I made a mistake with the encoding on the charset for Clarissa regular, so the utf8 page is missing. What is the procedure for posting updates and corrections? I do not want to clutter this forum with new postings.

I have also just noticed that the italic is showing signs of its evolution: 'q' has no 'dink' at top right, whereas 'a' and 'g' have dinks. I have tried both styles. The current style should have dinks everywhere, except on the 'b'. 'Dinkless' is cleaner, but tends to make the relevant letter shapes a bit vague, I think. Comments on this would be welcome.

Here is the full character set for Clarissa Regular: www.adgie.f9.co.uk/typophile/charset-Clarissa-Regu.pdf

I will leave the inconsistency in the italic for now.

Very balanced, if also conventional.

The two main things for me:
- The "Q" is indeed way out there.
- The head of the "R" is too small.

Spacing is good. The blank space seems slightly wide.

The Italic I'm unsure about.

hhp

Just wanted to say that for your first attempt at a complete face, it looks fantastic! Some us other amateur type designers need to step our game up (meaning me). :)

Here is another sample, with a narrower word space, and reworked 'Q' and 'R'. Note that there is no kerning yet.

www.adgie.f9.co.uk/typophile/critique-wordspace.pdf

The word space is good.

"r" right: too loose.
"i" right: too tight.
"a" right: too tight.
Kern resultant issues.

The "U" should be wider.

I would complain about the tail of the new "Q" being too long, but it is a glyph typically afforded some flair, so...

hhp

Thanks for all the detailed help, Hrant. Just a few words before I work on the stuff you described.

I agree about the 'U'. The all caps sample really shows it up.

Letter 'a' should indeed have a more space to the right. The tail at bottom right takes up some whitespace relative to, say, 'n', so there must be a bigger distance to the main stem. The distances to the main stem in 'a' and 'n' are identical at present. I forgot to adjust the spacing for 'a' after adding the tail: it originally did not have a tail.

I feel the right sidebearing of 'i' is less of an issue: I added 26 units to the main stem spacing to compensate for the tail. If the right of 'i' is too tight, I would expect 'l' to have to same problem, as the tails are the same. I will try some small variations.

On the subject of the tail on the 'i' and 'l', I did this partly to provide more space around these glyphs, to reduce the "logjam effect" that occurs in words like "filling", which I consider to be a defect of most sans serif fonts compared to serif designs. Do you think this works? A more common approach is a left facing serif on the top of the stem of the 'i' as well as a tail on 'l', but this produces different stem spacing for 'il', 'll', and 'li'. Trebuchet does this. I will try some sample text in Clarissa and Trebuchet to compare.

With the 'r', I am aiming to make the spacing halfway decent without kerning, so the font will work in a word processor. Looking at other fonts, the policy seems to be to space 'r' correctly next to a stem, kern it closer to rounds, and increase the space next to 'f' and 't'.
Like the 'i', I think the adjustment I need is small.

Anyway, I think it is about time for some kerning, so it might be a while until my next update.

> to reduce the “logjam effect” that occurs in words like “filling”

But haven't you replaced the logjam* with an over-crowded fishing boat? :-)

* BTW, don't foget to kern the "gj"!

> I am aiming to make the spacing halfway decent without kerning

Here's the (generally ignored) twist to this though:
People who would use a typesetting app that doesn't support kerning (by default) are not the type of people to mind slightly off spacing anyway! So you can save yourself a lot of extra kerns by making your base spacing more "aggressive". This is especially relevant in some non-Latin scripts, where such a technique (not minding the occasional touch when the kerning goes unused) can actually save you thousands of kerns.

hhp

Hrant, I think I know what you mean about "over-crowded fishing boat", but I am just guessing. The ticks on the 'l' and 'i' can pile up, and be distracting, much like a sequence of closely-spaced plain verticals. In combination with 't', we can really go fishing: e.g. "little". So I have just exchanged one distraction for another. Having said that, I think my distraction is prettier :-)

Concerning the kerning of 'gj': well, of course! In the spirit of test word nerdism, I searched for all occurrences of 'gj' in the American English dictionary that comes with the Unix spell checker, ispell. Just three hits: 'logjam', 'jogjakarta', and 'Kwangju'. No hits on 'qj'. That's saved me a bit of work.

Back to serious stuff.

By 'agressive base spacing', I think you mean base spacing that is motivated by making kerning easier or better. For a Latin font, this could mean spacing 'r' correctly with a round to the right, so we do not need to kern all the accented variants of 'ro', 're', etc. There would be less work kerning next to verticals, as there are fewer accented variants of such glyphs. My thoughts on this are that a smaller kerning table is not much of a benefit in itself. A kerning-aware application will work just as well kerning 'ro' and not 'rn' as it would kerning 'rn' and not 'ro'. However, a non-kerning application will produce a better output if the base spacing is designed to work reasonably well without kerning. So why not add a bit to the utility of the font by making it work better in a word processor?

I suppose that reducing the number of kerns could make some improvement in real type quality, in that the type designer is less burdened with the drudgery of producing excessive kerns, and can concentrate on more important things in the design. I get the impression from what you say that this is a real issue in some non-Latin scripts. Is it such a problem in Latin fonts? I only have a little experience with kerning. I had a go at the basics, with accented variants, using class kerning. I stripped it all out, because I was still messing with letter shapes, proportions, and spacing.

Anyway, I have rambled on enough, more than likely completely missing the point. It is a hot, thundery night in Brum, and I have nothing better to do.

> No hits on ‘qj’

Feqjakuqe. Some Albanian name.
Thanks to "Making the Alphabet Dance" by R Eckler.
:-)

> My thoughts on this are that a smaller kerning table is not much of a benefit in itself.

I have to agree. It just seems more elegant.
But yeah, the relevance to Latin is minimal, almost nil even.

> and can concentrate on more important things

Actually, yes. Since a font is never really finished.

hhp