Such is the one alternative. The other, to which I myself incline, is that they did not need to fetch the pillars at all, but utilised them on the very spot where they originally stood. According to this view we have here an example, unique in Britain, of Roman work _in situ_. The very arcading which we see I take to have stood north and south of the central hall of some large Roman mansion. Such a mansion usually contained an oblong central hall of this kind (often roofless), with a peristyle, or cloister, on either side opening into it, a portico at one end, and a smaller _tablinum_ or guest-chamber at the other. Lanciani has pointed out how this structural arrangement suggested the nave, aisles, porch, and chancel of the earliest ecclesiastical edifices at Rome.[169] The same suggestion may have influenced the builders of Ickleton Church to utilise this old Roman arcading, roofing in the enclosed space, but with a clerestory to prevent too great loss of light. If this view is correct the narrow north aisle probably represents the width of the original peristyle. [Footnote 169: See my _Roman Britain_, p. 266.] The south aisle is far wider, as wide indeed as the nave and north aisle together; and one asks why the fourteenth century architect planned his work so very unsymmetrically. The answer, I think, is to be found in the remarkable architectural development of the steeple. The piers of the tower are, as I have said, unmistakably Norman, but upon them are set, quite unconformably, arches at least a century later in date. The tower is pierced by these arches on all four sides, and was evidently meant as the centre of a cruciform church with transepts. For some reason this Norman plan was never completed, but it is very probable that the south wall of the church marks the limit to which the transept (which may have been actually begun) was meant to extend. The church has also later features of interest. There are some good mediaeval seat finials, shaped with the axe and bearing grotesque figures, musical instruments, and symbols; the word ORATE being decipherable upon one of them. The rood-screen is fifteenth century, and is placed across the eastern arch of the tower, with no trace of there having ever been a rood-loft.