not that i’ve studied this at all, but my gut reaction is that simpler is more readable. I would think that ornamentation and high-contrast make an italic less readable, so shoot for simpler, lower-contrast forms???

I actually always liked the Linotype-technology Sabon italic. Some say that the fact that the italics had to be drawn on the roman proportions was a huge limitation, and of course it was. The Sabon italic was a bit too wide and unnatural, but on the other hand, it actually was quite widely spaced which I fould worked nicely. I see the same charasteristics in Verdigris italic, and I think it works very well!

Adam

I have no idea but it is a very very good question.

I am by no means an historian and have no developed theories nor have authoritative reference material to back it up, but it seems that even in the earliest of comparisons between Roman (inscription based) and italics (cursives) that indeed italics seem not only more fluid but facilitate reading.

It may (and I emphasize ’may’) be that informal penned forms were originally the most common expression of individual written communication (handwriting). Since most literate individuals can quickly write and decipher the majority of other peoples’ handwriting, we may have developed a mental facility for decripting diverse and variable script forms.

Even in the poor reproduction above, to me it seems easier to read the italics.

Since most italic typefaces remain faithful to the earliest of typeface designs, and since the earliest of italics approximated Chancery scripts, the visual link back to human written forms remain.

I would tend to agree with Paul and Adam, and taking it slightly further I would say that in most cases, what makes a legible roman makes a legible italic, for example: simple terminals, wide apertures, a reasonably large x-height, etc.

Something you probably wouldn’t expect (and which some may disagree with me about) is that swashier caps (within reason) tend to improve legibility! I think this is mainly because it’s an indication that the caps were designed to fit with the italic, as opposed to being ripped from the roman and skewed.

“Even in the poor reproduction above, to me it seems easier to read the italics.”

Actually, I think the opposite is true. To me the Roman reads much better than the italic.
I interpreted Williams question to be about the readability of italics among themselves, not so much compared to their upright counterparts.

To me, the main issue with Italics is that they have sufficient separability from the upright to be noticeable without being alarmingly different in color. Since it is rare to see long passages of italic text, readability issues (tiring of the readers eyes) are not very relevant. What is sometimes done to create separation from roman is to slightly condense the italic yet lighten the weight a bit to counterbalance the color change. As Adam mentioned the Sabon went the other direction but was still handled quite well. It isn’t so much how you do it as how well you do it.

ChrisL

Readability and contrast-with-Roman are necessarily strongly opposed, for the simple reason that -hopefully- the Roman is the one made highly readable! That said, the dimension of difference that might least impede readability in an italic is color. So I think the “best” general italic is one with slight slant, a little bit of compression, maybe slightly smaller on the body, and visibly lighter in color. Also, to prevent it from skewing the “voice” of the Roman in an arbitrarily fluid/casual direction, the “ideal” italic is structurally quite close to the Roman. Hmmm, Harrier anyone? :-)

http://www.themicrofoundry.com/other/nour&patria/dev/nour-latin.gif

BUT: Since an italic is generally for subordinate snippets, its readability requirements are on the order of display types (ie not very great). Just as long as errant fixations are not triggered though.

hhp

William, I apologize ahead of time if this doesn’t follow the original intent of your topic, but I’ve thought more about the cognitive relationship between italics and handwriting.

Up until recently, handwriting was an individual’s most common means of informal communications and also the most practiced. Thoughts and feelings translated by one’s mind through manual activity into visual notations that are then fed back through the eye for mental confirmation must have some benefit in one’s ability to read script(my personal musing with no facts).

Now all daily written communication (business and personal) is done by keyboard or phone keypad, with less time writing with pen or pencil. Eventually our facility for written communication will be heavily screen and typing based. In time reading and writing handwritten scripts will become less common and more difficult.

Whether or not that might have any bearing on our ability to read italics is yet to be seen, that is of course, if there’s any correlation at all between the ability to handwrite and the recognition of script forms.

On a personal note, for decades, I hand print my notes and letters much faster and more legibly using roman letters. My own handwriting has deteriorated and is slower for me to do than writing block letters within the past ten years. I would also say that at least 70% of my daily reading is done on screen, and maybe less than 2% of what I read is someone else’s personal handwriting.

I bring these thoughts up as just an awareness that may or may not have an affect on how the majority of us read and write. If it does, then that will affect how, why and what new typefaces are designed for... roman, italics, scripts, et al.

Signing off... no more off-topic intrusions.

I think Norbert is right that for e.g. Aldus’ contemporaries, italic text was closer to what they were reading on a daily basis than the more formal roman types, and that readers were equally comfortable with both and may even have been more comfortable reading the italic types. It is important to remember that our contrastive use of italics in predominantly roman text is a later convention, and that the primary distinction is between formal and cursive writing models, between the book hand and the chancery hand. This is a distinction that is not limited to the Latin script: you find exactly the same thing in other mature manuscript traditions, e.g. Hebrew (with the added, third class of the semi-formal rashi style). The convention of use is not usually contrastive of different kinds of text within the same book, but contrastive of different kinds of books.

> I would be interested in people’s views on what makes for a readable italic.

National Geographic Caption (Matthew Carter, 1979) is a fine, and well-known, example of a stand-alone italic (no companion upright style) designed to be as readable [to the present-day reader], as a roman typeface.

Norbert, I don’t really buy that italic was ever as readable as Aldus’s Roman. What Jenson and Griffo and Garamond figured out was, I think, how to make type *more* readable than written letters.

I forget where I read that the most readable is lower case serif type, and whenever you depart from it it slows the reader down. I believe it.

My question was, given that italic is slower, what can you do to reduce the difficulties. Thanks, Maxim, for the example of Carter’s National Geographic italic. It solves the problem, it seems, by making an italic the same weight and width as the roman. That works admirably for his purpose, but wouldn’t be so good as a companion italic. So of the different ways you can move toward the roman, which is most helpful? Or are there othere ways to achieve greater readability, other than increasing width, reducing slant, increasing weight?

> It solves the problem, it seems, by making an italic the same weight and width as the roman.

It’s not just the weight and the width...

