For calibration of a psychophysical visual acuity test, it is indispensable that a reference test with Landolt rings be constructed accurately in accordance with the EN/ISO 8596 [9] and commercially available tests. However, we report here that commercially available near vision charts show inaccuracies in print quality and aberrations from the nominal values for the height, thickness of the lines, and size of the openings of the Landolt rings. Although the results obtained at smaller print sizes (smaller 0.1 logMAR) will not be comparable with angular visual acuity, these aberrations might not be of high relevance for routine clinical work but must be taken into account for research, as these might lead to artificial ceiling effects. The extent of these inaccuracies excludes such near vision charts as reference tests for a calibration.

For this study, we microscopically pre-selected the Landolt ring near vision charts with the best print quality. In other near vision charts, the smallest print size was only 0.0 logMAR for a test distance of 40 cm, and therefore not applicable for calibration. Screens of electronic devices could not be used because of an insufficient resolution at small print sizes [15].

Aberrations of offset print are due to an outflow of the color when it is pressed on the substrate (e.g., paper) between the rubber blanket cylinder and the pressure cylinder [16], or in case of screen print when the color goes beyond the edges of the stencil on the mesh [17]. Both aspects cannot be fully avoided and cause oversizing of lines and of the optotypes, as well as frayed or blurred edges, and rounded edges of the opening. For the RADNER Reading Charts, the print quality has been optimized with the printing company for text print. However, offset print is a technology that has been developed for text print, and thus is not exceptionally accurate for graphically constructed figures such as optotypes. We therefore made an attempt to optimize a set of Landolt rings for near vision charts by means of microscopic measurements and graphical modification. The custom made Landolt rings were constructed by a graphic designer. Lines of 8 Landolt rings were adjusted to the mathematically calculated nominal sizes from − 0.3 to 1.0 logMAR. These sets of Landolt rings were then printed by the printing company and the height of the Landolt rings, the thickness of the lines, and the width of the openings were then measured for every print size. The sizes were modified according to differences to the mathematically calculated values (e.g., when a parameter was too big, it was made smaller; when the opening was too small, it was increased). Then, the Landolt rings were printed again. This procedure has been performed in order to approach, step by step, the calculated nominal sizes on the print. Here, we used the set from the fourth round of modification (further modifications will follow). This level of improved accuracy gave values that were very close to the mathematically calculated nominal values for the diameter of the Landolt rings and the thickness of the lines. However, there were still openings of the Landolt rings that were too narrow and outside the 5% tolerance (− 0.2, − 0.1 and 0.1 logMAR). Although we believe that we can get even closer to the nominal sizes, these results confirm that manufacturing Landolt rings representing an appropriate reference for the calibration of near vision charts is technically much more complex than it is for distance vision, because the impact of the typical inaccuracies of print, such as oversizing, is much bigger in relation to the optotypes´ sizes (optotypes for distance acuity tested at 4 m are tenfold larger than those for near vision charts for a test distance of 40 cm).

Another concern is that reading charts and reference tests must cover a sufficient range of geometrically progressing print sizes. To avoid ceiling effects, the smallest print size for near vision charts with single optotypes should be − 0.3 logMAR, and for reading charts, − 0.2 logRAD (for a test distance of 40 cm; logRAD = reading equivalent of logMAR) [18]. However, the sizes of the optotypes of many near vision charts with Landolt rings only extend down to − 0.1 or 0.0 logMAR. Even the C-Test does not go down to − 0.2 logMAR. The smallest print size of this test is − 0.15 logMAR for a test distance of 40 cm. Although it might be clinically reasonable to use half of a geometric step at such a small print size, this approach and the limited sizing are not appropriate for the calibration of other tests.

In addition, we could not find a near vision chart that was entirely in accordance with the layout as required for the EN/ISO 8596 [9]. To avoid crowding effects, the EN/ISO 8596 requires that the horizontal and vertical distances between optotypes increase with smaller print size (e.g., from 0.0 to 0.4 logMAR these distances are two times the diameter of the largest Landolt rings displayed; for smaller sizes than 0.0 logMAR, they are three times the diameter) [9]. However, crowding is a phenomenon that is typical when words and sentences are read from reading charts [3, 4]. This situation raises two questions: (a) How should the Landolt rings be arranged on a near vision chart used for the calibration of reading charts? and (b) What is a calibration with the Landolt rings intended to compare?

Reading and recognizing details of optotypes do not represent the same visual tasks [11]. Colenbrander recommends that “reading tests that show how well the patients function, should not be ignored in routine medical practice. Because the goal of all medical interventions ultimately is to improve the functioning of the person” [11]. Furthermore, tests of reading vision tell us how the patients perform, and they represent the functional vision needed to accomplish the visual necessities of everyday life, whereas single optotype vision evaluation such as that carried out with Landolt rings only investigates a retinal area that is smaller than 1 degree from the foveal center at a visual acuity of 1.0 logMAR—what Colenbrander calls detail vision [11]. In addition, for clinical purposes, reading acuity has its own place in evaluating near visual performance and also its own notation. This situation is reasonable, given that single-optotype distance acuity has been shown to be a poor predictor of reading performance [14] and that in several eye diseases, it is typical that distance acuity and reading acuity differ significantly (e.g., age-related maculopathy and amblyopia) [12, 13]. Therefore, it seems evident that, independent of the technical limits of calibrating reading charts with Landolt rings, their usefulness for reading charts can also be questioned psychophysically.

It should also be mentioned that once calibration has been done based on the specifications outlined by the ISO/TR 19498:2015 [6], the x-heights of a reading chart will have to be adjusted according to the differences determined by this psychophysical study. However, such studies compare the means and standard deviations obtained from two geometrically progressing tests expressed in log units. Converting differences in geometric means into the linear system of x-heights in millimeters is complex and complicates manufacturing. We further report that the investigated Landolt ring near vision charts cannot even fulfill the criterion of a 5% tolerance because of the technical limits of print quality [4, 6].

Nevertheless, as long as a calibration with Landolt rings has not been realized, the ICO recommends the use of the x-height of a font and calibration of this height according to the visual angle of 5 min of an arc, as specified for the related test distance of the geometric progression [4].

The present study shows that near vision charts with Landolt rings do not achieve a level of quality sufficient to meet the premises for calibration. Since the x-height related to the visual angle represents an already well-recognized standard that has been demonstrated in many studies to produce reproducible and reliable results for reading acuity and other reading parameters [2, 3, 5], it seems to be obvious to retain this definition as the standard for reading charts in Latin script.