Is this board broken!?

I'll post this yet again, in yet further abbreviated version:

http://www.gutenberg.org

Thousands of copyright free texts.

There's a difference between gobs of "real" text (certainly a very useful thing), versus *test* strings/paragraphs. In terms of test text, Andrew might benefit from a number of things, including possibly this type of thing:

http://www.microsoft.com/typography/links/news.asp?NID=2454

BTW, I've settled on calling it "superpangram" instead of "metagram".

hhp

Andrew, don't get hung up on short phrases, or kerning pairs - spacing first, kerning after. Just get some decent runs of text and go from there. It doesn't actually matter whether the font is for text or display usage, a decent run of text will show up the character and spacing anomalies - short phrases won't.

If you want to go all the way - in an abstract manner - generate every character next to every other. Again the anomalies are apparent.

> spacing first, kerning after.

That will indeed cover you 90%, at least in the mainstream.
But for that last stretch of quality, they really have to be done in parallel (as much as feasible). Certainly, few designers do that, but that's only because even fewer users care.

BTW, kerning even affects the letterforms themselves, not just the sidebearings.

> It doesn't actually matter whether the font is for text or display usage

?
I think it matters a huge deal.
Especially when you consider, for example, that cap-cap setting is much more common in display. But there are in fact deeper differences between display and text, even in spacing/kerning.

> generate every character next to every other.

That's of course completely impractical.
Which is exactly why you need *both* test strings and long "real" text to arrive at good results without spending 20 years on it.

So, Andrew: it's not that simple. :-/

hhp

Andrew, start with Tracy's system:
http://www.linesandsplines.com/tracy/

If/when that runs out of steam for you (different people have a different tolerance/appreciation for quality), you have to decide how much further to take the "quatitative" approach. I personally take it very far - like in assigning spacing values to serifs.

Depending on how far you take quantification (noting that it has to end somewhere), you'll be left relying on the "eye" to different degrees. The more you rely on the eye, the longer it takes, and the more error prone it is. But, again, it's not something to be avoided, just balanced judiciously against the "math".

Rely on real text only when you can't go any further with test strings. There are many reasons for this, not least that your font might be used for a language you don't even know exists.

hhp

Welcome to the forums!

Try this: http://typophile.com/downloads/

We can fill both your bills.
You say 'jump,' we say 'how high?'

Please tell us how they work for you.