“A drawing comparing Carter’s National Geographic Caption (1979) to the magazine’s previously used Roman and Italic typefaces. The overall weight and proportions are about the same, but the new design slants less than Italic and is as wide as Roman.” [Source: Washington University in St. Louis News & Information > University News > Typographically Speaking at Des Lee Oct. 10–Nov. 29 [2003]. Exhibition examines work of Matthew Carter, among the 20th century’s preeminent type designers. By Liam Otten]

> wouldn’t be so good as a companion italic.

True, but I think simply making it lighter would allow it to
click as a companion (without impeding readability much).

hhp

But the National Geographic Caption typeface was never intended to be a companion to a roman in the sense that the regular italic type was. It is an example of an italic designed to be used alone, and also at a smaller size than the regular types.

My Constantia italic is another example of a wide italic, proportionally similar to the roman and, generally, setting only slightly shorter than the same text in the roman. In this case, of course, the aim was not general readability but specific readability in the context of ClearType screen rendering. In that case is it slightly lighter than the roman, and intended to be a companion to it and not a stand alone italic type.

What Jenson and Griffo and Garamond figured out was, I think, how to make type *more* readable than written letters.

But they were building on top of a distinction that had already been made, between formal book hands and running chancery hands, and which already exhibited different levels of readability. The key distinction, in the manuscript tradition, is of speed: the formal book hand is written slowly, without returning strokes, and tends to be more regular and less compressed. The running hand is faster, and with this comes slant and compression. What is interesting is that development of the less readable italic type followed that of the more readable roman, and was very popular both with printers — since it allowed them to get more words on a page — and with the reading public.

Robert Bringhurst made the observation, during his keynote address at the ATypI conference in Vancouver, that at the time we are talking about everyone who was a reader was also a writer (usually in both sense of the word), and so recognised letterforms not only as a consumer but as a maker. This is similar to the point Norbert is making, I think. To know forms manually, through the regular making of them, may contribute to facility in cognitive recognition. So while italics may never have been more readable than romans, I have no trouble with the notion that italics may have been more readable for renaissance readers than they are for us.

> it allowed them to get more words on a page

But -as I once tried to explain in my Daidala interview- narrow type isn’t necessarily more economical; for the simple reason that you’re losing apparent size in the process. Basically, it depends on the proportion of line breaks. Which is why Luce and Fournier called their narrow type “poetique”: poetry -generally- breaks after every line, and that makes narrow type directly more economical.

Also, I have actually heard an expert or two claim that the view that Aldus used italic type because it was narrow because he was making smaller books is ballony.

hhp

Well, Aldus did use italic types for text in his smaller format books, and used roman in the larger and grander productions like the Hypnerotomachia, but it would indeed be reductionist balloney to claim that the motivation was economical. Unfortunately, most of the other motivations ascribed to Aldus by print historians are also dubious, because the people making them are print historians and they’ve largely ignored the manuscript culture within which Aldus is producing his books.

We shouldn’t imagine that because Aldus is doing something that has not been done before in type that he is doing something that has not been done before. He is applying the conventions of formal book hand and cursive chancery hand usage that were well established in Italian humanist manuscript production. The formal book hand is used in larger format volumes, usually prestigious productions intended to be experienced as objects of beauty as well as texts to be read. The chancery hand is used in more quickly produced, smaller books or booklets whose function is to transmit the text in a plain and convenient format. I’ve seen dozens of both kinds of manuscript books. Aldus was simply applying the distinction that already existed in the new medium of type. Unfortunately, most commentators on printing history know very little about manuscript production, so credit Aldus with the invention of the small format pocketbook, rather than the adaptation of an existing model to the new technology and market.

Verdigris Italic has quite open counters, especially in the h m n u — their shoulders are not the typical sharp angles but round. This can already be found in the italics of Goudy’s Truesdell and Village. It looks like Goudy adopted a form of shoulders which he first employed in his Mediaeval. Personally I like this trick because it allows to keep letterforms narrow.

“I don’t really buy that italic was ever as readable as Aldus’s Roman. ... I forget where I read that the most readable is lower case serif type, and whenever you depart from it it slows the reader down. I believe it.”

No talk about textur/fraktur goes without the obligatory Bismarck quote, that no type is more readable than fraktur (as opposed to roman type), which is to amuse the audience: Nobody would subscribe to this TODAY. But this nicely indicates that whether roman, italic, blackletter are regarded as being more readable is a matter of habitualisation rather than any form per se being more readable. Maybe we are already moving away from (serif) roman, towards sanserif, and in some decades nobody would question that sanserifs are “most readable,” and who needs these disturbing serifs which are a mere reminiscence of another technique of producing letterforms.

>their shoulders are not the typical sharp angles but round

Thanks Karsten, nice insight. That is the kind of thing I was looking for in my question—how to make the italic more readable without pushing it too far toward the roman.

Another example that had been puzzling me is Spectrum’s italic, which seems quite readable, at least in short passages, in spite of being quite narrow. Now looking again, the round letters are quite open. And maybe the disparity of the narrow mn is not so good for even color...

I can believe that fraktur is quite readable to anyone accustomed to it. Though I wonder if it can be printed at as small sizes without the hairlines becoming too thin or the letters clotted. Was fraktur generally printed at somewhat larger sizes than roman? I seem to remember old German books having bigger type...

For reasons that don’t apply to fraktur and that I gave in this thread, I don’t believe that sans will ever become dominant for printing extended text.

> Nobody would subscribe to this TODAY.

http://www.themicrofoundry.com/ss_fraktur1.html

Karsten, the total relativism of readability is a fallacy, and a damaging one.

hhp

“But this nicely indicates that whether roman, italic, blackletter are regarded as being more readable is a matter of habitualization rather than any form per se being more readable.”

Sounds more than plausible to me. The serif’s expected triumph in readability beyond our current time could also well be a self-fulfilling prophesy. If the serif is continually touted as superior, it will continue to be used more often whether it indeed has traits above and beyond other forms or not.

ChrisL

Yes Chris, the next time your optometrist wants to up your prescription, tell him that you’d rather just read legal disclaimers and simply get used to the small type. What better than getting one’s physical realities to promptly evaporate into the relativist ether, eh?

hhp

Was fraktur generally printed at somewhat larger sizes than roman? I seem to remember old German books having bigger type…

No. Frakturs can be small, and not break apart. I recently had a chance to photograph a bunch of old specimens, mostly from Frankfurt c. 1680. There are quite a few fraktur types there that are six to eight point, and very clear. Unfortunately, I can’t show any of these images before March.

