the vaulted apses had been designed too low for the statues in them: “If the goddesses wish to get up and go out, they will be unable to do so.” The first half of Apollodorus’ criticism is unjustified: Hadrian was designing a Greek temple, not an Italic one. About the second half we cannot judge, for certain, for brick stamps show the apses to belong to the fourth century reconstruction, but the proportions, as we saw, are Hadrianic (Fig. 11.7). The temple was set in the midst of a forest of sixty-six columns of grey granite. When it was re-excavated in 1932, some of the columns were re-erected; the positions of others were ingeniously marked by clumps of shrubbery trimmed to the proper shape. The excavators found under the pavement an octagonal room interesting in itself, and significant for its place in Roman architecture. The level at which it was found is lower than that of Nero’s Golden House. (Hadrian’s temple was built in the grounds of what had once been the Golden House; the reader will recall the twenty-four elephants needed to move the colossal statue of Nero and make room for the temple.) The octagonal shape appears in the dining room of the Golden House itself, in Domitian’s palace on the Palatine, and in a room in the Small Baths at Hadrian’s villa (O on the plan, Fig. 11.2). The cupola of Nero’s octagonal dining room, together with its lighting through a hole in the roof, reappears on a grand scale in the Pantheon. This is what we mean by saying that Hadrian adapted to his own new architectural language the vocabulary of pre-Neronian, Neronian, and Domitianic buildings. Here once again modern archaeology illuminates the development of Roman architecture by demonstrating and dating the classical use of new things in old ways, and old things in new. [Illustration: FIG. 11.5 Rome, Temple of Venus and Rome, Gismondi model. (F. Castagnoli, _Roma antica_, Pl. 27.2)] [Illustration: FIG. 11.6 Rome, Temple of Venus and Rome. (Castagnoli, _op. cit._, p. 85, Fig. 2)]