Then, in the search for dignity and also to give more protection from weather to the walls of the cella, the eaves of the roof would be further prolonged outward and made to rest on sills that were supported by a series of posts. In this way the cella was completely surrounded by a colonnade or peristyle. As the use of stone or marble was adopted, the platform became the _stylobate_, which was approached by three steps, carried along the entire length of all the sides. The cella was built of marble or stucco-covered stone, and marble or stone took the place of the sills and beams of the roof, but the latter continued to be constructed of wood, supported by small columns resting on the capitals of larger ones. The outside sheathing of the roof was of terra-cotta or marble tiles. Unlike the roof of an Egyptian temple which was raised in the centre to admit clerestory windows, that of a Hellenic temple had an uninterrupted slope. Whence then was the light derived for the interior? =Lighting.=--Since all roofs, being of wood, have perished, the explanations that have been attempted are purely conjectural. A remark by Vitruvius, the Roman architect and author of ten books on architecture, regarding the =Temple of Zeus= at =Athens= that it was _hypæthral_ (open to the sky) has led to a suggestion that part of the roof may have been open, as in the case of the Pantheon in Rome. But, at the time he wrote, the cella was exposed because Sulla had carried off to Rome some of the supporting columns. Another Roman writer, Strabo, describes the decastyle =Temple of Apollo= near =Miletus= as hypæthral, but gives as the reason the enormous size of the cella, in which precious groves of laurel bushes grew. So, it is purely a surmise that the portion of the roof may have been omitted and that the temples were hypæthral. Another theory, founded upon the discovery in a temple at =Bassæ= of three marble tiles, or thin slabs, pierced with holes about 18 inches by 10, is that some five of these, let into each side of the roof, would have lighted the interior amply without admitting much rain. Again, the use of marble tiles has afforded a suggestion that, Parian marble being very translucent, the light might have penetrated through. James Fergusson, on the other hand, conjectured that a trench was let into each side of the roof; but this would have needed drains to carry off the water and no sign of a system of drainage has been found in any temple. Other authorities, however, maintain that it was only through the open doorway that light was admitted, which owing to the clear atmosphere of Greece and the reflection from the marble pavement, would be sufficient.