Small Frakturs just have less stroke contrast, probably like all small type, I guess.

Schwabachers were also made in quite small sizes, incidentally. But I wonder if Texturas ever really were. Gutenberg and his successors in Mainz printed with type that I find to be quite large.

None of this is scientific. I’ve never done any real studies on this, but this is all a gut feeling.

One thing about italics, at least in the pre-digital era, was that they fitted a lot better than romans, without kerning.

This was an idea I discovered and explored in Oneleigh Italic, where I was able to design the face no kerning between lower-case letters, and only one or two Cap-lowercase kerns.

I’m not convinced this smooth fit is necessarily an advantage, as it can tend towards an indiscriminate greyness; but on the other hand, roman text is prone to occasionally exhibiting really awkward glyph combinations which can’t be good for fluent reading.

But this nicely indicates that whether roman, italic, blackletter are regarded as being more readable is a matter of habitualization rather than any form per se being more readable. [My emphasis]

Karsten is right that what people regard as more readable isn’t necessarily related to whether that form is actually more readable. I’m not sure that this is the point that Karsten was intending to make, and other respondents seem to have assumed that he meant something else, but that word ’regarded’ is key. In determining whether one form of lettering is actually more readable than another, one needs to compare speed and accuracy test results for competent and habitual readers of either form. Obviously if someone had spent his whole life reading fraktur and is suddenly presented with roman type, he is likely to find the latter more difficult to read. But it is an error to presume from this experience that fraktur is per se more readable than roman.

As I wrote above, I have little doubt that renaissance readers found italic text to be easier to read than we find it, but that is an entirely different question from whether they actually found it to be as easy to read as roman. And that in turn is a different question from which they regarded as easier to read. I might regard typeface X as more readable than typeface Y, but until I do a comparative test of reading speed and accuracy for both types, I have to accept that my perception may be incorrect.

For me the question is not whether italic is less readable, but rather the factors that make it so. Karsten notes the ’pointy’ arches. Here what is interesting about Verdigris is that it rounds the arches while keeping the letters on the narrow side. (Or so it seems, I don’t have the font.) John, I notice your Constantia Italic has small serifs for an italic—perhaps because of the demands of cleartype. Do you think that this contributes to readiblity?

Overall, I wonder which of the factors—wider letters, heavier weight, less steep angle, rounded letters, small serifs—have what impact. Anyway, thanks to all for your contributions. Any more ideas are most welcome.

[I must apologize in advance for a very long post!]

I am not an advocate of sanserifs either. Spectrum italic is really interesting. But I think it tends to being monotonous. This may relate to

Nick Shinn:
“One thing about italics ... was that they fitted a lot better than romans, without kerning. ... This was an idea I discovered and explored in Oneleigh Italic, where I was able to design the face no kerning between lower-case letters, and only one or two Cap-lowercase kerns.”

Yes! This is the game, not only for italics: keep letterforms as compact as possible and avoid elements extending too far, yet at the same time make it LOOK vivid and irregular. By the way of irregularity and readability. In “The Case for Legibility”, p.77, John Ryder shows a nice illustration of the different slants in minuscules of Jenson italic, and mentions that this unfortunately has been evened out in certain re-cuts.

John Hudson:
My use of “regarded as readable” is a bit ambivalent indeed.
On the one hand, I doubt the value of readability tests. As you also seem to imply, you can never be sure WHAT the results actually proove — that people are used to certain forms? or that certain forms are “objectively more readable” by whatever criteria? These tests can only show THAT, at a particular time and place, people read certain typefaces faster than they read other ones.
But I didn’t intend to make a difference between “scientifically provable” readability and merely “felt” readability. My emphasis was on “FORM PER SE being more readable” which notion doesn’t make any sense to me. See the following off-topic part.

This may be a bit off-topic, but ...

hhp:
“Karsten, the total relativism of readability is a fallacy, and a damaging one.”

Sorry, I am “relativist” by default. Yet I am not fond of this term — whenever I encounter it in philosophical writings, it is the last weapon used by naive-realist dogmatists. Ernst von Glasersfeld has some nice reply to that in his “Radical Constructivism” (the book’s biggest fault is its silly title). ;-)

That aside ... you don’t need to convince me of the advantages of fraktur. Yes, it is more condensed and its baroque-by-default forms may allow for more variation than roman. BUT: Give someone a text set in fraktur to read, and he will be in trouble. What IS readability? Of course you can conceive of some criteria and “measure” fraktur and roman against them, “objectively.” (Now you have to tell me what “objective” means. HAVING some criteria at all? How do you decide which criteria are the relevant ones?) But then you also have to tell the rest of the world (Germans included) that fraktur is “objectively” better and they should learn to read it — so obviously there is more to readability than criteria like letters per page, &c.
Roman typefaces of the last decades have proven that much is possible in this domain too, in terms of readability as well as expression. It is a fact: We are USED to reading roman letterforms.
(I don’t regret that. Except maybe for the long-s / end-s distinction which nicely indicated where you could divide a word, and where you couldn’t.)

The link to your website, part 2: I haven’t been at ATypI Leipzig, and only heard about the panel discussion. (Volker Küster was my type design teacher.)
I cannot help thinking that “modernizing fraktur for contemporary use” is cheating. The result will either be too much of fraktur and of no use, or not fraktur any more ... There already were such attempts, around 1930, to “modernize” textur type and bring them “up to date.” Element, National and Tannenberg. If you study these typefaces carefully you will realize that they are the exact equivalent to Futura and its contemporaries — reduction. If you look at National you’ll notice the normalized S and s and k forms, or the one-storey a (this is Futura translated into blackletter!), but also the additional vertical part to the r’s ear and the additional vertical connecting part at the e’s slope — both of which have a visual function, but can also be interpreted to utter a deep misunderstanding of textur laws of form. These are very extreme examples (in many respects) but, to me are proof that something is wrong about this approach.

Which finally leads me back to the topic ...

I consider it more fruitful to study good fraktur and textur type, or early roman types like that of Sweynheym & Pannartz, to find out if they have features which would enhance type today. For me this is not particular letterforms but very “abstract” things like a dark color (remember the complaints about Dolly on a thread next door? someone got it but didn’t understand), compact letterforms which make spacing and kerning easier (see Nick’s remark above), or a certain irregularity (like varying slants in italics). Unfortunately, Grotext is not finished yet which is such an experiment.

