Diaeresis/umlaut height

Ori Ben-Dor

Some fonts have them at the same height as the dot of the i, while in others fonts they're lower.
What's your approach and why?

Nick Shinn

I generally make them a little smaller and lower, not least because of combinations such as Tü and fü, which occur in German and other languages. 
 

Vasil Stanev

For a font with 700pt cap hight, I place the bottom of the acute 50 pt above the cap, and as tall  as the design needs it to be. The dieresis is in the middle of the acute hight, except when the design needs it to be lower or higher. These basic metrics vary according to the overall vertical metrics of the particular font, case, and other factors, most importantly the design style. 

Florian Pircher

When reading text in German I like the umlauts to be of the same size and height as the i-dot. Text looks clumsy when an i or j is close to an ä/ö/ü and the dots of the two letters are of different size and position. I just read a book set in Bembo and it bothered me on every page ;‌)

Of course, it depends on the design of your typeface. A display typeface has much more freedom in this regard, at least for German. Also, if you use triangles or diamonds as dots, try to avoid those for umlauts and keep them as dots or at least closer to a round shape. For example, Iowan Old Style does this:



I would try to use the same size and position as i-dot or dot-above mark. If common pairs such as /f/udieresis touch or are very close to touching, consider making the dots smaller. Use repositioning/lowering of the dots as a last resort.

John Hudson

I tend to align the bottom of the diaeresis dots to the bottom of the i/j dot, but make them slightly smaller in most weights. In heavy weights, the diaeresis dots may need to be quite a bit lighter and narrower than the i/j dot.

Craig Eliason

Florian Pircher said:

Also, if you use triangles or diamonds as dots, try to avoid those for umlauts and keep them as dots or at least closer to a round shape. For example, Iowan Old Style does this:

Interesting. Would you advise the same for the diacritic on letters like /cdotaccent and /zdotaccent? (I notice Iowan Old Style uses the rounder umlaut-style dots for those also.) Which also provokes the question whether /i and Turkish /idotaccent should have different dots...

Ray Larabie

My dot accents are aligned with the dieresis. The i and j glyphs are generated using anchors and collision checked along with all the other dot accent glyphs. In some heavy /condensed weights, the dieresis might have narrower dots. I don't see any reason to make them differ except in cases where historical accuracy is required. The way I see it, if a dot accent is good enough for c, e, g, n etc. it should be good enough for ı and ȷ.

Florian Pircher

Craig Eliason said:

Interesting. Would you advise the same for the diacritic on letters like /cdotaccent and /zdotaccent? (I notice Iowan Old Style uses the rounder umlaut-style dots for those also.)

Yes. The design of the dots on /i  and /j are often designed separately from the dot-above-accent and diaeresis/umlaut. Here is Iowan Old Style again:

You can find this treatment in other typefaces such as Expo Sans (and Expo Serif):

Or even in reverse, such as in Adobe Jenson:

I do not like this particular design decision and can not imagine that it would be good for long-form reading. Maybe the design references historic norms that I am not familiar with.

Whether the transfer of the i/j dot to other diacritics is successful depends on the degree to which they deviate from a circle. Since the i/j dots of Arek merely hint at a diamond shape, they still look good on other letters:

Warnock features a much more pronounced diamond-shape. It may work for long-form reading, but I doubt that it is a good choice for languages that use dot-accents or diaeresis. The umlaut-dots are, however, far apart and on the same level as the i/j dots which I like a lot compared to the Arek positioning.

Taking it to the logical conclusion, Maiola uses pointy shapes for both i/j and umlauts. Looks great in display settings. I can also see myself reading a short poem set in Maiola. I like this typeface a lot, so I will not say anything about my feelings regarding setting more than one or two pages.

Here are some more Slimbach designs showing how typefaces with calligraphic influence can address this issue. First up, Brioso uses subtle adjustments to make dot-accents and umlauts more dot-shaped:

Even more subtle is the approach in Sanvito. Here, the dot-accent follows the i/j dot and only the umlaut differs (which could be slightly higher, but the adjusted dot-shape looks great).

Poetica uses circles for the umlauts. (/edotaccent and /zdotaccent are not available.) The design is readable, if somewhat uninspiring. Here, a localized glyph-set could be used to substitute the three german umlauts with versions that look closer to a double-acute (“Hungarian accent”) since that references the predominant handwritten style of german umlauts.

Finishing off, Caflisch Script uses distinct forms for the i/j “dots” while the umlauts and dot-accents stay muted, low positioned discs. I like this a lot more than the Poetica design since muted dots fit in better in a mono-line typeface such as this one.

Craig Eliason said:

Which also provokes the question whether /i and Turkish /idotaccent should have different dots...

Great question! None of the fonts presented above change the visual appearance of /i, /j, the umlauts nor the dot-accented characters when switching to Latin script with Turkish language. Maybe some of them should. I do not know, but it would be interesting to hear a native speaker’s opinion on this.

Ori Ben-Dor

Thanks everyone for your input!

Nick Shinn

Type designers have to make up our own minds.
These subtleties are imperceptible to the end reader, even those of Caflisch Script.
The people who care are art directors, graphic designers and typographers, those are to whom we address our nuances—and one another.
Certainly, native speaking professionals will have their opinions, but short of a broad scale survey there is unlikely to be consensus; every culture has its progressives and its conservatives, its differences of taste and opinion.
The thing to do, as a type drawer, is to study the field—historical and modern, native and colonial, megacorp and indie, bland and outrageous—and identify the norm and the outliers, then position one’s own design in relation to what one perceives as its “comparables”. But most importantly for the dieresis, consider how it will interact in worst-case scenarios with adjacent characters in real words, in the typeface you’re working on.