Well put.

Let me be naughty again:

2. Contrast in type designs are bad for the purpose of screen fonts.
...but displays fine with optimized photofonts

3. fractional point sizes are not possible on the screen.
...with a little more thought, yes they are - at least at a nicely faked level.

*** and a lot of people do like plum pudding (the British kind)***

>But proof of the pudding will be in the eating.

I was talking about the C-fonts and Bill's critique of them - not about ClearType itself. I was making the point that until the fonts are shipped and people start using them, it's difficult to properly judge them from a bunch of scans on the Poynter site.

I'm not sure I agree with all the quotes you attribute to Microsoft employees, but a few of them look familiar. 3. is definitely wrong though - fractional point sizes are definitely possible in Avalon WinFX.

Si, I agree with you absolutely that it is difficult to judge type from scans on a website. But the problems with the CT Collection are so egregious that they are perfectly obvious as presented there. I may as well say that I had made up my mind not to post until I had seen the types and the book, until I discussed the matter with Type God A, very familiar with the types, who told me there was no significant difference between the types as demonstrated on Poynter and the types as available in any other form.

In any case, nobody has come back to me and said ' 'Bill, your criticism is wrong. On screen, or in print, the problem you are referring to is not evident. The fitting error you mentioned is not in the font - it is a rasterization mistake on the site, or a printing error in the book.' There's a reason for that. The problems are evident at a glance.

Look, the problems are so great that it is obvious that most of these designers have no grounding in text type vernacular and can only be trusted to make clones of text or fancy display faces.

What is equally clear is the degree to which features in these types impede readability. Let me give just two examples.

Take Corbel. Look at the upsweeping curves at the top of a and r. Beautiful! Just lovely. I love them. I really do. They are aesthetic triumphs.

But they are functional disasters. They impede readability because they draw attention to themselves as beautiful artefacts -- and because their excessive slope and area gets you thinking vertically rather than horizontally.

We see here the traditional mistake of the amateur type designer - love of aesthetic form versus love of typographical function.

Let's look at Constantia. Did someone say contrast impedes readability? Then what are all those contrasty glitches doing on r, f, c, etcetera? Even in print, they are unidiomatic. But in a type intended for reading, they are traffic-blockers.

In so many of these types, you see lovely aesthetic ideas that create traffic jams in the race to read through a word. The very last thing you want to see is an aesthetically remarkable feature. All text type is about function, function, function.

On the other hand, there's the low contrast Dutch type by Bosma -- very safe, very generic, very Lexicon. There is one superbly judged feature I think I like, the ball of the a. But the overall widths of the letters are all wrong. Most are too narrow, giving a choppy, vertical effect which is the very opposite of what you want in text, which must have horizontal flow. The excessively narrow o is one of the clumsiest o's I have ever seen, and I am a dreadful drawer of o's. There isn't enough overhang in the f, a fault which is especially distressing in the bold, and the bold teeters on illegibility because the x-height has been raised too much in that weight. And then, in the midst of all this compression (which is the very last thing you want in a readability font), you have the r sticking out like a sore thumb, because, though lovely entirely on its own, it's too wide to fit in with the rest of the characters.

But do I have to go on? Everywhere in these types, you see a thousand areas for improvement. Everywhere you see that where there is a choice between beautiful aesthetics and pure functionality, aesthetics always wins.

You know, one of Bud Kettler's tests in proofing a font was to study it upside down.

Did anyone do that here?

Well, there's a reason why Courier is a great typeface.

What I see here is a bunch of types that need some expert, thoroughgoing art direction. I see some nice work with promising ideas that needs a lot of firm revision. These look to me like first drafts from the designers. I don't think there has been any substantial critique. I would be astonished to hear that these types had been extensively reviewed by a knowledgeable panel of any kind and sent back for cycle after cycle of revision until they just looked . . . . transparent.

Stylistically, I regard these types as fashion statements, not as functional text type.

Please, Si, just hire someone competent. Someone who just knows how to make a functional type rather than a pretty artefact. Well, you know my recommended list -- and you know where they live.

David says time is running out. I am not sophisticated enough to respond to what he is talking about. But that particular statement resonates for me in the different, simpler sense in which _I_ see that time is running out:

Time is running out for the designers who know what they are doing. There are a few designers over 50 who really know what they are doing. And there is virtually nobody under 50 with anywhere near that degree of skill. Microsoft has done some wonderful services to type by getting some excellent work out of the old pros. Why stop now? Especially at the point where you really need some _excellent_ type? But forgive me for repeating myself. I am not advancing the discussion here. Let some better brains steer the till.

