Regular and italic with the same character widths

Sander

Hello all,

I am designing a sans serif typeface and want to apply the same character widths for the italic as I used for the regular font. What are the advantages and disadvantages of this decision? Are there specific problems that I will come across? And are there publications about this specific subject? 

Piotr Grochowski

If the sans-serif typeface is monospaced, then of course this is the right choice. While many designers struggle to keep the glyphs optically corrected in the same width, some type designers have no problem with making thousands.

Sander

@Piotr Grochowski Thanks for your reply. My typeface is not monospaced, so I'll have to concentrate on the visual balance.

Raphaël Ronot

I don't have any publication in mind, but you can look at Sabon for reference. It's a good example of a font with identical metrics between italic and roman.

Craig Eliason

Why do you want to do that? 

Hrant Հրանդ Փափազեան Papazian

Uniwidth is a wonderful design feature, especially in this screen age. And if the Italic is a variable axis, it becomes even more crucial.

Hopefully your Italic is properly subservient to its Roman (versus being a flighty calligraphic primadonna) which would make things work out much better.

John Hudson

I tend to favour wide italics — unless specifically working in the style of renaissance types —, but would avoid committing to matching specific widths of roman letters unless doing a strict sloped roman design. Be flexible.

Scott-Martin Kosofsky

Sabon is, indeed, an especially good example, It was released simultaneously, in 1967, for metal casting machines by Linotype, Monotype, and the Stempel Foundry (for handsetting). It was the Linotype system of duplex matrices, in which the roman and italic letters were contained on the same matrix, one above the other, that drove the design of roman and italic at the same width.

Nearly all types designed for metal-casting Linotype (and later photo machines) shared this feature. Therefore, you could look to any of the early Linotype (and Intertype) catalogues for inspiration and guidance for designing romans and italics of equal widths. You’ll found the work of some remarkable designers, including W.A. Dwiggins, Rudolph Ruzicka, and Chauncey Griffith, who was the head of typographic development.

Hrant Հրանդ Փափազեան Papazian

Yes, be flexible in annulling the Aldine shotgun-wedding with an arbitrarily chirographic subordinate style.

Sander

Thanks all for your replies. I will have a look at Sabon. Another inspiration for my choice is Fago.
There are several reasons to use the same widths for (almost all) characters. It makes the spacing of the italic easier :-) But also for typesetting it can be handy, when you can change regular into italic and vice versa without changing the design of a text., for instance in a bilingual publication. 
The italic won't be a sloped roman and I realise that for some characters the width of the italic probably has to be different. But the equal widths approach intrigues me, so I give it a try. Why not!

Hrant Հրանդ Փափազեան Papazian

There's a compromise option, which I've actually used with the outlying weights of my Nour&Patria system, the middle weights being uniwidth: make it what I call "fixed-offset", where the Italic's characters are all the exact same amount narrower than their Roman counterparts; so a tracking adjustment (hopefully a small one) would make the Italic become uniwidth with its Roman.

This would of course require a user of the typeface to be aware of this, and willing to make the effort to apply it when desired.

Sander

@Hrant H. Papazian Good idea. That can be an escape route in case the equal widths result in a disbalance between regular and italic. 

Piotr Grochowski

Hrant H. Papazian said:

There's a compromise option, which I've actually used with the outlying weights of my Nour&Patria system, the middle weights being uniwidth: make it what I call "fixed-offset", where the Italic's characters are all the exact same amount narrower than their Roman counterparts; so a tracking adjustment (hopefully a small one) would make the Italic become uniwidth with its Roman.

This would of course require a user of the typeface to be aware of this, and willing to make the effort to apply it when desired.

At some point the rounding errors could go wrong. They definitely will go wrong in a reflowable environment, and they might go wrong in scalable environments if the internal fractional precision is incompatible with the font unit at some point size.

Craig Eliason

Sander said:

There are several reasons to use the same widths for (almost all) characters. It makes the spacing of the italic easier :-) 

It won't make the spacing of the italic easier, it will make it harder, because you're introducing an additional arbitrary constraint to have to satisfy. 

But also for typesetting it can be handy, when you can change regular into italic and vice versa without changing the design of a text., for instance in a bilingual publication. 

I don't really follow why this would require letter-by-letter matching of spacing. I can see why a bilingual publication might want blocks of italics to appear equally spaced, but I think you'll find that that effect is best achieved by varying widths. 

The italic won't be a sloped roman and I realise that for some characters the width of the italic probably has to be different. But the equal widths approach intrigues me, so I give it a try. Why not!

Because italic letters are going to demand their own treatment. When you introduce slope, then suddenly an /s spine is shorter, but a /z diagonal is longer. An /o no longer has its side extrema halfway up the x-height, but now has a lower extremum on the left and higher on the right. In other words, the shapes change so the spacing will want to change. 

Respect Sabon for its ingenuity, sure, but recognize it's drawn that way not because it's a good idea for an italic, but because the Linotype machines forced it to be that way. 