Hope I did not waste your time. ;-)

Karsten

“It is impossible to say anything about the autonomous typographic letter without calling to mind this historiographic falsification. Falsification is a familiar phenomenon in science. Scholars revert to it when the theory on which they have spent a lifetime threatens to be swept away.”

pp. 18, Gerrit Nooordzij, “The Stroke”

ChrisL

John,
I don’t know if you count me as one of the “other respondents [who] seem to have assumed that he meant something else” or not. I assure you that the word “regarded” is the prime factor I addressed. “Regarded” is more often what we act on than what may be indicated in any data collected. Too often people look at the conclusion drawn or postulated reasoning for the effect being measured instead of scrutinizing the way the testing was done and what we may draw from that limited spectrum. Testing that measures complexity and redundancy of saccades in types in use today can easily be “regarded” as a measure for types to come in use in the future. I don’t pretend to know what they will be any more than the blackletter reading people of years past knew would come after their time. The concept of “habitualization” has never been fully addressed perhaps because it is too difficult to measure. Those who make type selection decisions make their decisions mostly on what they “regard” as the best selection whether it in fact is or not. They may listen more to those who TO THEM are “regarded” to be the experts whether they in fact are or not.
My intent with all of this is that we not close the doors to what might prove to be a better tool or methodology because “the theory on which they have spent a lifetime threatens to be swept away.”

ChrisL

I always try to avoid the psychological aspect. Comparing readability of fraktur vs roman may be easy though — I would not consider a type readable if readers have to guess first whether a form represents an A or U, or an S or G, or a long-s or f. Comparing readability of two serif typefaces may be a bit more difficile; the measure today is reading speed, which includes questions like, we filmed his eye movement and made sure he viewed all text, but did he understand it? So it even touches the meaning of “reading.” (I know, this is also covered by text related questions to be answered.) But my point is, if by tests we find out about reading speed, do we get to know MORE than this fact, more than information about reading speed related to texts set in a certain typeface (in a certain size, with certain leading, spacing, printed in a certain color — black is not the same as black —, by a certain printing method, on a certain paper, of a certain texture, and color &c)? Similarly, what does it mean if the outcome of such a reading test is that “black on white is more readable than blue on yellow”? Which blue, which yellow, what were the lightning conditions, or the atmosphere in the room? You can make everything look scientific by systematizing factors and eliminating others. But which factors are important and which aren’t? My point is, by the sole fact that a typefaces is “read” a higher speed, which inferences may we draw, and what would be over-interpretation?

But, after Bill’s last clarifications, this turns out to be off-topic musing.

The question was easier to read, not faster to read.
Whaddaya want, the Mickey D of typefaces?
These are banal parameters with which to measure the worth of a typeface.
Why not ask whether it’s more enjoyable to read?
Whether it enriches the text more?
Whether it makes your head spin and transports you to a magical place?
But of course, that can’t be measured.

> One thing about italics, at least in the pre-digital era, was
> that they fitted a lot better than romans, without kerning.

?!
(But I’ll wait a little bit more for some other people to counter this...)

> keep letterforms as compact as possible and avoid elements extending too far

This is anti-readability.

> make it LOOK vivid and irregular.

This is artificial and not inherently supportive of readability.

> These tests can only show THAT, at a particular time and place,
> people read certain typefaces faster than they read other ones.

You are basing that merely on a preconceived ideology. Tests can
show a lot more than that. The fact they haven’t yet means little.

> Sorry, I am “relativist” by default.

I’m sorry too.

> Give someone a text set in fraktur to read, and he will be in trouble.

1) For how long?
2) Then what happens?
3) What kind of fraktur? What IS fraktur?

> The result will either be too much of fraktur
> and of no use, or not fraktur any more

Hey, I thought you were a relativist?
That’s really way too puritanical, no?
To me Design is about serving, not sermons.

> There already were such attempts

Those were junk. Not what I’m talking about.

> I consider it more fruitful to study good fraktur and textur type,
> or early roman types like that of Sweynheym & Pannartz, to find out
> if they have features which would enhance type today.

Assuming you would then actually do something about what you find, how is it different than what I’m proposing? What you will find are things that are anti-Roman. Then what will you do?

But anyway, why bother, since you’re a relativist?

> “Falsification is a familiar phenomenon in science.”

And thick blinders are a familiar accessory of relativists.
Gerrit “printing was a fall” Noordzij indeed...

> The concept of “habitualization” has never been fully
> addressed perhaps because it is too difficult to measure.

I think it’s because almost everybody who puts it on a pedestal and
worships it worries -rightly- that it will turn out to be a diversion.

The centrality of Familiarity is used as an escape by all kinds of people, from Luddite chirographers to PoMo hooligans. What’s telling and hilarious -if in a sad way- is that none of these people ever bother to actually research it!

Familiarity is difficult to measure?! Try readability!
But the real fallacy here is this classical Modernism:
“What can’t be counted doesn’t count.”

> The question was easier to read, not faster to read.

Same thing (when measured long enough).

What’s truly banal is Style, expression for the sake of erecting a designer’s nipples.

hhp

> I think it’s because almost everybody who puts it on a pedestal and
> worships it worries -rightly- that it will turn out to be a diversion.

OR:
The type of person who latches onto Familiarity is the type of
person likely to shun analysis; he will avoid true, objective
research into it, so never really discover it’s a diversion;
and maybe he knows that - and that’s an impenetrable
fortress right there.

hhp

I confess to getting a little tired of musing about readability in the abstract—which I have done my share of. Because we don’t have good tests the talk doesn’t lead much of anywhere.

I am looking not at what is tested or proven or in general true, but what people find for themselves. Eg. Karsten’s observation about Goudy and how van Bronkhorst applied the idea in another way. And at least one astute observer—John Berry—saw the difference in readability—and so do I. Others may disagree, but I am interested in what people’s eyes and experimentation tells them, even though it may not be generally true.

Someone mention Jenson’s variation in slope. I just checked Adobe Jenson, and indeed it has a lot of variation, but still manages to be pretty readable even though it has a lot of character. What did Slimbach do to pull it off?

John, I notice your Constantia Italic has small serifs for an italic—perhaps because of the demands of cleartype. Do you think that this contributes to readiblity?