I've responded to David's original post via e-mail. If anyone, other than Bill, is interested in what I said, contact me.


Bill, phrases like 'the race to read through a word' suggest that you don't know the basics of the visual and cognitive mechanics of reading. Your idea of 'traffic jams' impeding reading is quaint but lacks any scientific support. If you want to critique type designs based on aesthetics, or even on whether or not you like the designer, then that is a matter of taste. But if you are going to start criticising typefaces on grounds of readability, then you need to get at least a basic understanding of how we read. You could start here.

John, could you please make a more advanced suggestion? There is nothing in Larson's paper I am unfamiliar with, and I have never subscribed to the bouma theories bandied about by some people on these lists. For one thing, I find it impossible to think of the word 'bouma' without laughing - it is so undignified. For another, well, let's just say that word shape recognition runs counter to the bulk of reading science, and finally, I just have never accepted it because it didn't make sense to me. So I expect we are in perfect agreement on that point. If there is something radically new in legibility research, I would certainly like to know about it.

However, here, I would prefer, rather than discussing general readability issues, to get down to the nitty gritty of some of the issues I have raised, such as the u in Calibri. Can't someone just tell me that what I am seeing is a rasterizing error, or just a weird typesetting mistake? Nobody would be happier than I to find this out. I mean, first I got slagged by Leonidas and Issac for knowing too much because I had the fonts. And then when I said I didn't have the fonts, I got slagged by Simon for not being in a position to judge. Can't win! And tell me, are you still an avid supporter of InDesign's optical kerning? I remember you taking the very strong stand that InDesign's optical kerning provided superior spacing in most fonts, so much so, that you said the application should be allowed to determine spacing, rather than the font's original metrics. Is this still your position?

And to get to some of the issues David has raised, using his point numbers:

0. Could you tell us more about the mall test? What were the C-types compared to? As you describe it, it sounds pretty lame.

1. What is the justification for the OS not to know screen dpi? And in that case, is there a manual setting? (And is there any research on how many hundred people would use it?) How does this contrast with whatever Apple and other systems may be doing?

2. What degree of contrast is bad for readability? Could you be more specific? Is there any actual proportion of thin to thick that has been established? Like, is 1:8 or something like that too much? Where is Times on this scale? Georgia? Bodoni? Low contrast SON Dutch generics? Has any of this affected the CT collection?

3. Could you be more specific about fractional point sizes? Is it that you can't specify them? Or is it that if you do specify them, they will be faked, or ignored, as has so often happened in the past?

4. I have heard for many years that CT will ultimately be good enough for fonts not to have to be hinted anymore. If that is the case, where is Quartz? I have been concentrating so much on the current glories of ClearType that I have failed to notice that my Mac displays don't look that bad, after all. I was kind of surprised. Will there come a point, reasonably soon, when ordinary PS fonts look just great on screen?

5. These terms are incredibly general, but in these general terms, and granting that there is much room for misunderstanding, I would say that rhythm is more important than color for the simple reason that I have always found that types of strikingly uneven colour could be not only gloriously readable but quite attractive -- so much so, that I have sometimes wondered how important consistent colour really is? Could we usefully define these terms more stringently? I realize that type design and readability are too complex to reduce to simple science, but I think we need better definitions here. For example, how do you qualify a type that has strikingly uneven colour but phenomenal fitting and that just works really well? Maybe I know what you're talking about - - - - but maybe I don't?

8. That's hilarious. You know where I think the money should be spent.

9. If that's what they're saying, I don't believe it. They're just kidding themselves. But I do have some experience with designing scientific experiments, and I understand how easy it is to be led astray. And I'm not sure that I understand what you mean when you say '"read" literally' - sorry to be so dense!

Bill, if you are familiar with everything in Kevin's paper, I'm left wondering why you used phrases like 'the race to read through a word'. As Kevin's paper points out, we've known with certainty for a hundred years now that there is no reading 'through' a word, there is no passage from one end of a word to another.

