Proper way to draw µ

Type Minds's picture

I'm wondering how to correctly draw the Greek character μ. Upon inspection, many fonts' μ's look like a u with a leg pasted on like this:

This doesn't look right to me, but I'm new to this stuff. Is this how it's supposed to work? It seems like there should be some stroke thinning where the leg descends.

oldnick's picture

Is this how it's supposed to work?


Type Minds's picture


clauses's picture

In most (latin script) typefaces these greek glyphs are for use within latin script text eg. 10µF. In these cases they can follow the design of the latin (general ductus metaphor). They have however the same unicode indices as the proper greek glyphs, so if the font is expanded to include the greek script they will have to be redesigned, or a separate greek font will have to be developed.

quadibloc's picture

To be more informative:

The Greek letter mu does not have an agreed-upon way to represent it as a letter with the characteristics of a letter from the Roman lower-case alphabet - that is, with serifs.

While there is a conventional Greek upper-case which is built from the same basic shapes as Roman capital letters, the traditional Greek lower-case is cursive.

This has led to what is known as "Porson Greek". This is the type of Greek typeface with which people outside of Greece are most familiar. It looks like using a normal Roman face for capital letters, and an italic face for small letters.

This is appropriate when one is using Greek letters in mathematical formulae. It has been used for text, simply because, outside of Greece, printers have had little else available. But most Greeks think it looks just as stupid as most non-Greeks would, encountering a page printed that way for the first time.

Even Greeks had to use Porson Greek for a while, though, because Greece spent many years under the control of the Ottoman Empire, which suppressed the use of Greek in printing. At present, there are many Greek typefaces, which generally belong to the following three groups:

1) Take Porson Greek, but italicize the capitals.
2) Take Porson Greek, but optically remove the slope from the small letters.
3) Go sans-serif, possibly including aspects of ancient Greek letterforms.

Greek typography is still, in my opinion, in a parlous state, therefore. The upper-case and lower-case letterforms have been bolted together from unrelated originals, and they haven't really yet been harmonized into a generally agreed-upon way to represent the Greek alphabet in printing with a matching upper- and lower- case. Unless one goes sans-serif, and avoids much of the issue.

But for your purposes, having a look at [Bad link] should suffice.

Type Minds's picture

Thank you very much. I'm just beginning to learn type design, so I'm definitely sticking with sans type for now. Here's the mu from a project I've got going right now:

Does this look okay?

Igor Freiberger's picture

Your design is very regular, without the cursive elements proper to Greek lowercase. It may work in a Latin font where Greek letters just appear as symbols, although I think even there it will seem a bit rigid.

You can try to insert some small adjusts to make it less regular –minor variations in stems, non square shapes, terminals in angles different from 90°. I'm still learning how to handle Greek ductus, but this may serve as an idea:

u/mu and d/delta/eth comparison.

Although I also built µ from u, terminals are slight inclinated and left leg is irregular. I don't think one can say this is a "genuine" µ, but it has enough elements not to be seem as just-an-u-with-leg. To be used besides other Latin letters, this is a possible intermediate solution –not completely Latin, nor truly Greek.

Delta sample is just to show you can easily reuse elements from one character to build other within a same script (d–>eth) but this is much harder with different scripts (Latin/Greek). For delta, use of o, d or g elements would simply cause bad results and I chose to begin a new design to build it.

1985's picture

@ Freidberger, how would you suggest TypeMinds apply this to his typeface? Perhaps some flex in the descender? Some shear at the top of each stroke?

"Unless one goes sans-serif, and avoids much of the issue." As quadibloc says…


@ TypeMinds, I have just started drawing µ, in a sans serif design. I also overlaid a descender and a /u/. I decided to leave the left joint congested though. Not sure this was right, but it seemed more natural. Certainly no expert.

1985's picture

I guess you already answered my question. Removing the serifs and sculpting the stroke endings makes a big difference in the Roman font, less to work with in sans serif perhaps.

TypeMinds, perhaps shift the optical correction from the bowl to the descender? So, taper the descender rather than pinch the bowl on the left.

Igor Freiberger's picture

How would you suggest TypeMinds apply this to his typeface?

This is a possibility:

mu modified besides original.

1985's picture

It's a really beautiful character. The tapered descender looks great. Thanks Freiberger, always generous.

dezcom's picture

What you have shown as a Greek delta looks too symmetrical and much more like the Serbian Cyrillic form of afii10066.
Lowercase Greek is more loose and less constructed looking, at least to my eye.

Igor Freiberger's picture

Thanks, Chris. It's always good to have feedback from native speakers –and you're right about the Serbian/Bulgarian d, I was not aware of this until your observation.

I compiled a set of deltas from various serif fonts. Which ones do you consider adequate? I plan to keep some degree of regularity due to general design I used in my font. But I also want to avoid an appearance like Greek-made-in-Hong-Kong.

