The church was probably only of medium size; the length of the present church is about 250 feet, its vastness being in its width. The _Paschal Chronicle_ speaks of “its stupendous and marvellous columns all being ἐκ τετραέντου”; but owing to a variant reading it is difficult to determine whether it means that the pillars were square, or were set in a square, or formed four bays. Glycas and Codinus, who wrote a thousand years after the foundation of the church, say that it was basilican (_dromika_), and had a wooden roof (_xulotroullos_), and the latter says that the church of Theodosius had cylindrical vaults. As it is evident from the rapid destruction by fire that the roofs of the early churches were of wood, they were probably Basilicas. Only a few minor particulars, such as the existence of an atrium, and the right of sanctuary in the bema (_thusiasterion_), can be gathered from the homilies of S. Chrysostom. Socrates tells us that this patriarch was wont to preach “in the ambo for the sake of being better heard.”[37] From Palladius we learn that there was a baptistery (in which the Sixth Council of Constantinople, A.D. 394,[38] appears to have met) attached to the church, and it was here Chrysostom took leave of the deaconesses at his banishment, as described in a passage difficult to interpret. “He went out of the baptistery on the east side, for there was no western (exit). The mule which he usually rode was made to stand westwards before the gate to the church, where is the porch, so that he might escape the people who were expecting him.” The passage from the same author about the waters of the font being stained with blood does not, as is sometimes supposed, necessarily refer to S. Sophia. In applying the plan of a church of mean size so that the doors should face eastwards, we are at once struck by finding that the western hemicycle of the present church would lie about the apse; and we cannot but suggest that in this we may have the very _raison d’être_ of the remarkable plan of the present church, which it would seem might be properly classed with those churches which have apses at both ends, like the early basilica at Orleansville near Tunis;[39] the MS. plan of S. Gall is the best known example; our own early church at Canterbury was another instance, the result of adding to a church with a western apse; France furnishes Besançon and Nevers, and Germany numerous examples. It is indeed possible that some parts of the old structure may have given practical and positive reasons contributing to this result, and a thorough examination of the cisterns beneath the present floor of S. Sophia may yet yield full evidence of the first basilica; or if these vaults were entirely built for Justinian’s church, their material would almost certainly be derived from the earlier building.