I'll be frank about why I have not directly responded to your criticism of the Calibri u: I didn't read it. I have not read it, and I'm not about to go dig it out. As I've explained, I give up on discussions in which you are involved, especially when you are in full histrionic attack mode, because you just make me very angry. Maybe if you actually want people to respond to you, you might consider the way in which you put yourself across. Every once in a while you manage to say something reasonable, but generally it gets ignored, and not just by me.

Regarding InDesign's optical kerning, when it first came out I experimented with it quite a lot and was impressed. I think the ideas behind it -- Hermann Zapf's ideas -- could indeed enable an application to produce better spacing than that in many fonts. But remember that this was pre-OT fonts with class kerning, when many font developers didn't bother to kern accented characters at all, and many fonts had only minimal kerning and generally less than they needed. In this environment, the idea of an application being able to handle spacing was an attractive one. I think it is less important now and, I found that once one got away from fairly typical Latin text faces the InDesign optical kerning was less impressive. It has problems with many non-Latin scripts, for instance. That said, there are still a lot of poorly spaced fonts out there, and sometimes InDesign's optical kerning improves things. In recent years, I've only used it for specific combinations in display settings, but then I have access to more and better spaced fonts than I used to.


I've been thinking today about your responses to my specimen graphics, particularly your comments about Verdana, and I think you -- and perhaps other people -- are confused about the nature of the CT fonts. Because these fonts are labelled 'screen fonts', I think there is an assumption that they have been designed to meet the same goals as Verdana and Georgia were. This is not the case. Verdana and Georgia were designed to very specifically address legibility issues in b/w, low resolution display: an acknowldgement that all the super-hinting applied to Times New Roman and Arial were not enough to compensate for the underlying unsuitability of those designs to screen readability. Georgia and Verdana are likely the most legible fonts for screen reading in existence, and they will probably remain so. I do think Georgia suffers under ClearType, because you lose the even colour ensured in the b/w rendering; Verdana performs better with ClearType. It was never the intention of the new CT fonts to try to out-do Verdana or Georgia for readability. Not only would it be difficult, it would be pointless: as if every print typeface were designed solely with the aim of being the most readable typeface in existence. It is worth noting that in designing additional screen fonts for Microsoft -- Nina and Meiryo Latin --, Matthew has returned to the Verdana design and reworked it: it has achieved a pinnacle, and there is nowhere much else to go in terms of sheer readability.

The new CT fonts were commissioned to meet a very different set of design briefs. The fonts needed to be readable, on screen and in print, but sheer readability was not a stated goal. The whole point of ClearType is that it removes much of the restrictions of b/w rendering that made the approach taken with Georgia and Verdana necessary. The new fonts were each commissioned for a particular use, either within Microsoft or among users. Cambria, for example, is intended to provide a more modern choice for business documents and other situations in which users have been churning out page after page of Times New Roman for so long. Constantia was specifically commissioned as a typeface for journals and other publications, and needed to be able to work for both online and print editions.

In addition to specific uses, the fonts were designed to show off certain aspects of ClearType rendering: to show how well ClearType renders a variety of different approaches to letterform design and detailing. Candara, for instance, was commissioned specifically to show how well ClearType can render those subtly flared stems; at the same time, it too was designed for a particular use: informal correspondence (I use it in my e-mail program).

I've been using most of these fonts for well over a year now, and I really do think they are very good and that you are simply wrong. I don't much care for the style of Cambria, and it's Cyrillic and Greek are not as good as in the other families. I don't use Calibri except in the occasional subhead. But I use all the other fonts every day: Candara for e-mail, Constantia as my default browser font, Corbel as my system UI font, and I'm typing this in Consolas. They work very well: they're all nicely readable on my screen, and much as I admire Verdana I like having a choice on screen as much as I like having a choice on paper.


I'm going to post the message I sent to David via e-mail (and to a couple of other people who requested it), on the basis that your last message in this thread was quite civil and also because my responses to some of David's points are quite similar to your own. Like Si, I don't recognise some of the statements as anything like what I've heard from people at Microsoft, so I presume that what we're getting is David's interpretation.


> O. If someone thinks the C-types are not the best, they are
> in the 4% minority of people who chose unanti-aliased Italic
> over aliased italic in the now famous "mall test."

What do you mean by 'the C-types'? Do you mean ClearType font rendering in general, or the
CT Font Collection (Calibri, Cambria, etc.)?