The length of the serifs is related to their horizontality. If the serifs were steeper they could be longer, but my aim in Constantia was to avoid cluttering the limited pixel space in the x-height range, so the serifs are quite flat and short.

hhp:
I don’t get your point. You are argueing as if you want to counter me, but what you say actually supports me.

KL — “keep letterforms as compact as possible and avoid elements extending too far”
hhp — “This is anti-readability.”
KL — “make it LOOK vivid and irregular.”
hhp — “This is artificial and not inherently supportive of readability.”

Too quick. Let me illustrate what I mean. Nick’s Oneleigh Italic, at www.shinntype.com (left column, 4th from bottom), is a really good example. It combines both aspects:
(1) Letterforms are very compact. By “compact” I mean things like: Hooks at n m i are rather small, so the n or m fits into a square. p descender slant is very steep. f is anything but swashy which avoids collision. (If you [hhp] like it more swashy, it’s no problem to do this by calt where the surrounding letter allow it.) The t horizontal bar is short towards the right side, so keeping it withing the square, but it doesn’t hurt. The k bottom finial extends below the baseline, this again keeps the entire form within a square and avoids possible white space (and thus kerning) rightside. The g’s lower bowl does not extend much either. &c.
(2) Nevertheless — maybe because of this — the type looks very quirky. Which I find *enhances readability* (let me define: eases recognition of word images). Like, it is easy to make out p’s alone by the different descender slant, the s by its movement to the left (again to fit the square, obviously), the k by descending below the baseline. &c. There is nothing “artificial” about it and can be fully accounted for by reason. ;-)

Nick Shinn:
“Whaddaya want, the Mickey D of typefaces?”

:) Readable, legible ... there’s a distinction, but in English terminology I don’t know which is which.

Karsten

Legibility: “the ease with which one letter can be told from the other”.
Readability: “the ease with which the eye can absorb the message and move along the line.”

p. 85, Types of Typefaces, by J. Ben Lieberman, 1967. This page also has pictures of Beatrice Warde of Monotype and Paul A. Bennet of Linotype, to whom the book is dedicated. I suspect the distinction of legibility and readablity at least goes back to them.

So I think Karsten, you mean ’legibility’.

Oneleigh has its charm, but I do think readability is compromised. In the accompanying note on the face, Nick says that the italic is influenced by Goudy’s ’Companion’ typeface. Looking at Companion in Mac McGrew’s book, the stress is more regular, but the strokes less ’within the box’. For example the k descends more below the basline, and to the right, requiring kerning. Oneleigh follows other Goudy italics in the rounded shoulders that Karsten points out. I would put Companion higher on ’readability,’ though neither face is really aiming for that as a priority.

The challenge I am interested in here is how to achieve both expressiveness, which I think is to an extent desirable even in a text italic, and readability.

Adobe Jenson has a great variety of slopes in the italic, but to me is more balanced and even in color than Oneleigh. The quirky look of Oneleigh is no doubt a deliberate goal of Nick, but here I am interested in how Slimbach pulls off the even color inspite of the variety of slants. It does have a *lot* of kerning, which you could say is “cheating” with a revival of a metal face, but the resulting look is great, which for me justifies it.

Incidentally, by contrast, the italic of ’Goodchild,’ Nick’s Jenson-inspired font, is much more regular. Nick says he wanted to get away from the usually more calligraphic look of Jenson revivals. Also he gives it a large x height, so it is a different animal—more modern, as he says.

Thank you for the legibility and readability definitions!

“I would put Companion higher on ‘readability,’ though neither face is really aiming for that as a priority.”
“Incidentally, by contrast [with Jenson], the italic of ‘Goodchild,’ Nick’s Jenson-inspired font, is much more regular.”

So you tend toward the rather calm or modest typefaces. ;-) It cannot be denied that Oneleigh is expressive, yet the exaggeration makes it easy to see the principles at work. These don’t differ from the ones in Adobe Jenson Italic, though in Jenson they are much more subtle. Even in Jenson Italic, the t bar is very short, the f endings are far from being swashy. Especially note the y which tries to be compact; this becomes more obvious if compared to the swash alternate y. Part of the evenness in turn seems to result from similarly wide counters.
Another interesting thing which I see in Jenson Italic is that e.g. m n r stems are leftside curves while i or t stems are rightside curves. This may also add to (I think) legibility & thus readability.
A lot of kerning doesn’t mean cheating at all, and digital type need not imitate letterpress type. Not even in a revival. What I am interested in is to keep efforts in type design reasonable. One aspect is to avoid too much kerning, and “compact” letterforms help a lot. I see that this sounds worse than it is meant — I don’t make squarish letterforms myself and do kern terribly much.

Karsten

Thanks for pointing out Oneleigh Italic Karsten, now you made me want to buy the darn thing :-)

ChrisL

Karsten has explained my point about italics fitting better, by mentioning how the characters “fill the square”. The character forms that differ most between roman and italic are v and w. In combination with round characters, it always seems like there is too much space around these in the roman. Some digital romans address this by giving v and w negative sidebearings, not something that was done in metal. Many others, like Adobe Garamond, have negative sidebearings and also negative kerning.

The round forms of v and w in the italic “fill the square” better.

I think Oneleigh and Adobe Jenson are much different in principle. Compare the a and c in the two faces. The stress in the a and c in Adobe Jenson—based on the calligraphic italic of Arrighi—is the same, with just some variation due to the width of the a. In Oneleigh the stress on the a and c seem to me to be quite different. The more uneven stress in Oneleigh is one contributor to the quirky feeling.

>m n r stems are leftside curves while i or t stems are rightside curves.

I don’t understand what you mean here. What do you have in mind? —The lead in or serif of the mnr, including the more vertical part is thinner than the bottom of the stem; this seems to be general with italics. The slant of the i , though, seems more like the stem than the branch of the n.

> You are argueing as if you want to counter
> me, but what you say actually supports me.

Is it strange that we agree in some ways and disagree in others?
But if you’re saying that we’re agreeing completely, needless to
say I don’t see that at all.

> fits into a square

And you think this helps readability? I don’t.
And really, based on other things you wrote,
you shouldn’t either! :-/

> The t horizontal bar is short towards the right side,
> so keeping it withing the square, but it doesn’t hurt.

