The arrangement of the interior of Grecian temples necessarily depended on the mode in which they were lighted. No one will, I believe, now contend, as was once done, that it was by lamplight alone that the beauty of their interiors could be seen; and as light certainly was not introduced through the side walls, nor could be in sufficient quantities through the doorways, it is only from the roof that it could be admitted. At the same time it could not have been by a large horizontal opening in the roof, as has been supposed, as that would have admitted the rain and snow as well as the light; and the only alternative seems to be one I suggested some years ago—of a clerestory,[142] similar internally to that found in all the great Egyptian temples,[143] but externally requiring such a change of arrangement as was necessary to adapt it to a sloping instead of a flat roof. This could have been effected by countersinking it into the roof, so as to make it in fact 3 ridges in those parts where the light was admitted, though the regular slope of the roof was retained between these openings, so that neither the ridge nor the continuity of the lines of the roof was interfered with. This would effect all that was required, and in the most beautiful manner; it moreover agrees with all the remains of Greek temples that now exist, as well as with all the descriptions that have been handed down to us from antiquity. [Illustration: 153. Section of the Parthenon. Scale 50 ft. to 1 in] [Illustration: 154. Part Section, part Elevation, of Great Temple at Agrigentum. Scale 50 ft. to 1 in.] This arrangement will be understood from the section of the Parthenon (Woodcut No. 153), restored in accordance with the above explanation, which agrees perfectly with all that remains on the spot, as well as with all the accounts we have of that celebrated temple. The same system applies even more easily to the great hexastyle at Pæstum and to the beautiful little Temple of Apollo at Bassæ, in Phigaleia (Woodcut No. 149), and in fact to all regular Greek temples. Indeed, it seems impossible to account for the peculiarities of that temple except on some such theory as this. Any one who studies the plan (Woodcut No. 149) will see at once what pains were taken to bring the internal columns exactly into the spaces between those of the external peristyle. The effect inside is clumsy, and never would have been attempted were it not that practically their position was seen from the outside, and this could hardly have been so on any other hypothesis than that now proposed. An equally important point in the examination of this theory is that it applies equally to the exceptional ones. The side aisles, for instance, of the great temple at Agrigentum were, as before mentioned, lighted by side windows; the central one could only be lighted from the roof, and it is easy to see how this could be effected by introducing openings between the telamones, as shown in Woodcut No. 154.