If you mean CT rendering in general, then there is a known percentage of people who prefer
b/w rendering. Adam Twardoch is one of them; I'm not. Some people are more colour
sensitive than others. I don't think this is an issue of typography, per se, at all, which
is why the statement doesn't make any sense if you're referring to the CT Font Collection.
It is an issue of rendering preference, and it doesn't matter how good a particular
typeface is: people who don't like ClearType rendering are not going to like any typeface
rendered that way.


> 1. Longhorn will not query the driver to find out what the dpi
> of the screen is.

Personally, I think ClearType is pretty bad at less than 120ppi, so I could see some
benefit to querying the screen pitch and having the option to turn CT off on screens with
lower pitch. But see No.4


> 2. Contrast in type designs are bad for the purpose of screen fonts.

I take it this means stem contrast. What kind of contrast? Obviously not all contrast,
since plenty of good screen fonts -- b/w, AA and CT -- have some contrast. Very high
contrast? That can be plenty problematic on paper too.


> 3. fractional point sizes are not possible on the screen.

As Si Daniels notes, fractional point sizes are possible in Avalon WinFX, but you're still
stuck with discreet pixels in the y-direction so rendering fractional sizes will produce
an interesting variation in height:width relationships. 11.5pt type might not be any
taller than 11pt type, depending on the resolution, but it would be wider: insta-extended
type!


> 4. the CT rasterizer has gotten "so good" that as resolution
> improves hinting will no longer be required.

Hmm. I think y-direction hints are going to be required for a long time to come, even well
past 200ppi screens. The question is what kind of x-direction hinting will be necessary in
sub-pixel rendering environments, and what kind of compatibility do you need with other
rendering environments, low-resolution printers, etc. Frankly, if a font is designed
purely to be used in a CT rendering environment at 120+ ppi, then I don't think much, if
any, x-direction hinting is necessary *for most designs*. Hinting -- as you well know --
is about all design: it is design specific, and there are always going to be some kinds of
design that require hinting more than others.

ClearType is in an awkward place, because it has to do something with those b/w
superhinted fonts (Times New Roman, et al), which look truly awful if you apply the
x-direction deltas within ClearType, so MS made the decision to disable x-direction deltas
completely in CT. Unless MS were going to replace every single core, web and other font
they ship with new versions optimised for ClearType, I don't think they had much of an
option in this regard: they had to ditch x-direction deltas. This means, of course, that
if one did have a typeface for which some kind of subtle deltas aided CT rendering, you
would be out of luck.

If one had the luxury of starting from scratch -- of designing a new rendering system that
didn't need to handle legacy fonts, for which one would only design new fonts that would
be optimised for that environment --, one could avoid many compromises. We are, as you
say, in a technological struggle, but then that is the history of type technology.
Gutenberg was in a technological struggle.


> 5. In general, color in type is more important than its rhythm.

I don't even know what that means. Depending how you think of colour and rhythm in text,
the two are either so separate that it doesn't make sense to speak of them relatively, or
so closely bound to one another that it is impossible to consider one without the other. I
mean, one can consider colour in terms of darker or lighter overall appearance of text, in
which case the rhythm may still be good or bad. Or one can consider colour as it relates
to rhythm, in the way that an unusually dark individual character will distort or destroy
the visual rhythm of text. When we talk about text looking 'uneven', we're generally
describing an effect in colour, but the impact on rhythm is pretty obvious. When we talk
about text overall looking too dark or too light, e.g. because of poor inking, then we're
describing something independent of rhythm.

It is difficult to respond to this statement without knowing in what sense colour and
rhythm are being discussed.


> 6. The human physiology of reading evolved to completion millions
> of years ago.

That's probably true in purely *physiological* terms, e.g. the structure of the human eye,
the arrangement of rods and cones, the way the eye connects to the brain. And the
cognitive processes used in reading are not something that came along after we invented text.

The interesting question for which there is probably no answer is whether those cognitive
processes have evolved further to specifically assist the reading process.


>7. Having the fonts done nearly two years ahead of release is a good thing.

You know how big software companies work as well as I do: a group has a budget for
development, so they'd better bloody well do the development or they'll lose the money and
any future funding. And there are small windows of opportunity for checking components
into an operating system. If you are shipping an operating system in 2006, you'd better
have your fonts ready two years ahead of time.

