul coup aussi vaste entreprise."

NOTE 5.-Martini speaks of the public clock-towers in the Chinese cities, which in his time were furnished with water-clocks. A watchman struck the hour on a great gong, at the same time exhibiting the hour in large characters. The same person watched for fires, and summoned the public with his gong to aid in extinguishing them.

The tower indicated by Marco appears still to exist. It occupies the place which I have marked as Alarm Tower in the plan of Taidu. It was erected in 1272, but probably rebuilt on the Ming occupation of the city. In the Court of the Old Observatory at Peking there is

preserved, with a few other ancient instruments, which date from the Mongol era, a very elaborate water-clock, provided with four copper basins embedded in brickwork, and rising in steps one above the other. A cut of this court-yard, with its instruments and aged trees, also ascribed to the Mongol time, will be found in Chapter XXXIII. (Atlas Sinensis, p. 10; Magaillans, 149-151; Chine Moderne, p. 26; Tour du Monde for 1864, vol. ii. p. 34.)

NOTE 6. Nevertheless," adds the Ramusian, "there does exist I know not what uneasiness about the people of Cathay.”

CHAPTER XII.

HOW THE GREAT KAAN MAINTAINS A GUARD OF TWELVE THOUSAND HORSE, WHICH ARE CALLED KESHICAN.

You must know that the Great Kaan, to maintain his state, hath a guard of twelve thousand horsemen, who are styled KESHICAN, which is as much as to say "Knights devoted to their Lord." Not that he keeps these for fear of any man whatever, but merely because of his own exalted dignity. These 12,000 men have four captains, each of whom is in command of 3000; and each body of 3000 takes a turn of three days and nights to guard the palace, where they also take their meals. After the expiration of three days and nights they are relieved by another 3000, who mount guard for the same space of time, and then another body takes its turn, so that there are always 3000 on guard. Thus it goes until the whole 12,000, who are styled (as I said) Keshican, have been on duty; and then the tour begins again, and so runs on from year's end to year's end.'

NOTE 1.—I have deduced a reading for the word Quescican (Keshican), which is not found precisely in any text. Pauthier reads Questiau and Quesitau; the G. Text has Quesitam and Quecitain; the Crusca Questi Tan; Ramusio, Casitan; the Riccardiana, Quescitam. Recollecting the constant clerical confusion between and t, what follows will