No? I think it hurts plenty. Horizontality and a strong
head are how the “t” supports readability. Neither are there.

> quirky ... eases recognition of word images

When applied without an understanding of
readability, it can only do so by dumb luck.

I’m the biggest fan of divergence in order to improve
readability! But the good foundation has to be there,
otherwise it’s just random noise.

> Readability: “the ease with which the eye can
> absorb the message and move along the line.”

This I don’t agree with. An essential dimension of
readability is duration/length. Without including this
any definition just won’t cut it in my book, so to speak.

> kerning ... is “cheating” with a revival of a metal face

For some reason I don’t have a problem with that sort
of thing. And few people -like Dan Carr- seem to, really.

For one thing, a discerning typographer could always file
down sorts to make them fit better (although of course it
takes too much effort to do it to pages and pages of text).

> Karsten has explained my point about italics fitting better,
> by mentioning how the characters “fill the square”.

(We’re sort of migrating across threads now, but OK.)

The big problem with this direction of thought is that
it’s the lack/weakness of serifs promoting the boxiness,
not the slant. This is even before we get into the issue
of the detriment of boxiness to readability...

The fact is that italics, due to the need to limit breakage
and hence kerns, are necessarily fitted more poorly in the
pre-digital era!

> it always seems like there is too much space around these in the roman.

The fatal flaw of this logic is the assumption that only black space
conveys information*. Not to mention the direct opposition to the
other thing in Oneleigh that Karsten points out: irregularity.
The two combine to imbalance the leveraging of notan.

* A natural symptom of creation-centric (as opposed to
user-centric) design and/or the perception that a font
is a bunch of shapes (as opposed to a notan machine).

hhp

When I wrote that “m n r stems are leftside curves while i or t stems are rightside curves,” reference is neither to serifs nor to overall slant, but to the form/shape of the stems themselves:
— with n m r, vertical stems are like this: )
— with i t u, vertical stems are like this: (

“I think Oneleigh and Adobe Jenson are much different in principle. Compare the a and c in the two faces. The stress in the a and c [...]”

When using the word “principle” I didn’t appeal to stress but to the square-trick in which respect they share a lot. When talking about stress however, you are definitely right. Btw, Oneleigh is very “Goudy” as regards stress! This kind of stress is Goudy type design humor at its best.

hhp:
“fits into a square” — “And you think this helps readability? I don’t.”

There is no direct link between fitting-into-a-square and readability. In my post I made two sections, (1) describes what I mean by “compactness”, and (2) mentions that *some* effects of trying to keep letters compact (like different slants) enhance readability (for which I gave my then-definition).
The square-trick just makes type designer’s job easier. We may not cut letters into lead blocks any more, but even with the latest and most advanced technology, the basics have not changed: Each letter is placed in a box with leftside and rightside flesh, and if you find that a particular pair would need special adjustment, you kern it. But: Kerning = hard work. So: If you have a chance to avoid it, do it. ;-)

About irregularity and readability. In his text about Today Sans Serif, Albert-Jan Pool made an interesting remark: “It [Today] looked so irregular, with each character featuring its own peculiarities. To my eyes it was sparkling with detail in all of its elements. It was one big idiosyncrasy. I showed him [Volker Kuester] my design and told him that I was striving for evenness and harmony by repeating as many forms as possible. Volker explained to me that he was doing exactly the opposite in order to achieve the same goal. This was how I started to get a visual grip on the interaction of readability and legibility [...]” In Robert Norton: “Types Best Remembered,” Parsimony Press 1993.

“But if you’re saying that we’re agreeing completely, needless to say I don’t see that at all.”

Never mind.

> *some* effects of trying to keep letters compact
> (like different slants) enhance readability

1) It seems your thoughts on “compactness” are black-centric.
2) How do you think compactness helps, exactly?

> The square-trick just makes type designer’s job easier.

Yeah, and not designing fonts is even easier.

> In his text about Today Sans Serif, Albert-Jan Pool ...

I read that in the 90s.

> Never mind.

Sure, whatever - it’s your mind.

hhp

> fits into a square
And you think this helps readability?

You’re projecting again.
I don’t mention the “R” word or the “L” word anymore.

I said that italics fitted better than romans, and qualified that by saying “I don’t know whether that’s a good idea”

> fits into a square
And you think this helps readability?

You’re projecting again.
I don’t mention the “R” word or the “L” word anymore.

I said that italics fitted better than romans, and qualified that by saying “I don’t know whether that’s a good idea”

The fact is that italics, due to the need to limit breakage
and hence kerns, are necessarily fitted more poorly in the
pre-digital era!

I don’t understand. Could you explain that a bit more?

it always seems like there is too much space around these in the roman.
The fatal flaw of this logic is the assumption that only black space
conveys information*.

In saying “it always seems”, I was recognizing that most people today, type designers and typographers, prefer to see the space around the v, w, and y tightened up. I don’t think this is a logic, it’s just what people like to see. It could even be considered a preference for a certain flavour of notan.

You’re creating a straw man here, in order to launch your “you don’t know from notan” missile.

— with n m r, vertical stems are like this: )
— with i t u, vertical stems are like this: (

Thanks, Karsten, I now see what you mean—and I do think this might contribute to the success of the Adobe Jenson italic. Nicely observed.

These features are a part of the very clever way Slimbach handles the extremes of the strokes as they reverse themselves. This way he is able to introduce a lot of ’flare’ to the strokes, which is part of the overall look of the italics, and which gives them a lot of energy but keeps them readable. The roman of Adobe Jenson has a lot of flare in the stems also, so it also unites the look, even though the models from the roman and italic are from different people.

I see that Centaur’s italic doesn’t have the flare, so perhaps this is Slimbach’s innovation. Impressive!

This is also a feature of the earlier Poetica as I see now.

Interesting. I see what you mean about Poetica. I think Poetica looks great large, but in small sizes—and not very small at that—it crashes in readibilty, whereas Adobe Jenson keeps on reading well at small sizes. Part of this is Poetica’s narrowness, which in large size gives it tension and drama, whereas at small becomes very difficult to read. Jenson is wider—the m is 25% wider than Poetica, which is a lot—so that is an important factor, but I also wonder whether the darker weight and flairing stems of Jenson, and their affect on color is also helpful.