Delta from various serif fonts, all in regular weight.

This became a bit off-topic, but I hope the question about delta also help TypeMinds in his project.

cuttlefish's picture

This talk of deltas is not too off-topic. It's always good to get some advice for Latin script users on designing other scripts. I've learned many such things over the years here at Typophile. It would be nice to one day consolidate it all into a tutorial of some sort for designers who have never read Greek before to designing Greek type, and similarly for other scripts.

dezcom's picture

"I compiled a set of deltas from various serif fonts. Which ones do you consider adequate?"

Hard to tell if you have soup from only one ingredient. So much depends on how well it dances with its countrymen ;-) Opa!

Nick Shinn's picture

…the cursive elements proper to Greek lowercase…

Present day Greek typography has left that stricture far behind.
Centro is a type from Panos Vassiliou's foundry, Parachute.
Let's assume he knows what he's doing.

Igor Freiberger's picture


Surely a contextual evaluation would be better, but if the excessive simetry could be perceived in one single glyph, it also may be identified in others under similar condition. Thus your opinion (and also those from other Greek-speaking experts) would be valuable anyway.

If the font just includes Greek characters as Math symbols or to support less knwon alphabets*, delta, mu and others will dialog mostly with Latin-script. So the dance would be with strangers, with a risky mixture between tango and syrtós. ;-)

Most of deltas I verified (about 60) are very simetrical and the bottom half is usually based on o. So I guess the majority of these fonts did not follow traditional Greek style in lowercases. Happily, here in Typophile we can verify these issues to improve Greek characters.

As long as I could conclude, my delta would benefit from a completely different top segment and a right curve with less mirror effect with the left one.


If I understood correctly, there is a vivid discussion about the style to be adopted in Greek lowercase, with designers favouring cursive, traditional, less regular shapes, while others reputes this as a surpassed approach. And even about the cursive ductus there is a reasonable variation about what would be genuine and proper.

As a sample, compare what Chris told about simetry and the very simetrical delta from Centro Serif (it's the first glyph of the second line).

Parachute surely knows what they are doing. But I'm not Greek nor can speak Greek language –and hence I don't have the measure about how much I can let behind in terms of traditional style. In doubt, it seems safer to use shapes more usually recognized as typically Greek while your knowledge does not permit higher flights. My suggestion to TypeMind was based on this idea.

* delta, lambda, theta and omega are used in native North American languages, mixed with a Latin alphabet.

Nick Shinn's picture

…my knowledge does not permit me higher flights.

Then rely on your good taste.
If a type style is non-cursive, it would be gauche to cursify the Greek characters.
When Karpathakis introduced a Greek version of Futura in 1939 (Olympia), he ran with the strict geometry of the original.
And that face was as popular in Greece as elsewhere, so after 70 years may be considered traditional.

Igor Freiberger's picture

Then rely on your good taste.

This is highly personal but, for me, good taste is not enough when I'm designing something I don't know deeply. This is because I came here and did ask other's opinions about things like B with hook, insular G or Shekel.

If a type style is non-cursive, it would be gauche to cursify the Greek characters.

Agreed. This is what I tried with mu and delta I showed above –they follow very much the Latin font style, although top of delta is plainly wrong. The same with the mu suggested to TypeMinds, with subtle modifications.

(...) after 70 years may be considered traditional

Thanks! See, now I have more than good taste to rely on. You helped me giving a concrete info about how long the cursive style is not the only "traditional" Greek design. It's essential to learn with the experts.

