I wonder if some readers might like to know of a technique which I use, which I devised myself, which, for purposes of discussion might perhaps be called "mathematical hinting", though whether it is hinting as such is something which perhaps needs to be discussed: maybe it needs to be called "mathematical pseudo-hinting" instead.
I devised the method for use with my Quest text font: I have used the method successfully with other fonts too, though my experience is of using them on a Windows 98 system with an 800 pixel wide by 600 pixel high display where 12 point is 16 pixels. So maybe on other systems the effect is not so good, I do not know.
A copy of the Quest text font is available from the following web page.
http://www.users.globalnet.co.uk/~ngo/fonts.htm
Quest text is designed on the basis of being as if a calligrapher were drawing the letters with a pen 256 font units wide by 168 font units high, upon guidelines on a 256 font unit by 256 font unit grid, locking a line edge to a grid line wherever possible.
I imagined having contours made out of wire and imagined placing them onto courtyards paved with square slabs. That is, each paving slab in the courtyard is to be either black or white in its entirety, whether it is black or white depends upon whether the area of the particular paving slab is mostly inside or outside the black area of the contours. It is not only the shapes of the contours yet how they are arranged in relation to the paving slabs which is important.
The slabs are 128 font units for 12 point, 85.3 font units for 18 point, 64 font units for 24 point and so on. So I used lines 256 font units wide and 168 font units wide wherever possible, locking the lines onto the edges of the square paving slabs. The 168 aligned on the grid gives 128 plus 40 at 12 point, so a black pixel and a white pixel. At 18 point, 168 font units aligned on the grid gives 2 black ones and a white one. And so on for various sizes.
The figure of 256 font units was selected on the basis that Microsoft Word 97 and WordPad on the Windows 98 platform which I am using uses 16 pixels for 12 point, so the 256 font units for the grid spacing is from 2048/16, where 2048 is the number of font units in whatever point size is being used to display the font.
The intention of this design strategy was to produce a font which looks very clear and distinct at 12 point and reasonable at 18 point yet at larger sizes shows a distinctive style. So I rounded some of the corners with curves 128 font units in curvature distance.
So, I wonder if the symbols mentioned in the original question would benefit from being drawn so that vertical lines are at x equals a multiple of 256 font units, that is 256n font units, where n is a whole number such as 1, 2, 3, 4 and so on, or at x equals 256n + 88 or x equals 256n + 168. Similarly for the y values of horizontal lines. I have found that symbols and letters look good at 12 point, 18 point and 24 point using this method, though 14 point can look awful in some cases!
I used the technique in my Style font, making the lines 80 font units wide, always with one side of each line aligned on a 256 font units measurement. So, from 176 font units to 256 font units horizontally, 432 font units to 512 font units and so on: that is, from 256n -80 font units to 256n font units. Vertically, I went from 0 font units to 80 font units, 256 font units to 336 font units and so on: that is, from 256n font units to 256n + 80 font units.
I hope that this helps. I would be interested to know if this technique works for you and whether the way that I have used it works on displays which are other than 800 pixels wide by 600 pixels high.
William Overington