Is it a good thing or not, from a design perspective, that the CT Font Collection is
'ready' two years ahead of development? I think it is probably a good thing to be where we
are in terms of these fonts, in that if there fixes to be made we may have some
opportunities. Would I have liked to be able to spend another year on my fonts? Yes.


>8. Spending a million dollars on research to prove the quality of
>the fonts, ex post producto, is a legitimate process.

Did someone really say that? 'to prove the quality of the fonts'?

I think there is legitimate testing that should be done on any font that someone claims to
be 'highly readable'. I know that I fulfilled the design brief from the client for
Constantia, and I like reading it, but if the client is going to describe it as 'highly readable' then I'd like to understand what that means relative to other types or to some mean of readability scores.

Do I wish that more rigorous testing had been available as part of the design process?
Yes. But would I have done anything different if, for example, such testing told me that
really loose letterspacing is good for readability? I'm not sure.

I'm convinced that when we talk about reading, we're actually talking about two things:
the cognitive process of reading and the conscious experience of reading. When we sit down
to read, we are not aware of saccades and regressions and fovea and parafovea and all the
visual and cognitive mechanics of the activity: instead, we have an experience of text and
typography, the latter less conscious for most people than the former. I'm not at all
convinced that trying to design type to optimise the cognitive processes of reading --
which basically consists of trying to increase speed while not decreasing accuracy -- will
serve the experience of reading. So I'm wary of attempts to either direct or justify
design decisions from the laboratory.


>9. Making type that reads better, and I use the term "read"
>literally, will improve the way all small type appears to users.

I don't understand that one.



Now, Bill, do you think we can have a civil discussion? I picked up on one thing that you said in the past few days that I think is a good suggestion: in these kinds of big, multilingual font projects -- and by no means only the MS project --, there is a tendency for the non-Latin components to be more thoroughly reviewed and revised than the Latin. There is a presumption that type designers have appropriate experience and skill to design Latin letters without the kind of oversight and guidance that benefits e.g. Cyrillic and Greek. I think this is a problem, and I for one would be really happy to have someone with years more experience than me critique my Latin designs with the same attention that Maxim and Gerry critique my Cyrillic and Greek. Now, not all designers would take to that: there are some who refuse to accept advice on scripts that they don't know well, let alone on their native script. On the CT font project, I had the job of overseeing the design of Latin diacritic forms, ensuring that they corresponded to the expected norms of readers of the languages that employ them, but the idea of having someone reviewing the actual designs with a greater degree of attention is something that actually appeals to me a lot. I know how valuable such reviews are in designing non-Latin types, and I for one would welcome that kind of process for Latin in future.

Finally, while we're on the subject of Cyrillic and Greek, the multilingual, multiscript aspect of the new CT fonts is being overlooked in criticisms of their design and the way in which they were commissioned. The designers were selected in part because of demonstrable ability and/or experience with non-Latin type design.

Posted on Tuesday, March 29, 2005 - 4:47 am:

P.S. When John says "11.5pt type might not be any taller than 11pt type...[but it'll be wider]

another ray of light.

> I think y'all should make one...11 point grid....

No thanks, cher Maitre! Two or three book types, with a reasonable life expectancy of more than six months, is the summit of my typographical ambitions!

Bill
>the problems are so great that it is obvious that most of these designers have no grounding in text type vernacular

You really don't know how to judge this until you've done it, "found out" the difficulties and then perfected something that actually works at small low resolution sizes, and, small high resolution sizes, and large high resolution sizes.

It's unlike print text design...I mean, larger x-ht, wider spacing, etc. are common to small sizes of both high and low resolutions, but then the difficulty curve ramps up, all the issues of good type design are recolored, like when you sink below 50 meters of water, and you lose all the reds? Here you lose all the subtleties, and more, you're forced to make decisions about the screen results that effect the print results...

So if you've reached your summit, either just keep looking up, or just go back down but, please, don't shout up at them to jump, 'cause they might land on ya!?

There's so much to respond to but let me just address David's last points.

Yes, you are absolutely right, you don't know how to judge until you've done it, and I don't do screen type and I don't _want_ to do screen type -- ever.

But - I have every right to say what I said, because:

after spending let's say $5 mil minimum (considering everything) on these types, MS decided to allow the very first public viewing of these fonts, on the Poynter site, to be settings of text lines that (we now know) had been annihilated by InDesign's optical kerning engine (of destruction). We all had every right to expect that the first widespread showing of these types would show them at their best.

