GUIDE FIGURES
The acuity of the eye, ability to separate dots, is 1 minute of arc. The characters on the 20/20 (euro 6/6) row of an eye chart map into a 3 wide by 5 high grid where the size of the elements is 1 minute of arc. The ability to judge the relative positions of dots is the Vernier acuity of the eye, which is 5 seconds precision (repeatability – think of a good marksman who’s sights need calibration) and 10 seconds accuracy (closeness to truth– think of a poor marksman who’s sights are calibrated).
1 A 0.005-inch dot viewed at 18 inches subtends 1 minute of arc, [assume a circle of radius 18 inches. It has a circumference of 113 inches. Then 113/(360*60) give 0.005-inch.
2 A 15-inch (12 inch wide) monitor with 1000 wide pixel resolution gives a dot size of 0.0120 inches 2.4 minutes (0.3 mm).
3 A printer with 1200 dpi resolution gives a dot size of 0.0008 inches give 0.16 minutes of arc, 9.6 seconds of arc (15 times better).
4 1-point size is 1/72 inches is 2.8 minutes of arc.
5 12-point characters printed should be 34 minutes of arc high compared with 5 minutes of the 20/20 row of an eye chart.
NOTES
1 My recollection from work I did back around 1977 at Reuters was that there is a significant (10% to 20%) difference in legibility between Times and Arial characters displayed at about 15 minutes of arc. The reason is that, given a fixed size of character, the serifs take up space and reduce the size of the core shape of the glyph, i.e. the part that conveys the information. And, as I recall that difference in size was about 10% to 20%. However, when characters were viewed at normal reading size 28 to 35 minutes (10 to 12 point size) there was no significance in legibility.
2 Again from early work; Time types, those with serifs, are more readable. There was a distinction made between legibility, readability and comprehensibility. Readability is in part about maintaining the unity of the characters that form words and that’s what serifs and proportional spacing does. In addition I feel that large masses of heavy Arial font pull the words together into a mass of squiggly vertical lines, I just can’t see the words. (a slight exaggeration)
3 Tests we did at Reuters used tatistascope (a projector with a shutter) to flash large characters on a screen at progressively reduced time, but not size. His showed a marked transition from subjects recognizing all 5 random characters to only 3 of the 5 as exposure time reduced. The exception was where the 5 random characters included common sequences such as ere and air. This is useful knowledge for the design of rapid perception in critical display systems such as aircraft instruments.
4 I read the thread on hinting referenced by Bhikkhu Pesala above and went ahhh gocha. My understanding is that there are two general classes of algorithms.
One class takes a matrix of pixels and multiplies each be some weasel number and adds them together, This class has applications in digital sample rate conversion, TV line standards conversion, anti-aliasing, image edge detection and I presume gray scales. It is used in digital photo editing applications. This class of algorithms is ubiquitous and I am familiar with it.
The other class is grid fitting, which I am not familiar with and it is difficult to search for but [“grid fitting” polygon] produces some results. This class seems to have applications in the general area of image recognition and characterization, e.g. recognizing whales by processing images of their tails. This class is also likely to be concerned with extracting and presenting fine detail in graphics such as geographic maps. Hence the “ahhh gocha” because the referenced thread makes the point that gridding improves characters that include fine detail. I would confirm that in the comparison I did above, the hinted/gridded fonts had finer detail.
- I guess this means that manual / auto hinting is important for two reasons. First it improves the detail in fonts and thereby the discrimination legibility. Second, many people, like me, will not have enabled character smoothing and so for general acceptability one should hint the fonts.
John Cork