[Text reproduced from a reply on Typophile in March 2005, as was]
(Apologising in advance for skipping over some things with insufficient explanations]
Paul, there are essentially no resources online for what you’re looking; Yannis’ papers have been written primarily from the perspective of monographs to a standards body, and pretty much anything else will have its roots in classical or biblical studies – an important but narrow perspective, and also one which tends to colour the authors’ approach. You should note that a lot of the information online on personal sites is not trustworth y on this subject: often inaccuracies, misleading statements, bias, or incomplete information are found alongside fairly ‘innocent’ passages.
There are very few print resources; my article for which Chris kindly provided the plug is getting a bit long in the tooth now. The latest text touching on the subject, albeit not deeply, is Microsoft’s Now read this on the ClearType fonts. There are other texts that you should have a look at, such as John Bowman’s Greek printing types in England…, Victor Scholderer’s catalogue of the British Museum 1927 exhibition, and a few articles in the Greek printing types: from tablets to pixels volume (primarily John Lane’s article); I note items you may reasonably expect to find in a decent university library, or a better public one.
The reason for both the potential for inaccuracy in online resources, and the paucity of print resources is simple: the Greek script is quite difficult to design well for: not only becasue of its extremely strong ties to calligraphic models, but becaus these very models are very different from those proposed by any western calligrapher or theorist. Throw in a complicated, changing cultural environment for the script over almost three millennia and you begin to get the picture. (It takes me three to four hours of talking around a shopping-trolley full of material and many slides to give my ‘introduction to designing Greek for non-Greek designers’ to the MA students…)
So, cutting to the chase, I will try to address the points rasied from the general to the specific. Firstly, Hrant is right in pointing out a fundamental error in Brent’s statement: that you may be designing a Greek to package together with a Latin does not mean the Greek is in any way subsidiary. Even if you are designing a Greek to go with a pre-existing Latin, your goal should be to produce a typeface that would be internally consistent, i.e. be perfectly acceptable to well-informed native readers irrespective of the Latin. To be more precise: we should seek the equivalence of typefaces across scripts at certain typeface-wide levels and mid/macro-typographic levels, not glyph or sub-glyph level elements. (I’m going to skip writing further on this specific point because it will take me off-topic to the general problem of approach to cross-script designing.)
Brent is right is separating mathematical symbols from Greek: disregard anything you see in such typefaces as irrelevant to Greek typography. (But got wrong the note about alternate forms; see further down.)
If you are developing Greek at all seriously, you will need to tackle polytonic at some point. All texts prior to 1982 (classical, Byzantine, Ottoman period, and post 1830) are polytonic, in addition to the ecclesiastical. Post-1982 only periodical publications and stuff off DTP bureaus is monotonic; all ‘serious’ prose and all poetry is in polytonic, as well as many academic texts. The reasons for this are multiple, but – more importantly for this discussion – what we may call ‘polytonic’ could be anyone of at least five different combinations of character sets, grammatical rules, spelling and hyphenation conventions. [Cut here long paragraph on misinterpretations of the character/glyph model in Unicode.]
However, whereas OpenType may make polytonic development significantly more practicable (e.g. on managing composite glyphs and spacing/kerning) it also imposes on the developer the problem of encoding in the system that is the font in conjunction with the text engine significant behaviour specification that in the realm of ‘dumb’ T1 or TT remained in the user’s head. For example, I’m only half-joking what I say that some very serious people are now of the opinion that ‘polytonic case conversion’ is penance for past sins.
In the relationship of the two cases, the Greek typographic script takes the principle of different lettering roots already present in the Latin and magnifies it. About half of the capitals are identical to the Latin: it is absolutely fine to just duplicate the glyphs into the Greek character slots; the rest of the Greek capitals range in difficulty, the main parameter being the degree to which the forms can be adapted to the particular typographic style. On the other hand, it is wrong to say that Greek small letter have any form in common with the Latin: this is a misconception borne of poor research, bad design process, and the ease with which designers use a copy-pasting approach when a strong theoretical or experiential frame is absent. (Evidence for this abounds from Gill and van Krimpen to the very latest releases; in the earlier cases it was due to Morison’s later abandoned theories; in the latter to a lack of exposure to original material and a coherent narrative for the script.) This is not to say that some elements within glyphs cannot be identical across scripts – but if they are they should be so because there is a consistent rationale behind it, rather than the expediencies of design processes.
Re secondary fonts in Greek: the practice of spacing out (or using alternate fonts) began to be eroded in 1910 onwards, when Monotype started marketing Series 90/91/92 as a family. The two practices coexisted for some decades, mainly depending on how a publication was produced. For example: I’ve got next to me a late-1950s History of the Greeks by Karagatsis (then still alive) in Monotype Series 192 which does not have a secondary font, based as it is on Scholderer’s New Hellenic, and a 1937 edition of Aristotle’s Constitution of Athens which uses the inclined Series 91 for emphasis within the upright Series 90. In fact, throughout most of the previous century this pattern of spacing for emphasis only survived in texts typeset with typefaces like the new Hellenic or Porson, for which no secondary fonts were developed. Any successful text typeface designed in at least the last seventy years will have had a secondary font, in almost all cases a [differently] cursive or inclined one. The issue of the style of the secondary font is quite interesting in Greek (and a rather large topic in itself, touched on briefly in the MS book).
I am not sure how to interpret Brent’s assertion that ‘Greek characters really don’t have ascenders (although there are a few tall miniscules that reach what we’d consider ascender height)’. In Greek typographic forms there is a clear horizontal banding between the baseline and the kappa-height (the equivalent to the x-height), and typeforms that ascend as well as descend beyond this band. Call it what you will, but I see this as ascending and descending elements. Also, the inference about the relative x-height (kappa-height) being larger is puzzling, since it is exactly the opposite that we often see in typeset texts, because of the space occupied by the diacritics.
The point made about the angle of the monotonic tonos is somewhat misleading: it should never be vertical. All Greek reference works (including the original 1981 text passing monotonic into law) state unambiguously that the tonos is identical to the acute/oxia, therefore inclined to the right. Anything else you see is either a bad hack of an older polytonic font, or a badly designed contemporary one. It is possible to vary the angle between the monotonic tonos and polytonic oxia within a typeface, but both should be clearly strokes inclined to the right.
The way Brent writes about alternate forms of letters can be read to imply that the alternate forms are evident in mathematical typesetting only, which is not strictly correct. The beta, theta, and phi both forms are very much in use both in writing and typography; the alternate pi is used in writing but not typography; and the alternate kappa in writing and typographically depending on the style of the typeface.
I’ve just scratched the surface here, and have not at all addressed your original question: ‘how do you go about designing a new Greek?’. Recently I did a survey for all the documents I ‘ve put together over the last eight years while helping to develop Greek typefaces alongside Latin ones: there were about 45,000 words and over a hundred separate documents. I’d love to edit and put all that together at some point for wider circulation, but I can’t see this happening in the next few months. But if you’re in TypeCon seek me out and we can make some sketches.