I would never have guessed that InDesign was the culprit unless I had simultaneously been working on a review of XPress 6.5 where I had to learn, for myself, what Indy's optical kerning really was doing. And I learnt that several days after I posted my initial comments about these types.

What I first saw really was as bad as I said it was. And I still think the types need work. John has made clear that the quality emphasis in the project was on the Greeks and Cyrillics (about which I could not possibly offer an opinion), at the expense of the Latin, and my view is that the Latins should have been perfected first.

However, I now know that the types are not as bad as I feared. But, I (we, all of us) only know that because I made a fuss about what I saw. In response John kindly posted some great images without all that InDesign garbage, so we could all see that all these people had not taken total leave of their senses, as it appeared that they had done at first - and that first showing to a predictable chorus of oohs and ahhs from ... whomever.

One irony is that none of this would have happened had the book been set in Microsoft Publisher.

That said, I now have the book, as of an hour ago, and must admit that though there is much to like, I still stand by the initial remarks I made, except the one about Lucas's fitting of the u.

You don't have to be an expert screen font designer to understand that there are still some fundamental problems here which should be resolved. An italic a that is too small or an e that is too wide is still that, no matter what the technology. These are simply out of 'idiomatic' bounds and will appear that way no matter how they are displayed. There are still too many weird looking, horizontally elongated s's and I cannot believe this is a restriction of the technology as has been suggested to me as a possibility by someone very skilled in the design of screen type.

One thing I must say is that I no longer am so troubled by the glitch serifs in John's type - whatever it is called, one gets so confused. (I vote the types be renamed to John's Type, Jelle's type, etc. - this C thing is a very bad idea.) However, although the numbers look good to me in print, they still seem to me too small in the 2nd series of screen shots John provided. In print, the t of John's type seems to me to be angled too far rightward. I may be somewhat to blame for this, as I was the first to bring this subtle design corrective to John's attention about a million and a half years ago, and I expect I still preserve the email I got back from him promising that he and Ross would take my advice in all future fonts. What I have later come to realize is that this rule is different for every t. Some need it, some don't. For instance, the t in Miller doesn't. What I can't tell, from the book, and just a casual glance, is whether the t really does lean rightwards, or if my eye is playing tricks on me, or is being tricked by some interesting feature of the design that I have not yet identified. My only other comment on this type in print is that the default cap spacing seems too tight in relation to the lower case. But this could be kerning I don't like. One example is Jo(hn) ... another. Hmmm. Yes. I think the caps are slightly too tight, but that kerning is making some combinations seem tighter than they really are. And yes, the glitch serif on the r really does seem to pull left, slightly, opposite to the direction of reading. It needs a fractional inclination to the right.

One problem with the book is that the only type you can really judge from it in continuous text is John's type. Although it is wonderful to be able to see so much of it, I do wish I could see extended passages in the other fonts as well.

I dunno, David, what do you think? (And by the way I hate what you did with the f in that readability type of yours based on Century OS! There! I said it!) (uh-oh - I think something heavy just landed on me) (And ... I have to say it ... though I have the highest opinion of wonderful Lucas .. I still think Calibri, as I have seen it, is alternatively clotted, and picket-fencey.) And as for Lucas's claim, in the book, that you could set a small type Bible in Calibri ... I have this to say. Designers working by themselves get hypnotized by their children. I have said a hundred thousand times that great type is usually only achieved by several people working together towards a common economic (and of course aesthetic) goal. Here, we actually have a chance that such a scenario could occur. I suggest that everyone involved with this project get together for a few weeks and take the gloves off. There is enough solid good work here for something miraculous still to be achieved. But only through rigorous review. Of that I am certain; on that I stand.

>Cambria, for example, is intended to provide a more modern choice for business documents and other situations in which users have been churning out page after page of Times New Roman for so long.<

Can't resist responding to this one: there is only one comparably compressed typeface that has ever made significant inroads on Times, and that unfortunately is Minion. Everyone hates Minion -- even Robert -- but I have never seen any type succeed in coming as close as it does to functionally replacing Times (which is Robert's favourite type). Speaking of Cambria, do you notice how here the t really does lean too far leftwards? I prefer your approach, even if I think you overdid it, and I recognize that there might have been technical reasons that forced you into a compromise. That said, there should have been an art director who ensured that things like the t's all did the right things, etc. etc. etc.

