Many scholars have drawn their attention to the question – did Jan van Eyck have perspective methods in paintings? Perspectival configurations in works of art have achieved a greater consensus by attempting a greater degree of accuracy. Nevertheless, art historians continue to disagree on the perspective methods of a number of major paintings. They disagree not only on the kind of perspective should legitimately be found in a painting, but also on the directions of the lines that are involved: a curious occurrence, since the paintings are static objects, and only one direction of a line is possible.
Indeed, a closer inspection of the Arnolfini Portrait reveals many more lines than previous analyses have shown. In such reproductions – I have found it useful to number the lines – so they may be more easily compared with other theories. The ceiling has eleven beams, and since each is seen from one side except the central one, there are theoretically thirty-two lines, not eleven as drawn by Kern or seven by Collier. Of the thirty-two, I could only see twenty-eight. The left wall and window, including its saddle bars, yields nine major orthogonals (drawn in the plate, in which I have given numbers 3, 4, 5, 6, 7, 12, and 15 for reference), as opposed to the two usually drawn. It is not including six lines implied between the rows of bottle glass, one (numbered 4) in the grating, and twenty-three more in the very short segments of stained glass along the far margin of the window-frame. Therefore, the painting got a total of thirty-nine orthogonals. Nine clearly visible lines recede from the near shutters to a lateral vanishing point outside the picture (numbers 1-9 along the left margin). The further shutter yields three more lines receding to a second lateral vanishing point (shown, unnumbered, above the brim of Arnolfini's hat). The rug contains five orthogonals (three of which are drawn); and the canopy of the bed has four orthogonals (all are drawn and numbered 1-4; the tassels are drawn very neatly in three almost parallel lines).
Even Collier’s reconstruction is considerably idealized, for example, in the first six lines between the floorboards, which he draws receding to a single point. Some lines in the original are quite accurate (the third floorboard line is both long and almost perfectly straight) and others are off by wide margins. The fifth floorboard line, for instance, can be drawn in two quite different directions, depending on whether the segment below or above the terrier is taken as standard. The same is true of lines 6 and 7, and it is interesting that line 5 does not diverge more strongly than the four lines of the border of the rug, which are meant not to look straight. Its two possible directions X and Y differ by approximately three degrees, providing a margin of error that may be used to judge other elements of the picture. “Margins of error” needs to be given several senses, depending on what is being measured. In the references in this article, I use it to denote the angle between two alternate reconstructed lines, each of which is meant to trace a single line on the original. In this sense, the term corresponds to the relative direction of the lines. The tangle of lines can be pared away by discounting shorter segments, those with the greatest margin of error, and those in darkness. In the roof, darkness swallows many of the cross-beams and it is hard to tell their shadowed sides from the shadows they cast on the ceiling. The most reliable orthogonals by these criteria – intersect in four small vanishing areas. Two of the vanishing areas are near the pictures vertical axis and two are disposed at either side.
Limited time Offer
The analysis can help distinguish between the theories we have considered. Since orthogonals like 5, 6, and 7 on the floor have large margins of error, Kern’s thesis that Jan drew to vanishing points instead of to vanishing areas is unsupportable: we simply cannot be sure that lines converge with that accuracy. The same margin of error makes it impossible to say if Jan’s painting contains five, six, seven, or more vanishing points as Collier found because the points are not separate from one another by a greater distance than the margin of error of the individual lines. Collier’s mention of five, six, seven, or more vanishing points did not mean to imply that Jan had some system or rule that required drawing to so many points. The researcher suggests that while the artist appears to have had some notion that straight lines in any planar surface appear to converge, his works reveal that even internal lines within the confines of a single surface do not always align to the same vanishing point. This is not a systematic perspective “system”, but one based upon empirical observation. Nonetheless, it is important to note that, regardless of whether or not Jan intended multiple vanishing points, the inaccuracy inherent in the picture does not permit the conclusion that there exist five, six, seven, or more vanishing points.
To conclude, Jan van Eyck probably had no interest or knowledge of such analytic approaches, and we need not consider he supposed to have any “system” in mind. He accomplished by eye, and with consistency between paintings, a compromise between medieval and Renaissance sensitivities.
Related Research essays