Which seems truer to you?
A. In optical-size cuts, for smaller settings it makes more sense to gravitate to single-story /a and /g forms, because their simpler topography and larger counters make them more readable.
B. In optical-size cuts, for smaller settings it makes more sense to gravitate to double-story /a and /g forms, because their unique structures are better at differentiating them from other letters.
(For those of you who predictably want to reply "it depends," on what does it depend?)
0
Comments
I like the the shape of Book Antiqua (see the cursive) for readability. Bodoni is a no-go.
"gg" is relatively common in English, and I have seen several books where you can spot "gg stamps" all across the page immediately
I usually give \æ two stories, notably in the italics with a one-storey \a.
I like the way that the loopy bowl of \e flows easily into the bowl of two-storey \a.
It’s story in the USA, storey in the other main varieties of English.
Human visual resolution is somewhere near 120 lines per cm, or ~300 dpi. At 300 dpi 2 pt (US Printers Points) equals 8 pixels. If the output device can only render pixels on or off in a strict rectangular array, an \E needs a height of 5 pixels and \M, \W a width of 5 pixels. Also 1 pixel is needed for spacing.
5 x 5 is used for eye charts to measure visual acuity. There is the old one by Snellen using serifs and the more modern LogMar (sans serif). They have only Latin capital letters. There are attempts to include the full ASCII range like http://mkweb.bcgsc.ca/snellen-optotype-font/. Optician Sans is a font for LogMar.
8 x 8 was usual in the old days for bitmap fonts. This would allow 2 pixels for descenders, but they shifted up \g and \y. It also allowed a bold version. Looking into my collection, they all use 2 counters for \a to avoid mismatch with \o and \c, and open tail for \g:
3 x 3 is the smallest and would allow 1 pt. Maybe it's readable but not legible if some unknown words need disambiguation of single letters:
IMHO white space is important. If a Swiss (sans serif, monoline) font is just scaled down, the dot of \i can connect with the stem and then looks like \l. Had this problem while reading a scientific PDF using a narrow "Swiss" and had to zoom twice to read some unknown scientific terms.