Bringhurst et al on spacing kerning etc.

[Bill Troop billtroop@xxxxxxxx]

> BT> There is no absolute rule, but see Bringhurst's The Elements of
> BT> Typographic Style, an invaluable handbook. His position --
> BT> somewhat extreme -- is that punctuation should always be set
> BT> in roman.
>
> Even if the punctuation is part of the italicized matter? That does
> sound extreme. What's the rationale, I wonder?

Going to look it up, I find that Bringhurst (p. 85, 2nd ed.) is discussing
only a limited range of punctuation:
"Parentheses and brackets are not letters, and it makes little sense to
speak of them as roman and italic. There are vertical parentheses and
sloped ones, and the parentheses on italic fonts are almost always sloped,
but vertical parentheses are generally to be preferred. That means they
must come from the roman font, and may need extra spacing when used with
italic letterforms. [Illustration] [Sidebar: This rule has been broken more
often than followed since the end of the 16th century. It was followed more
often than broken by the best of the early typographers who set books in
italic: Aldus Manutius, Bershon Soncino, Johann Froben, Simon de Colines,
Robert Estienne, Lucovico degli Arrighi and Henri Estienne II.]
"The sloped square brackets usually found on italic fonts are, if anything,
even less useful than sloped parentheses. If, perish the thought, there
were a book or film entitled _The View from My [sic] Bed_, sloped brackets
might be useful as a way of indicating that the brackets and their contents
are actually part of the title. Otherwise, vertical brackets should be
used, whether the text is roman or italic: 'The View from My [_sic_] Bed'
and '_the view from my _[_sic_] _bed.'"
Bringhurst recently amplified these views in a letter to me anent a
cyclical controversy regarding this very issue that broke out on one of the
type design lists:

"Dear Bill,

"I don't see the point of arguing about italic parens,
myself. In fact I don't see the point of arguing generally.
People who want to use italic parens should use them --
and will. But people who want to believe that italic
parens are necessary or eternal or God-given will have
a rough time proving their case.

"Because most italics are sloped, the lazy and uninformed
(i.e., most people) have come to think that italic "means"
sloped, or that italics are always and necessarily sloped.
Because I think understanding things is a lot more fun
than not understanding them, I try to teach people to
grasp italic structure, and thus to understand that slope
is optional in italics. I also like to show them that early
italics -- which are frequently more beautiful than later
ones -- were always used with upright (roman) caps.
You know all this, of course; I'm just rehearsing the
argument (in the older, less combative, sense of the word).

"Guyot, Garamont, and other members of "the generation
of 1500" cut sloped italics and started sloping everything:
caps and punctuation included. One might think that this
would lead to increased independence of italic, but in fact
it coincides with the opposite : a demotion of italic to eternal
subordinate status.

"Aldus, Soncino, Froben, Colines, Robert Estienne, Henri II
Estienne, and Arrighi, who all set entire books in italic,
routinely favored modest slope, and routinely preferred
upright ("roman") parens and [if they used them at all] square
brackets. For them, parens and square brackets were linear
forms, in effect, bent pieces of rule. "Italicizing" them made no
more sense than italicizing an em dash.

"If you have Harry Carter's book handy, you can see an illustration
of this practice in a page using one of Arrighi's types cut by Lautizio
Perugino -- the one with extra vowels, cut at the request of Gian Giorgio
Trissino [Carter, fig. 80]. Same procedure wouldn't work so well with
the adjacent italic [fig. 81], which belongs to the later style.

"Guyot became something of a specialist in the new manner of italic
and imparted a slope to everything -- except for numerals. Even
Caslon, 200 yrs later, cut no "italic" numerals -- for which he perceived
no practical need. We do have a practical need for italic numerals, so
we add them to recreations of Caslon's and Guyot's and Garamont's
type. But do we have either a practical or an aesthetic need for
sloped parens? Some people think so. I find that I don't. I prefer,
most of the time, to treat parens and square brackets as nonletters, like
the em dash or the asterisk, the dagger or the fist.

"There are those who insist that everyone join the union. They even cut
sloped daggers and ellipticized asterisks. They are not, in my opinion,
wrong to do so. But my preferences are different.

