Width Reduction In Sans-serif Italic Styles

Michael Jarboe

I'm curious to hear thoughts on the idea of width reduction in sans-serif italic/oblique styles. Not so much with serif types as they are naturally reduced in width due to their root in cursive letterforms and handwriting.

I've observed in many sans-serif typefaces that the italic styles are often reduced in width. Process wise it's likely this width reduction occurs before slanting and can often be as little as 95-97%. I understand that optically, this allows the italic styles to closer match the width of the romans being that slanted letterforms are naturally elongated and wider than their roman counterparts.

My curiosity is peaked though as I believe that not all designers utilize this technique, so I've been debating the reasoning of various points of view. Understandably, I don't believe I've seen this technique utilized in any truly condensed typefaces, nor have I seen it in any monospace typefaces. Maybe it is utilized in monospace sans-serif designs that utilize a more true italic where the letteform construction varies greatly from the roman?

For condensed faces of course there is a point of no return wherein the counters are so narrow that it wouldn't be realistic to further reduce the width of the letterforms in an italic. Similarly, the non proportional sidebearing space would increase, albeit subtly, if this technique is used in a monospace face.

Since I have multiple condensed and monospace typeface designs in development, I've found myself at odds with this theory, being that it doesn't seem applicable at large.

George Thomas

It is a natural geometric occurrence, so I follow that rule.

Nick Shinn

When a vertical stem is skewed, the area remains the same but the thickness decreases.
So “the amount of ink on the page” remains the same, which gives the impression, en masse, of the same weight in roman and italic, which is more significant than any suggestion that the italic is lighter because of its thinner stem.

For a typical slant of 8°, the stem thins by around 2%, which, if it is even perceptible, is no doubt desirable, as our eyes are accustomed to expect such angled strokes to be thinner. For two reasons: (1) Because vertical stems look better thicker than horizontal, and angled is on the way to horizontal, and (2) Because of our habituation to seeing typographic strokes angled that way being thin, derived from the “broad nib” contrast of traditional serifed type, apparent in A and V etc., and often carried over into sans types with ostensibly even stroke thickness in those letters.

Chris Lozos

as is typical for type design, the Eyes have it.

Igor Petrovic

One more point worth considering is how one sees the purpose of italics in a font. I remember a topic here on TD discussing italics, and the takeaway for me was that there is no clear consensus should italics try to be "as similar as possible" or "kind of different" compared to upright.

The classic purpose of italics is to emphasize a part of the running text. So it makes sense to be a "kind of different". But usually in a way that emphasizing using regular-italics is slighter than using bold-upright weight.  

On the other side, if the font is more display than text, and more oblique than true italics, maybe it makes sense to be "as similar as possible".

So to some extent, it's a matter of decision/concept I guess.

Nick Shinn

I checked all the sans types I’ve designed, and the only one with noticeably different italic widths is a humanist style, in which many of the letter forms are categorically different from the Roman.

Sometimes the non-humanist italics came out slightly different in width from the roman (Chris’ “eyes have it”), and sometimes I made a point of making them uniwidth.

But the humanist italics were a horse of a different colour.

Linus Romer

@Nick Shinn

For a typical slant of 8°, the stem thins by around 2%

Wouldn't that be 1% for a vertical stem, because cos(8°) = 0.99?

Nick Shinn

I didn’t do the math; I skewed a rectangle in FontLab, and used the measuring tool to approximately determine the new thickness.

(This was the method that occurred to me, easier than remembering where my “Four Figure Tables” book was, or looking up a trig value online, which I‘ve never done—or remembering how to perform that kind of calculation anyway, which I haven’t done for 50 years.)

Michael Jarboe

Thanks for the insightful replies. I think this is a case where I want to stick with the rule I've internalized and utilized, yet have found it isn't really absolute, or applicable in all cases.

Hrant Հրանդ Փափազեան Papazian

BTW much better to use a ratio of integers for slope rather than degrees or percentages.

Ray Larabie

@Hrant H. Papazian
How does the reader benefit from geometric purity of slants?

Hrant Հրանդ Փափազեան Papazian

@Ray Larabie I guess the same way the reader benefits from us moving a node one unit.  :->

It's actually nothing ideological (for a change :-) but simply about the practicalities of working on a grid (versus polar coordinates for example). Although you could say that can get ideological pretty quickly... Previously on One Font To Give:
https://typedrawers.com/discussion/comment/49687/#Comment_49687