By the way comparing Poetica and Adobe Jenson at small sizes will show anybody who is doubtful that this readability stuff is for real. In large size both are perfectly legible, in the sense of telling readily one letter from another. But at small size the difference in readability is dramatic.

As you can tell, I am a big fan of Slimbach, especially his italics, which I think are really strong.

I have found that italics with a steady and monotone rythm are harder to read, axis variation helps to reduce this effect.

Héctor

> I don’t mention the “R” word or the “L” word anymore.

Which is admirably true to yourself. Thank you.

> I said that italics fitted better than romans, and qualified
> that by saying “I don’t know whether that’s a good idea”

1) I was replying more to Karsten.
2) Again: it’s not italics, it’s sans-ness you’re actually talking about.

> Could you explain that a bit more?

Actually, my wording was quite poor, sorry.

I guess basically it ends up meaning that -in metal type- the undesirability of kerning* causes italics to be incomparable to Roman in terms of success of fitting.

* In the metal sense of the word, not the digital.

> most people today, type designers and typographers, prefer
> to see the space around the v, w, and y tightened up.

1) I would say that people always preferred that, but the physical limitations of metal made negative sidebearings not impossible but much more problematic, so it happened less.
2) I thought you were providing an example of how Roman type is spaced more unevenly than italics; but like I’ve already pointed out: that’s from the serifs, not the slant; and totally even color is totally bad (something we’ve argued about before). This is in fact something Karsten and you should talk about (if you two get over the thing I mentioned in that other thread).

> a preference for a certain flavour of notan.

Here you’re once again leveraging your disregard for the subconscious
needs of the reader by classing everything under Style. Banal to the max.

> Impressive!

On the other, the proper way to do that is via linguistic
analysis, not the context-free formalism of shape congruence.

> will show anybody who is doubtful that this readability stuff is for real.

But they have to be looking. I don’t mean occularly, but psychologically.

hhp

your disregard for the subconscious needs of the reader

I care very deeply about the subconscious needs of the reader.
I’m a caring kind of type designer.

I don’t believe readers have subconscious needs; frankly, I’m far from convinced that there is any such thing as subconsciousness as a state distinguishable from unconsciousness. There are physiological and cognitive functions in reading of which the reader is unconcious during the reading act, just like we are usually unconscious of the functions of breathing. But as with respiration, we can apply our conscious mind to reading if we wish, and like thinking about breathing, it doesn’t help the process. :)

The other thing to note is that there is a huge amount of stuff of which we are unconscious that we will forever remain unconscious of, because it is outside of our sense perception. Again, like breathing: no matter how consciously we think about the act of breathing, we have no consciousness of what individual oxygen receptors in the lungs are doing, because we cannot sense them individually. Similarly, in reading, we are actually unconscious of how we recognise words in the brain, and no amount of conscious effort will make us conscious of that process. We can study what the brain does during reading, and we can make guesses about the process it uses based on tests designed to reveal things about the process, but that is not the same thing as being conscious of the activity. So if you want to talk about non-conscious needs in reading, call them what they are: unconscious, and beyond the grasp of consciousness.

Leave the ’subconscious’ to the Freudians.

“I would be interested in people’s views on what makes for a readable italic.”

I did an interesting exercise. . .

I became curious about this issue first because people who generally don’t know what they are talking about typographically, (which always leads to good avenues of investigation), “decided” that low contrast fonts were better for the screen than other fonts, presumably better than those of high contrast. After nodding along for while, I did some experiments that proved this low contrast bias is not only partly incorrect, but a really bad idea in some cases ;)

What convinced me to investigate are three typographic truths: 1. that low contrast is not a good idea in the branches of letterforms, (they obviously nust taper toward the main stems), and 2. that low contrast serifs along our Latin baseline provide little or no help in reading, in fact they may well clog up the system (place argument here). And 3. I believe that in the brave new hintless world of CT, there has to be a way to make Romans and Italics that work for both print and screen... So, in pursuit of an answer, and because it don’t lay in Sans, I drew Aldine Italic where the branching is seriously stressed, there are many fewer serifs along the baseline and, at least in Aldine and Kis Italics, contain more flexible topologies with which to experiment.

Along the way in this investigation, I had to pull apart “our” definition of Italic and reconstitute it, for myself at least as, “the styles derived from the Aldine italics,” and not the modern, san serif, or those Italics made to serve as very occasional companions to a Roman. So I drew. and drew, Andrew, (a decent font name, no?) What I found, is that Italics whose main strokes are lighter than the x-ht features, (and lighter than the main strokes of a Roman if companionized), and Italics with branch strokes and serifs at the same angle, or with serifs only1-2 degrees less acute, but not much more, and using the unique features I could, e.g. a “k” with a good loop at the top of the x-ht, as opposed to a curly ball or a plainer serif... all really help reading, assuming you have not chosen bad angles.

And, perhaps more important to a future chalk full of font fuzz, you can easily design an italic like this that does work with Roman in screen and print applications even with the dumbest rasterizer, and zero hints, proving the paraphrasing of a Jurassic park theme that type will find a way. Otherwise, I don’t know how to make a series of weights, with matching Italics, for printer and screen, where screen = CT, and I don’t mean Connecticut.

Wow, I haven’t had to re-read a post so many times for a while...

In 1-bit onscreen rendering, there is of course a threshold below which [too] many features become the same weight. That said, I do think the threshold MS adopted with their 90s core fonts was off by a couple of PPEMs - this because they took users’ stated preference for “light & airy” type too seriously, ignoring the difference between deliberation and immersion. And this couple of PPEMs makes all the difference.

In gs onscreen rendering, I definitely agree that we can use more contrast than we’re being given.

> low contrast is not a good idea in the branches of letterforms

Yes, even Renner accepted that. But could that not be classed
as an optical correction, as opposed to an “intended” feature?

> low contrast serifs along our Latin baseline provide little or
> no help in reading, in fact they may well clog up the system

Wow, an idea new to me. Where was it hiding all these years?!
So you’re saying that you want baseline serifs to be light.
I wonder if that’s what people like Fleischmann were thinking
back then too... Fascinating.

> in the brave new hintless world of CT, there has to be a way
> to make Romans and Italics that work for both print and screen…

Here I’m not making a [good] connection.
Do you mean better than what MS has done?
If so (or otherwise) could you elaborate?

