I too did the same thing. Read Form of the Book cover-to-cover (noticing that it was not necessarily typeset according to the what was written) and of course Elements. However, I'll make a few comments. 1. In Elements, Bringhurst explains how it is preferable in theory to optically align pages with hanging hyphens. I have even seen in another book how to do it in Quark, but of course it's completely impracticle to typeset it that way unless you happen to be using InDesign. 2. Regarding your problem on page 95. Firstly he was talking about hand-setting it, secondly, with programs such as InDesign and I believe TeX (not that I have ever used the latter) you have the option of glyph scaling. I am sure that Tchichold in his world of lead type would never have dreamed of such technologies.

I tend to use Elements as a guide, but even so, setting small caps in the time that Elements was done, was a tedious process (again I'm only talking about programs that I knew such as Quark, PageMaker, FrameMaker etc, I have no idea about Tex which I kinda understand can do anything), but then again today with programs such as InDesign, this is very straightforward. I only say this because when all type was set by hand, it was no more effort to use the correct figure then to use the incorrect one. Since the advent of the Apple Mac with Aldus Pagemaker, making DTP something affordable in the 1980s, many compromises were made to the point that good typesetting was very difficult to accomplish. Just take the word "difficult". Those of us who worried about switching the ligature to the "ffi" ligature ran into problems of letter spacing and hyphenation. And the mainstream apps such as Quark et al meant that these typographical nuances were simply ignored.

Today, when I offer my clients, true ligatures, hanging punctuation, real small caps, real superiors etc etc, they simply do not realise what they are getting since they are so used to the mediocre level of typesetting where large typesetting house happily typeset their books in Times New Roman and happily have very tight or very loose tracking so that they can fit there text to the page according to some book designers warped understanding of what a book should look like.

I guess this is why typesetting books are typeset badly! What I do is read the content of the book and for real application of the book I try and find old books that were typeset well in the days of lead where only the professionals knew that the location of the small letters were in the lowercase of the box of lead characters.

I

<quote>Like, I agree with a lot of Tschichold's advice wrt spacing issues yet much of this <em>runs contrary to Knuth's decisions in TeX</em>, so I've always doubted I knew what I was talking about.</quote>

what exactly do you mean?

Popescu,

Tschichold spends several paragraphs in a detailed discussion of using closely-set type and avoiding excess spaces. For example, he makes specific recommendations against extra spaces after sentence endings, and insists that using fixed spacing between lines is (in general) the right thing to do. I was also surprised to learn that he felt widows were less of a sin than using extra spacing, and I don't necessarily disagree with that.

By default TeX uses extra and variable spacing all over the place: spaces after sentence endings are 1.5x the width, flexible "glue" is usually inserted between lines, spaces after other punctuation marks are given extra glue, and additional complexities are used. I've never particularly cared for the resulting appearance yet I've always felt my judgement was a bit off. Turns out at least one other person agrees with me, at least to some extent.

The reason I'm asking about typography books is because I'm writing my own typesetting engine (hopefully it'll be available someday--for free, of course), and I'm looking for various opinions on what constitutes good and bad typography so I can try to make it flexible enough to accomodate them. I'm also thinking about other approaches to the whole "is this a good layout?" heuristic analysis than the somewhat simplistic ones used by most of the other systems I've worked with. Looking at "optical spacing" is a start, but it seems there may be additional information that would help to guide the process.

While human intervention is always going to be necessary for truly beautiful results, I want to try and get as close as is reasonable. And while WYSIWYG systems are an excellent choice for many things, the project that prompted me to do this just isn't feasable with such a system.

Soon I should have some output worth showing, and then people can tell me how full of crap I am and I won't feel so bad about criticizing Felici or Bringhurst

I think it's pretty widely accepted that some of the default settings in TeX run contrary to what most people nowadays would regard as good typographic practice. (You can't blame Knuth, who knew nothing about typography when he started the project, and whose prime exemplars were mathematical journals from the early 20th century.)

For example, I've never seen a serious typographer argue that it's a good idea to add extra space between sentences. Certainly the majority would disagree rather forcefully, I imagine!

The whole business of stretchable vertical space is interesting, because of course such a thing wasn't generally possible with earlier technologies, so it was a reasonable thing to experiment with at the time. But I think (and I'm not so sure here, but I would guess I'm in a substantial majority again) that the results of using this are not usually effective. I don't think the leading between lines is stretchable by default, but the spacing between sections and paragraphs certainly is.

Similarly Knuth's recommendation to use a spaced-out em dash for punctuation is rather eccentric. Most people would use either a spaced-out en dash or (less commonly) a pretty tight em dash.

Bringhurst is a much more reliable guide on matters of style, in my humble and rather ignorant opinion. (The only suggestion in Bringhurst that I find completely baffling is his insistence that, in a ragged-right setting, it's best never to break words. Would anyone care to convince me of this?)

Instincts...

I agree, it is respectable to learn form past experiences but future of type ore anything ells is not decided and not known from past as they created it from they past not present as we must do now. Learn from past not fours it to be present roles. Some mistakes ore wrongs of past could be correct now

PS. The documentation for Peter Wilson's Memoir class contains a lot of reasonably sane advice about book typography.

Don't forget that the defaults for TeX were set in the early 1980s, and that just because they are the defaults doesn't mean you can't change them. For example, if you don't like the extra space after a sentence, just put \frenchspacing in your preamble. That's no more difficult than setting up InDesign. (And yes, TeX and InDesign are my two main typesetting systems.)

