[249] Salz., plates vi., x.
[250] It is probable, writes Salzenberg, that originally the buttress masses reached only up to the roof of the gynaeceum, level with the springing of the great arches; as Cedrenus describes how Justinian, at the restoration after the fall of the dome, made outside the building, in the neighbourhood of the main piers, above the roof of the gynaeceum four staircases, “cochleas” which reached up to the dome “to strengthen the vaulting.” Theophanes also speaks of new piers which Justinian erected to strengthen the dome. The circumstances mentioned by Procopius seem to indicate that the abutments of the great arches were not sufficient. See our chap. [x]., § 1, for another interpretation.
[251] These Salzenberg thought later additions, “for the stone projections are not bonded to the piers, and the Silentiary says columns stood in these positions.” We do not so interpret the lines of the poem, and, although Choisy here follows Salzenberg, it is impossible to see, if there were additions subsequent to the completion of the building, how it is that the perfectly symmetrical disposition of the marble panelling shows no disturbance, and the beautiful carved cornice which mitres round these projections has had no additions made to it (our [Fig. 47]). The straight joints, which Choisy in another place specially notes as a method of Byzantine building, were here most wisely applied; for on one side the great pier was of stone, and on the other the buttress pier is of brick.
[252] Salzenberg conjectures from Agathias that these arches were a later addition made when the dome was restored by Justinian. But without them, as he remarks, there would not have been originally a square base for the circle of the dome. See explanation of original form, p. [210].
[253] Salzenberg assumes from Paulus that “the dome was surmounted by a cross”: the cross was of mosaic inside.
[254] See Salz., plate x. The right-hand side is a section through one of the domical vaults, and the left through the barrel-vault which connects two domical vaults. The plans, plates vi. and vii., and the section plate xi., show how close some of the columns stand to the piers, to which they are joined by small barrel-vaults, intersecting the domical vaulting.
[255] A mistake for vertical circles; the large number of cisterns where the vaults are uncovered make this certain. See our p. [221].
[256] Salz., plate xxiv.
[257] Salz., plate xxiii.
[258] See figure in Salzenberg’s text.