Reading Comprehension

That a flat surface can be made to represent three-dimensional space, with apparently coherent depth and proportion, was, before the fifteenth century, an unsolved problem. Earlier European painters had used various conventional devices — overlapping figures, diminishing scale, oblique angles — to suggest spatial relations, with results that often persuade the eye but rarely cohere into a consistent geometry. The technique that came to be called linear perspective, formalised by Brunelleschi and Alberti in the early fifteenth century, supplied a consistent geometry: a method by which all the depicted lines that, in the real world, would have been parallel and receding could be made to converge on a single point on the picture surface, producing an illusion of depth that ordinary perception accepts as faithful.

The technique's adoption is sometimes presented as the moment painting acquired truthful representation, with earlier work treated as a slow groping toward what perspective finally delivered. The presentation is misleading in two ways. First, perspective produces an illusion calibrated to a particular position in front of the painting, the so-called viewpoint; from any other position, the geometry of the depicted scene is, strictly, wrong. The eye accommodates the discrepancy without noticing, but the geometry does not become the world simply because the eye has agreed to read it as such. Second, the earlier devices were not failed attempts at perspective; they were different conventions, with different aims, whose displacement was a choice rather than an inevitable discovery.

A second consideration concerns what the technique does to the depicted scene. Linear perspective imposes on the picture a particular spatial logic: the world it depicts is one in which the viewer stands outside the scene, looks into a coherent depth from a single point, and is invited to forget that the picture is a picture. The convention of perspective, in this sense, is not the discovery of truthful representation but the production of a particular kind of viewer — one whose visual relation to the depicted world is that of the spectator at a window. Other conventions had produced other relations: the viewer of a mediaeval altarpiece is not a spectator at a window but a participant in a sacred scene whose figures look back.

To say this is not to dismiss linear perspective or to call its adoption an error. It is to call it a particular innovation with particular consequences, whose displacement of earlier conventions was a choice that has come, over the subsequent centuries, to feel like nature. The strongest defences of the technique have been those that recognised what it gained and at what cost; the weakest have been those that called it the discovery of how representation must work.