Cheers,
Robert"
> LB> Columbia University Press instructs its editorialistas to make
> LB> certain that punctuation following any word in italics or bold
> LB> shall be set in roman type.
>
> Bringhurst's rule again. Hmmm. A foolish consistency?
Bringhurst is principally concerned with parentheses and brackets, and
notes that additional spacing adjustments will be required. Of course he is
speaking of top quality typesetting; one could not easily make such
adjustments in a word processing program.
> WooF> Punctuation marks which look different in the normal and the
> WooF> italic font follow logic: In the question above, the question
> WooF> mark is not part of the name of the movie. Punctuation marks
> WooF> which look the same in the normal and the italic fonts (period,
> WooF> comma, single and double quotation marks) should be in the same
> WooF> font as the immediately preceding word, in order that the
> WooF> spacing (technical term: "kerning") will come out right.
That is not what kerning is. Spacing is spacing and kerning is kerning. A
kern refers to a specific adjustment made between two letters, such as T
and i. Bringhurst on kerning: "Part of a letter that extends into the space
of another. In many alphabets, the roman f has a kern to the right, the
roman j a kern to the left, and the italic f one of each. As a verb, kern
means to alter the fit of certain letter combinations -- To or VA, for
example -- so that the limb of one projects over or under the body or limb
of the other." It only needs to be added that fonts usually contain
somewhere between 200 and 3000 kern pairs, and typesetting programs such as
Quark and Pagemaker allow extensive modification of the kerning tables
manufacturers provide with their fonts; and that some kern pairs are
positive. For example, in the font Monotype Bembo, where the f overhangs
considerably, there is a positive kern between the f and the space
character, which would otherwise appear too small in words ending in f.
To get back to the point WooF addresses, no two fonts are spaced according
to the same standards. In a well-designed font, the italic o and roman o
will be centered identically, but spacing and offsets can be radically
different for other characters. For this reason, commas, periods, etc.
should almost always be set in the same font as the previous text. However,
this does not absolutely guarantee that the punctuation will have the
desired spacing. Automatic kerning cannot of course occur between a roman
and an italic font, but manual kerning adjustments can always be made.
Finally, tracking is something else entirely. Tracking refers to a space
adjustment that is made uniformly over a word, paragraph, page, or entire
document. Tracking text negatively means adding more proportional space to
everything -- including the space character and all kern pairs -- thus
tightening up the text. Tracking text positively does the reverse. It is a
rule of thumb that positive tracking is desirable the smaller the text is;
negative tracking is desired the larger the text is. Most fonts have their
spacing optimized for 12 points, and can often benefit from a little
positive tracking when used at 10 points. A few fonts (for example Lino's
Times Ten, or Matthew Carter's Miller, are actually optimized for use at 10
points and would probably be improved by a little negative tracking if they
were used at 12 points or 14 points.)
A kern is always an alteration of the natural space that would normally
take place between only _two_ letters. Thus the term 'kerning pairs'.
It may help to explain this a little further. In a font, each letter
occupies a box, and has what are called 'sidebearings' on each side. For
example, the letter o might have sidebearings of 35 units on each side; the
letter T might have 5 units of space (measuring from the extreme serifs).
The font may contain a kern pair instruction for the To combination that
may look like this in the .afm file: "T o -45". This means that in a
program that honors kerning (such as Pagemaker, Quark, or WordPerfect), 45
units of space will be subtracted whenever T and o occur next to each
other. To give some idea of what a 'unit' is, the stem of the T will
typically be about 100 units wide and the letter will typically be about
700 units tall. The width of the space character will typically be 240
units. In a font with an overhanging f (such as Bembo, Centaur, or Caslon),
the top of the f actually falls outside of the bounding box by 20 or 30
units, or more. This is referred to as a 'negative right sidebearing'. The
total width of the bounding box is different for each character in a
proportionally spaced font. The total width equals the extreme positions of
the character plus the right and left sidebearings. Each character has
different sidebearings, as required to make it harmonize with all the other
letters. In type design, setting the sidebearings is often far more
difficult than designing the actual letters. The 'spacing' of the font is
shorthand for the 'left and and right sidebearings'.
In a Postscript font, the kerning information is contained in human
readable form in the .afm file (both platforms), and in machine readable
form in the .pfm file (PC) and the suitcase file (Mac). In a Truetype font,
there is only one file (typically with the extension .ttf), which contains
both the font outline instructions and the kerning instructions. The
kerning information for a Truetype font is not human readable, and must be
decoded in a font editing or page layout program.
> ...
> A similar dilemma pertains to quotation marks: Does punctuation go
> before or after the close quote? If I'm not mistaken, British usage
> is "logical", whereas in U.S. publishing the punctuation always goes
> inside the close quote, whether it really belongs there or not. Until
> recently, I've resisted this convention. Adding marks to a quotation
> that weren't in the original -- it goes against the grain.
Bringhurst has this rather interesting amplification (p. 87) which
addresses a third style that automatically comes with many contemporary
fonts (for instance, if I am not mistaken, many in the Adobe "Originals"
series):
"Punctuation is normally placed inside a closing single or double guillemet
if it belongs to the quotation, and outside otherwise. With other quotation
marks, usage is less consistent. Most North American editors like their
commas and periods inside the raised commas, "like this," but their colons
and semicolons outside. Many British editors prefer to put all punctuation
outside, with the milk and the cat. The kerning capabilities of digital
typesetters, especially in the hands of the advertising typographers, have
evolved an intermediate third style, in which closing quotation marks are
kerned over the top of commas and periods. Typographically, this is a good
idea with some faces in large sizes, but a bad idea with many faces at text
sizes, where a kerned quotation mark or apostrophe may look much like a
question or exclamation mark. [Illustration] [And note, that with some
fonts, you will get this effect automatically, unless you turn kerning
off.] When quotation marks are not kerned, it makes no _typographic_
difference whether they follow commas and periods or precede them. But
typographers, like editors, should be consistent, whichever route they choose."