In striking agreement with this evidence since the Turkish Conquest, are the accounts given regarding the Church of St. Saviour by writers previous to that event. According to them, the church was in the neighbourhood of the Church of St. George Mangana, and to the east of that sanctuary; it stood close to the sea, immediately behind the city walls; its Holy Spring was enclosed within the walls, and yet could be reached from without; in front of the walls through which the sacred stream flowed, was a beach of sand endowed with healing properties.[921] Nothing can be more conclusive. This identification is of the greatest importance for the topographical reconstruction of the quarters of Byzantine Constantinople along the eastern shore of the promontory, for, with that church as a fixed point, it becomes comparatively easy to determine the positions of other noted buildings in the neighbourhood. By means of that landmark, for example, the situation of the Church of St. George Mangana can, we have seen, be fixed.[922] It enables us also to settle, without prolonged discussion, the question raised by the extensive ruins discovered behind Indjili Kiosk, when the ground was cleared, in 1871, for the construction of the Roumelian railroad. The walls of an edifice 322 feet long by 53 feet wide, were then brought to view, and among the _débris_ marble pillars and capitals were found in such numbers, as to prove that the building to which they belonged had been one of considerable importance.[923] Because some of the capitals seemed ornamented with the heads of bulls and lions, Dr. Paspates came to the conclusion that the ruins were the remains of the celebrated Palace of the Bucoleon.[924] On the other hand, Dr. Mordtmann thinks that here was the site of the Imperial residence, known as the Palace of Mangana,[925] an erection of Basil I.[926] That the latter opinion is the correct one may be proved by means of the fact that the Church of St. Saviour stood at Indjili Kiosk. In the first place, the Palace of Mangana was near the Church of St. George Mangana—so near that the destruction of that palace by Isaac Angelus, to obtain material for edifices of his own construction, was viewed as an act of sacrilege committed against the property of the great saint.[927] But the Church of St. George Mangana, we have found, lay a short distance to the west of the Church of St. Saviour,[928] near the site of Indjili Kiosk. Consequently the remains of a palace near that kiosk must be those of the Palace of Mangana. This conclusion agrees, furthermore, with the fact that the Mangana, which gave name to the palace, was in this vicinity.[929] It is also consistent with the circumstance that the Palace of Mangana was noted for its coolness,[930] as would be characteristic of a residence in the position of Indjili Kiosk, which is exposed to the north wind that sweeps down the Bosporus from the Black Sea.