Sander

Thanks @Craig Eliason. That are issues I didn't realize yet. Are there publications about the italicisation of typographic characters? It would be helpful to have some examples of the visual corrections that are necessary.

Craig Eliason

Your eye needs to be the guide. Do it the same way you'd space any font: establish control characters (/n/o/H/O) with good spacing first; then put other letters in the midst of those to set spacing; get all sidebearings set before starting any kerning.

Hrant Հրանդ Փափազեան Papazian

Craig Eliason said:
It won't make the spacing of the italic easier, it will make it harder, because you're introducing an additional arbitrary constraint to have to satisfy.

Unless the Italic is properly subservient to the Roman. Yes, there will be tension, because modifications are inevitable, but maybe it's not really "arbitrary", because:

Because italic letters are going to demand their own treatment.

Maybe they shouldn't, at least not in violation of our old shared maxim: a typeface is not a collection of beautiful letters, but beautiful collection of letters. So they should request, but cannot be allowed to demand.

Hrant Հրանդ Փափազեան Papazian

Piotr Grochowski said:
At some point the rounding errors could go wrong.

Interesting point. I do have to wonder: since the rounding should be less than 1/2000 (½ of 1 em unit) per character, how long would a line have to be to actually break differently?

Hrant Հրանդ Փափազեան Papazian

Sander said:

Are there publications about the italicisation of typographic characters?

Whatever you find is likely to be along the official-party-lines of Italics being more calligraphic than the Roman, which means greater deviation in width, going against your ambitions. You might instead look at actual successful slanted-Roman designs (even if people tell you that's an oxymoron :-) including these two recent ones:
— https://typographica.org/typeface-reviews/berthe/
— https://typographica.org/typeface-reviews/empirica/

Sander

@Craig Eliason Even if the eyes are the guide, there are two ways to adjust the italic. One way is to adjust the character shape and then adjust the character width. The second option is to adjust the character shape within the restrictions of the predefined character width. Perhaps the result of the second approach is not an italic that looks perfectly as what we are used at, but that doesn't mean that it is wrong. 
@Hrant H. Papazian Thanks for these two links. In particular, Empirica looks interesting related to my project.

John Hudson

Hrant:

Yes, be flexible in annulling the Aldine shotgun-wedding with an arbitrarily chirographic subordinate style.

To be fair to Aldus, his use of italic type was not as a subordinate style; it was just the normal text style of the kind of small format book that he was producing. He very seldom used roman and italic together, and always as complimentary styles, to mark distinction between different kinds of text, neither subordinate to the other.

The use of italic embedded within roman text is a convention that evolves gradually, but I've not seen any evidence that the editors, printers, and publishers who did this thought this implied italic was a subordinate style to roman. I think that is a much more recent perception, and has to do with the creation and marketing of coordinated typeface families under a single name and, much more recently, the linking of these families in functionally stylistic sets in which the roman is considered ‘regular’.

_____

Craig:

I can see why a bilingual publication might want blocks of italics to appear equally spaced, but I think you'll find that that effect is best achieved by varying widths. 

It is precisely for the case of bilingual texts that I favour a fairly wide italic, so that both styles have a similar visual presence. As a case in point, the Brill italic actually sets slightly longer than the roman.
This is what I mean by being flexible. Fitting different letterforms to the same widths is an artificial constraint that necessitates compromising the shapes; the Linotype Sabon italic o is a good example of such compromise. Unless there's a specific technical need to match the widths, e.g. for Linotype machine composition using duplexed matrices, why impose such a constraint?

Craig Eliason

Sander said:

@Craig Eliason Even if the eyes are the guide, there are two ways to adjust the italic. One way is to adjust the character shape and then adjust the character width. The second option is to adjust the character shape within the restrictions of the predefined character width. Perhaps the result of the second approach is not an italic that looks perfectly as what we are used at, but that doesn't mean that it is wrong. 

If your adjustments make the the letter worse for no good reason, that is wrong in my book. 


For a parallel, look at the /a of Arial. Next to its neighboring letters in the word above, it looks stretched wide. It's a pretty ugly letter to my eyes, certainly more disharmonious than the /a of Helvetica above. Now Arial was designed with predetermined advance widths--they were taken from Helvetica so that the font could be subbed in without changing textflow. But perhaps to avoid looking too copy-cat-like, the Arial designer shortened the little tail at the bottom right. This though meant that to keep the advance width and also not open up too much space on the right, the whole right stem had to come rightward, in other words widening the letterform. As a result Arial's /a fits its advance width and is well spaced, but appears too wide in shape.

A big part of this is understanding that interior space of letters is a component that needs to be kept in balance too; it's a bad idea to stretch or compress letters with only the interletter space in mind. 

Again, Arial's /a, like Sabon's italic /o, are reasonable solutions to a design brief that offered no flexibility in advance widths. But it seems pointless to me to adopt that inflexibility for no good reason. 

Cristóbal Henestrosa

Again, Arial's /a, like Sabon's italic /o, are reasonable solutions to a design brief that offered no flexibility in advance widths. But it seems pointless to me to adopt that inflexibility for no good reason.