As for the "glue" in between paragraphs, I agree it can be dangerous (and I frequently have to "defeat" it when I typeset with TeX). Systems built on TeX, such as ConTeXt, offer good facilities for typesetting on a grid.

And while Knuth did do a lot of research on typography while developing TeX and Metafont, I wouldn't use the TeXbook as my typographic guide except for matters mathematical. Bringhurst is a much better guide. (And I agree with you, Robin, about the ragged-right hyphenation issue.)

1. [42nd SSD] is because I'm writing my own typesetting engine.

Aren't you tryin to reinvent the wheel? i mean yes its true that some defaults in TeX aren't pleasing for everybody but you can change easily. and that's, in my opinion, the best advantage of TeX vs, let's say, InDesign.

2. [robin huston] Knuth, who knew nothing about typography when he started the project

im not so sure about that. one cant possibly write a system as TeX without knowing nothing about typography.

3. without the glue, TeX paragraph composition algorithm will be useless, isnt it?

4. why not use interline extra space if its so little that the eye is fooled and if the result will be more pleasant as a whole? the same question about extra compression/expansion of types pdfeTeX can do.

5. ragged-right paragraphs: i use this setting for its side effect - fixed interwords spaces (or is the other way - using fixed interword space have the side effect of ragged-right paragraphs?). so, if using hyphenation will give me a more straight left margin, why not using it?

6. [42nd SSD] what i want from a typesetting engine is this: first - complete detachment between form and content, second - let me choose/change anything (i dont use TeX and friends for their defaults, but for their full or almost full customization), third - dealing gracefully with various input/output systems/media etc.

although i encourage u with ur project, my personal opinion is that such effort deserve other purpose. have u consider joing etex/pdftex/omega team?

I think you are looking in the wrong forests. I

I appreciate all the thoughtful comments and sincere feedback.

I agree 100% you can't "learn" good typography from reading, as it's more art than science at this point. I'm not looking for magic formulas or rote approaches, because I don't believe there's any one answer to what is "good book design".

I do feel that contrasting and comparing various approaches is very helpful, and just in the last few days I've gained new insights into how various people regard the problems. And I appreciate seeing other people's opinions on what constitutes good design, as I figure it won't hurt.

The issues I have with TeX are legion, and I've been using on a regular basis since 1986. Honestly, it's time to say goodbye and move on. It hasn't kept pace with the world around it, and trying to drag it into the 21st century just isn't worth the personal effort to me. I don't need the advantages it offers (a community of users and existing styles).

What broke the camel's back was when I went to install Yet Another Font and realized what a horrible mess it was to use "non-standard" Type 1 fonts with TeX. Then I thought about the style I need, which uses lots of flowing parallel text; this is a non-trivial problem for TeX. And, its page-break algorithms are not that good... multilanguage support is iffy at best... and so on. All of TeX's weaknesses came to a head on this particular project, and none of its strengths. I know people are valiantly trying to make it work in modern times, but I believe the fundamental design is inadeqate given the amount of change which has taken place around it.

I totally agree that the tool doesn't make one a good typographer, because almost anything can be used to produce good documents. Heck, I could write raw Postscript if I had to, or bludgeon something like groff into doing what I need, or even scribble words on a page with a pen. My problem is that I need to format some very large documents, ones which will be updated frequently from a database, and I can't afford to spend the time setting each page by hand. (Tweaking, of course I expect that; but only a reasonable amount.) I need something that can generate decent output and handle the complex formats I've wanted for years without having to jump through TeX's ridiculous hoops.

Charles, thanks for the recommendation for The Art of the Book, and I'll be sure and take a look. I have my personal collection of rare books as well, many of which are beautifully printed (and others which are simply awful )

I have my own appreciation of what consistutes beautiful typesetting; the education for me is that many people disagree, not only with me but with each other as well.

Not to minimize your problems, but they are solvable with TeX (but not always easily). As for fonts, we always made a

<quote> One of the reasons we are (probably) leaving our TeX engine has to do with a need for 16-bit fonts (Unicode)</quote>

omega deals happily with 16bit fonts.

<quote>InDesign in a week or so. But it has many, many features that will be hard to overcome knocking out the existing hyphenation routine (I believe it is an algorithm),</quote>

it's the algorithm taken from hz program developed for monotype by Hermann Zapf. adobe bought this program early in 90s.

<quote>so you have control over all hyphenation.</quote>

in theory, never InDesign

Actually, Adobe isn't the only party that has rights to Hz...

hhp

<quote>omega deals happily with 16bit fonts.</quote>

But Omega seems to undergo al lot of change, it isn't "finished" like TeX.

And (unless I'm wrong yet again) if you have ALL the words over 5 letters that occur in a book you are setting in an exception dictionary, then you have control -- it doesn't matter what the algorithm might do, as long as the exception dictionary is checked first. Of course, if the author re-writes in proof . . .

Charles

> it's the algorithm taken from hz program developed for monotype by Hermann Zapf. adobe bought this program early in 90s.

That was URW rather than Monotype, and the mid-90s, I think.

Hrant, who licensed the patent from URW before Adobe bought it?

T

If I told you I'd have to distribute you.

hhp

> That was URW rather than Monotype

true. my mistake.

Beautiful. Please come write for Typographica.

I guess that isn't very helpful. I usually recommend Bringhurst
because his is the one that has benefitted me most.

There is more to dig through at Typophile Books.