This leads us to consider an hypothesis which embraces a purpose more aesthetic than symbolical. Put shortly, this hypothesis is, that the bend was designed to produce an artistic illusion--a perspective effect whereby a building appeared to be longer than it actually was. Thus, if one of the side walls of a church with a “weeping chancel” be viewed from the Western entrance, an impression of greater length and of undefined distance is received by the spectator. Again, where a rood screen exists, the bend in the wall, not a pleasing feature, considered separately, would be concealed, but a change in the direction in a lofty roof would still produce an illusion that the church is indefinitely extended. Moreover, as the beholder caught a glimpse of a portion of the sanctuary window, which was often richly ornamented, both in form and in colour, the beauty of the vista was much enhanced. On this hypothesis, the screen, instead of being the immediate cause of the inclined axis, was an accessory to a complete design, a feature of which the effect was foreseen and provided for. To round off this question, it should be added that not many relics of screen-work exist which belong to a period earlier than the fourteenth century. A few specimens of thirteenth century work remain, and doubtless others were swept away at the Reformation, but most of the examples which have come down to us are assignable to the fifteenth century[589]. In short, the middle stage of the development of rood-lofts seems to correspond roughly with the period in which deflection is most commonly observable. This coincidence does not, indeed, solve the riddle, since each of the opposing schools may produce it as testimony.