> there are many fewer serifs along the baseline

Now this is getting seriously interesting...
So we’re talking semi-serifs here? This is an area that
I’ve long thought needs much more serious attention then
has been given. And you did this -instead of making the
foot serifs lighter- because of screen constraints?

> I had to pull apart “our” definition of Italic and reconstitute it

Bravo.

> Italics whose main strokes are lighter than the x-ht features

Wait, you’re talking about stems thinner than like the top of the “o”?
Reminds me of a feature in Mandel’s Colorado, although chances are
there’s no real synergy.

> Italics with branch strokes and ...

From here on I’m pretty sure I’m lost.

David, pitchurz please!!

hhp

I have it in mind to design and cut an italic for “Sonnets From the Portuguese”. The size would be 16 point and since I will be making it to fit the composition system of the Monotype machine I will be restricted in that there will be fixed unit values because I will be using an existing device (a wedge) which creates certain character widths as they are cast.

I have my choice of one for 14 point which will make a narrower set of units widths for 16 point, and one for 18 point which I think will make the face much too wide; so I think the former is best.

This should be an interesting project, and I am aiming for a very verticalish italic; something like 1 degree, or perhaps even absolute vertical, depending on the italic form to make it italic.

I doodled quite a few characters on the vertical and there is a tendency for it to look almost like it’s falling over to the left!

I have always felt that the less slanted italics were the most readable, but they are in many cases quite narrow and tightly fitted. This can be intimidating to wade through as extended text, but then I’ve always been one of those people who tend to scan over italic inserts of any length. Don’t know why, because I like the grace of italics.

I love the work of FW Goudy, but I’ve always found Deepdene Italic has a tendency to pack up in printing by letterpress, and I’ve seen quite a lot of litho work where the same thing happens. When we made digital Kaatskill at Lanston Type Co. I redrew Deepdene Italic to a wider setwidth and brought it into line with the Roman. In the two printings Goudy had a hand in using Kaatskill, he cast Deepdene Italic to align with it, because he had no budget to make a paired italic. It really didn’t work. I think the Lanston pair are a little better. I want to make it clear that there was precious little of design on my part in either Kaatskill roman or “Kaatskill Italic”, since this was a simple redrawinbg of the work FWG had done, with some scaling. Anyway, I think it’s a friend italic in its present form.

Jim

> there is a tendency for it to look almost like it’s falling over to the left!

Due to the semi-serifs. BTW, I doubt
one degree would be enough; the slant
in Rialto is the least I’d apply myself.

hhp

Jim, I have a 1957 version of the Elizabeth Barrett Browning sonnets you mention in a Stichting De Roos (Dutch—bibliphile society), set in Palatino Italic, at a rather small type size. Henk Krijger (the designer of the Raffia Initials) was the designer, and provided it with an EBB monogram. The version might bear you out in your comment on ’less slanted’ italics.

I’ll post some images this evening.

David, thanks for the fascinating ruminations on italic, which I think I even understood.

Your idea that the x-height features need to be heavier than the stems applies to quite a few italics, I believe, including Verdigris and Adobe Jenson.

The serifs on an italic lower case do not define the line in a way that the roman serifs do, so I can see that dropping them might not affect readability much. On the other hand, I would think that it would be a problem keeping things in visual balance...

Jim, Spectrum’s italic, mentioned earlier, is I believe also rather upright, though not what you contemplate. Perhaps it does contribute to its surprising readibility, given how ’swashy’ and high contrast it is.

> The serifs on an italic lower case do not define the line in a
> way that the roman serifs do, so I can see that dropping them
> might not affect readability much.

But why not have full serifs in italics?
(Not that I think “defining the line” is the relevance of serifs to readability.)

hhp

Greetings.

Pierre Haultin’s relatively obscure italic attracted me for its strength and readability. It started the Verdigris project some ten years ago when I purchased a book from 1584 set almost entirely in it. Haultin’s italic was the basis for Verdigris italic.

I’ve yet to find a type specimen showing Haultin’s italic. After buying a large number of facsilmiles without success, I turned to Matthew Carter, who confirmed my hunch that I was inspired by Haultin.

You’ll find a sampling of it in Updike’s Printing Types (unidentified) in a title page printed by the Giunta.

I’ve always loved Sabon, but find the italic — forced onto roman widths — annoying. Verdigris was influenced by Sabon. For Verdigris I wanted the same bookish, workhorse kind of a face, but with nice proportions, using the Granjon ’Garamond’ as model for the roman (it has distinctive contrast in lowercase widths), Pierre Haultin’s italic for the italic.

Open counters and moderate-to-low stroke contrast, with slightly irregular rhythm, are what make an italic work for me. Avoiding delicacy and prettiness was deliberate — I wanted to hint at little pieces of metal smashed into paper (italic TYPE) rather than something handwritten. I ditched Haultin’s swashes I could have employed, and opted to ’modernize’ certain forms to avoid quaint affectation.

Goudy, whose types I once admired (and now can’t stand), had nothing to do with Verdigris’ inspiration.

Kerning is overrated. If faces could once deliver such even text in metal with a handful of kern pairs, well-designed forms can do same digitally. I try to design forms that fit well together first, then we kern. One must still decide whether to kern for text or display. Faces that are over-kerned to look good as display behave poorly as text. Optical sizes are a good thing.

Mark van Bronkhorst

> little pieces of metal smashed into paper

Ideally not smashed. Smashed, on Jennifer Lopez’s wedding invitation.

> Kerning is overrated.

Type design is over-rated (by us).
But once dedicated to it, it’s not only justifiable to worry a lot about kerning, but it’s generally commendable. Luc[as] de Groot has pointed out that kerning often gets shortchanged on this side of the Atlantic (not least by expats). I agree, and see that fact as a cowboyism. There are good things about cowboys, but there are reason they’re better at wrestling livestock than designing the little machines that are fonts.

> If faces could once deliver such even text

They couldn’t.
(Especially not serif ones with small x-heights = book faces.)

> I try to design forms that fit well together first, then we kern.

On the other hand, ideally, not only base spacing, but the black
forms themselves are often affected by the prospect of kerning.

hhp

Mark, thanks for your very interesting account of Verdigris, and your views on italics. Did you also change the angle of the slant of Haultin’s italic in making Verdigris italic?

The angle is pretty much as Haultin had it.
MVB

The angle is pretty much as Haultin had it.
MVB