>Cambria, for example, is intended to provide a more modern choice for business documents and other situations in which users have been churning out page after page of Times New Roman for so long.<

Can't resist responding to this one: there is only one comparably compressed typeface that has ever made significant inroads on Times, and that unfortunately is Minion. Everyone hates Minion -- even Robert -- but I have never seen any type succeed in coming as close as it does to functionally replacing Times (which is Robert's favourite type). Speaking of Cambria, do you notice how here the t really does lean too far leftwards? I prefer your approach, even if I think you overdid it, and I recognize that there might have been technical reasons that forced you into a compromise. That said, there should have been an art director who ensured that things like the t's all did the right things, etc. etc. etc.

to be settings of text lines that (we now know) had been annihilated by InDesign's optical kerning engine

We know nothing of the sort. Can you please respond to my request to identify the specific graphic on the Poynter site, and the problem therein, so I can compare to the book and the original InDesign file. I'm pretty sure that everything in the book except the character set showings did not use ID optical kerning.

As I noted in the other thread, the scans on the Poynter site have clearly been through Photoshop and touched up to remove paper texture and other scanning artifacts. So I wouldn't rule out that problems might have originated at that stage.

Re. Constantia ('John's type'): My only other comment on this type in print is that the default cap spacing seems too tight in relation to the lower case.

I'm not entirely sure what you mean. Do you think the caps are too tightly spaced to the lowercase, i.e. at the beginning of capitalised words? We might have a difference of approach in this regard.

That said, there should have been an art director who ensured that things like the t's all did the right things, etc. etc. etc.

As I wrote elsewhere, I'd be very happy to work with a suitable art director on virtually any project. But even with Maxim and Gerry's feedback on the Cyrillic and Greek, not all designers responded as positively or found the relationship easy. As someone who works primarily on non-Latin types, or on types that involve a non-Latin component, I'm very used to working with expert reviewers. It is something that other designers might rebel against, feeling that their creative independence is hampered.

John, sorry, the specific graphic on Poynter is the one that -- sheesh, there are so many urls in this discussion, who can keep track? -- started it all. Anyway, I think the one I mean is the Calibri one with the note that says it has been scanned from Now Read This. It's rather late and perhaps my eyes deceive me but, still, I must say that neither in the book nor at Poynter can I find a single word in Calibri that seems to have even colour. Letters seem either too close or too far, all in the same word. I _think_, however, that the fault is somewhat magnified in the Poynter scan, seen onscreen, as compared to the book but I may be imagining that. fi, ti, notice how they glob up. It's a pretty idea but yields too much density. Anyway, I'm reasonably sure that the Poynter scan makes things look slightly worse than they are. For example, the bold x diagonals really look too thick on the scan, and still too thick but not so bad in the book. Now, suppose that this is a technical requirement that would never be noticed at normal screen sizes? That would invalidate some of my comments. I also see what seems like a lot of bad spacing elsewhere, say, in Candara, but I can't know whether the problem is in the fonts or the settings. My assumption must be that the settings are as complimentary as possible but for any of number of reasons that might not be the case.

Re reviewers, I happen to know that you work with them well, and congratulate you on this difficult achievement. Those who can't do this don't belong in text type design. Creative independence has to be tempered with (let me call it) the idiomatic wisdom of the experienced community - accepting as a given that some compromises due to technological limitations may have to be imposed.

Mais je divague, and worse, grow pompous. To return to the point, I _think_ I see Indy spacing/kerning artefacts in the book which I _think_ have been slightly magnified by the scan and the imposed screen resolution for the whole process, none of which at all reflects what these types might look like when eventually used as intended.

Re John's type, yes, I mean some but not all of the cap to lower case combinations. Of course there are different approaches, and it is all a terrible headache to judge. For example, if I remember correctly, there are several Carter fonts where caps ending in stems are kerned to any likely lowercase letter, an approach I can't offhand recall taken elsewhere.

Speaking of different approaches to spacing, there is a tendency to make o spaces too large, for various reasons. I think this has gone too far in Calibri, but obviously can't be sure. I wish someone would invent a program that at the very least would let you space a minimum of four classes of letters -- lowercase and uppercase rounds and straights -- manually, and then let you automatically play with varying the proportions between each class. Of course, someone will say that this has already been done, but I have been shrieking about it for years and have yet to see it done.