The Indian or horizontal dome never can be made circular in section, except when used on the smallest scale, but almost always takes a form more or less pointed (Woodcut No. 111). From the time of the building of the Treasury of Mycenæ[252] to the birth of Christ we have a tolerably complete series of arches and vaults constructed on this principle, but few domes properly so called. After the Christian Era the first example is found in a singular tomb at Mylassa,[253] near Halicarnassus,[254] where the dome exhibits all the peculiarities of construction found in the Jaina temples of India. After this we lose the thread of its history till the form reappears in porches like those of the 11th century on Mount Abu, where it is a perfectly established architectural feature, that must have been practised long before it could be used as we find it in that building. Whether we shall ever be able to recover the lost links in this chain is more than doubtful, but it would be deeply interesting to the history of art if it could be done. In the meantime, there is no difficulty in explaining the constructive steps by which the object is now attained in India. These may also throw some light on the history of the invention, though this is not, of course, capable of direct proof. [Illustration: 112. Diagram of Roofing.] The simplest mode of roofing a small square space supported by four pillars is merely to run an architrave or stone beam from each pillar, and cover the intermediate opening by a plain stone slab. Unless, however, slabs of great dimensions are available, this mode of construction has a limit very soon arrived at. The next step therefore is to reduce the extent of the central space to be covered by cutting off its corners; this is done by triangular stones placed in each angle of the square, as in Woodcut No. 112, thus employing five stones instead of one. By this means, the size of the central stone remaining the same, the side of the square space so roofed is increased in the ratio of ten to seven, the actual area being doubled. The next step in the process (Woodcut No. 113) is by employing three tiers and nine stones instead of two tiers and five stones, which quadruples the area roofed. Thus, if the central stone is 4 ft., by the second process the space roofed will be about 5 ft. 8 in.; by the third 8 ft. square; by a fourth process (Woodcut No. 114)--with four tiers and thirteen stones--the extent roofed may be 9 ft. or 10 ft., always assuming the central stone to remain 4 ft. square. All these forms are still currently used in India, but with four pillars the process is seldom carried further than this; with another tier, however, and eight pillars (as shown in Woodcut No. 115), it may be carried a step further--exactly the extent to which it is carried in the tomb at Mylassa above referred to. In this, however, as in all instances of octagonal domes in this style, instead of the octagonal form being left as such, there are