Agree. For almost every situation, at least. It might be a good exercise though, even if it is only for concluding it was not a good idea in the first place.

On the other hand, I can certainly see it as an interesting feature when you want to change the style from Regular to Italic (and even Bold!) without any reflowing, just as Asap does. I am guessing this would be valuable if you are in a hurry.

John Savard

In the early days of Roman and Italic typefaces, Italic was much narrower than Roman. So making each Italic letter exactly as wide as the corresponding Roman letter will mean that if you were trying to do a Baskerville, a Caslon, a Garamond, or anything earlier from the Aldine or Jenson eras, your typeface would be inauthentic. In the case of the Jenson era, of course, even having an Italic is inauthentic.

For later typefaces, however, rough equality between the Italic and the Roman is the norm. Does that mean there's no problem?

Not quite. Ideally, the width of every letter in a typeface should be set to an optimum value based on its shape. Since the Italic is somewhat cursive in character, some of the letters will have different shapes from the Roman (of course, a slanted Roman doesn't have this issue) leading to the ideal width being significantly different. Specifically, an Italic l (lowercase L) is about the same width as an Italic t, while in the Roman the lowercase L matches the i.

However, every typeface for the Linotype which places Roman and Italic on the same matrix has to make them of equal width; thus, type designers have had a lot of practice in doing the necessary fudging of their designs for this. So the cost is small, however, I still concur with Craig Eliason that it is pointless "to adopt that inflexibility" without a good reason - particularly as they aren't using hot metal for printing any longer, making its mechanical limitations a thing of the past.

Sander

Thanks @Cristobal Henestrosa for pointing at Asap. In that typeface the constraints are even bigger, matching the widths of the bold with that of the regular. I was looking for such examples to see how other designers solved certain problems.

@John Savard There is no mechanical limitation nowadays which forces you to use a limited number of widths. But why should the shape be leading in a design and not the width? @Craig Eliason The Helvetica-Arial example shows that within the same width two letters /a can be designed from which one looks better than the other. So, the main problem of Arials /a is not the width but the design. 

John Savard

Sander said:

There is no mechanical limitation nowadays which forces you to use a limited number of widths. But why should the shape be leading in a design and not the width?

That's an... interesting question.

If you mean to point out that this is an unexamined assumption, you are right, so I will point out its basis.

In the usual case of typeface design, the designer wishes to craft letters that will work well together in text - and look a certain way. The shape of the letters is precisely the material that is manipulated to achieve those goals. The spacing of the letters is then something that is derived to serve the letters with their given shape, so that the words they form will be in harmony.

One might impose some constraint on the character widths, and adjust characters to fit that constraint, for any number of reasons, but that's not the typical design flow of a typeface.

Sander

Thanks @John Savard for your argumentation. I agree that starting with letter shapes and subsequently defining side bearings is the way we are used to think about designing type. But as @Hrant H. Papazian cited we want to design a beautiful collection of letters. So, the beauty is not only created in the individual shapes, but also in the relation between them. And that relation is (partly) determined by the character widths, especially when they are limited. 
Taking predefined widths as the starting point for a (in this case italic) design, urges you to look at the shapes from a different point of view. It can cause problems, but it can also result in modifications that you would not have considered when you work in the traditional way.

Nick Shinn

I’ve been investigating this, Sander.
It started when I made a monowidth, unicase alphabet system/typeface, Panoptica.

Several people used it with kerning, so I thought “Why not?”
And in fact, applying “Optical kerning” in InDesign was eye-opening.

Going further, I discovered that the squished characters, notably M and W, are not inherently low-readability. In fact, they are quite distinctive (easy to decode), in a way that widening them out into a general greyness diminishes. 

top: Monowidth
middle: The monowidth glyphs, with “normal” sidebearings and kerning, for even colour.
bottom: Glyph shapes adjusted for even more even colour (wider H and M, narrower F and T). 

I surmise that the “ugly” versions with dense M etc. may well be more legible than the “proper” version, although perhaps not as pleasant to read.

nb: Tracking closed up and width equalized for a better comparison of the differences.

Sander

Hi @Nick Shinn. I doubt if the monowidth M is more legible than the "proper" one when the text is longer than just a few lines.
Monowidth fonts are an extreme case of inflexibility, so they are a good source of information for me to see how the italic and regular may be combined. 

John Hudson

Nick,

Going further, I discovered that the squished characters, notably M and W, are not inherently low-readability. In fact, they are quite distinctive (easy to decode), in a way that widening them out into a general greyness diminishes. 

Would that seem the case, I wonder, if the type were not so tightly spaced?

Nick Shinn

That’s very pertinent. I do think that kerning the monowidth glyphs makes for an interesting effect, especially when tightly set, and the proportional version looks better open, like a “small caps” paragraph setting. So when I publish this family, I will default the “monowidth kerned” version to a tight set, and the proportional version more openly.

To return to the original topic of this thread, neither of the italics of these variants set to exactly the same width as the romans, despite being slightly tweaked slants of the same letter shape, for the reasons Craig mentioned above.