Nick Shinn's picture

Here's where I gleaned the story of the Greek Futura, from a real expert! —

dezcom's picture

Fixing the flag at the top would solve most of your problem and move the reading away from Serbian. What I meant about dancing was more about the feeling of rhythmic movement than rigid bilateral symmetry. Sans serif can certainly feel much more geometric. The "countryman" perhaps was a bit misleading. It is not just about Greek glyphs, it is about all the glyphs and how you intend it to be used. If you are only making the Greek that is also math symbols, then your purpose is far different than a complete script meant for reading Greek text. Personally, I wish that the math glyphs were not coded to language text slots so that you could adjust each for their separate uses. Spacing and kerning Greek text is a different beast than just using a few occasional Greek glyphs with Latin and figures. Even x-height, slant and weight could be directed to math formulae instead of text. The way a delta or sigma holds a line in a math expression, where there are mixtures of operators, figures, and cap or lowercase Latin glyphs, would perhaps be a reason not to saddle the math Greek to the text Greek. This is not just about Greek, it is also about the whole nature of the typeface. My personal opinion, which should certainly not be taken as THE Greek way by any means (Greeks rarely agree about much) so think more about your intended purpose with the face than with how "native" it looks. My opinion is that Greek text should dance more than say Times Roman Latin. Many Latin faces have glyphs which appear as a single gestalt on the page and group together as words with a single gestalt. I feel Greek text swings together as a flowing, undulating tango rather than marching lockstep in very even time. Think of it more as 6/8 time rather than 4/4 time. Like a tango, where both dancers must work quite together to avoid stepping on each-others feet, not like a march where one soldiers misstep will only betray his failure to match the rest.
Nick's mention of Futura is a perfect example of not throwing the baby out with the bath water. When someone chooses Futura for setting text, they are not choosing it to give the look of any traditional text. Futura was not meant to evoke Katzanzakis text let alone Homer. The typographer has obviously a different purpose and that purpose needs to be respected. My comment was just about that one glyph you first had shown and it did not look like it was in the modern vein of Futura or AG, or Univers. If someone were to do a Greek translation of a book on Paul Renner or Tschchold's "New Typography", or even some high tech subject matter, you might consider Futura to be a good fit.
That being said, that is not to say that fine Latin faces such as Greta or the more than lovely Maiola are not dancing perfectly in their Greek text. They were not designed by Greeks but they certainly can Tsamiko or Kritiko if asked.
The reason I hesitate from picking a delta from your image is that I don't want you to think there is only one way to dance. Fred Astaire and Ginger Rogers vs Nureyev and Dame Margo Fontaine have their own way of dancing but each is correct in its own right.
What I am saying is, put on the music that you ask them to dance to and then pick the best partners--don't just ask to see one of their feet.

Igor Freiberger's picture

Nick, thanks for the link. Very informative and useful.

Chris, the dancing metaphor is a very nice one and I think the whole idea is clear now. Many thanks for your detailed and highly instructive message.

I wish that the math glyphs were not coded to language text slots so that you could adjust each for their separate uses.

Many Latin-script fonts add pi, mu and omega twice, both as Greek letters and as math symbols, probably due to backward compatibility. It seems it will took some time until authors, editors and typographers fully adopt the correct Unicode characters.

In my case, initial idea was to use just the mathematical symbols. But later I discovered some languages using Greek characters mixed in Latin alphabets, as Wakhi and a few native North American. Some even mix Latin, Cyrillic and Greek glyphs (!), demanding additional glyphs to a complete language support.

When these glyphs are used in a Latin context and in small number, I believe a Latinish design does not cause great troubles. But this may be tricky in a font with Greek support. A Greek lowercase with the dancing dinamism you described would not fit well mixed inside a word with the more static Latin approach. Of course, linguists who built these mixed alphabets hadn't typographical knowledge.

I feel Greek text swings together as a flowing...

I love Greek letters, but I find they very difficult to draw. My hand is still a bit rigid to achieve their swings. Hope in the future I learn some Zorba-tango. :-)

The reason I hesitate from picking a delta...

Understood. Thanks again, the thread become a nice lesson. Hope TypeMinds forgive us to took it so far away from the initial question.

dezcom's picture

Thanks, Igor,
I don't think we have strayed too far from his interests ;-)

Type Minds's picture

Well, it seems that I forgot about this thread after a couple of days, and I missed a lot! I didn't get most of it, but that's okay. Thanks for all your input, everyone.

By the way, I have begun a critique thread for the project where that mu came from [Bad link].

Andreas Stötzner's picture

I think the images given by Igor Freiberger do show the point pretty much good.

I’ve struggled with this kind of question for years.

Catharsis's picture

Nick, do you consistently stress that diagonal in your Greeks...?

I guess I'm sitting right in the tar pit of Latinization with the /mu and /pi in Traction... :o( I currently only have those two (for use in Latin text), so I guess they can get away with not feeling too Greek. But still...

(Note to self: One of these days, extend Traction to Greek and Cyrillic... one of these days...)

dezcom's picture

I think you will have a problem with these not reading as the glyphs intended. You have gone so far towards matching the Latin style that you have lost communication. The pi is totally lost as a Greek letter and as a math symbol. The mu is not much better in that regard. There are no serifs in Greek script and math people don't want to lose their differentiation either.

riccard0's picture

I can’t speak for Greeks, but as a reader and typesetter of texts in which there could occasionally be a |π|, I would say your take has some problems of recognisability.

Catharsis's picture

Thanks guys, that's good to know. I'll keep that /pi for when I do Cyrillic, then. ;o)

Catharsis's picture


eliason's picture

Your thins get too thin, I think? And is the curve of that mu descender unusual?

Catharsis's picture

It's the display cut, so the thins are supposed to be thin. I presume the left leg of the /pi stays thin for too long, though. How about this?

I'm also starting to think that my OS figures should rise above the x-height...

eliason's picture

Those certainly seem better to me.
That thinning horizontal stroke of your pi looks subordinate to the vertical stem on the right, when I think of the reverse being more typical.

Syndicate content Syndicate content