Stem of r to thick?

1996type's picture

Hey everybody,

I'm currently developping a typeface called expletus sans, which is also (an older version) in the critique section. Spme how the stem of the r seems to thick. It's not a rendering issue, because it occurs at all sizes and prints. It really is the right width. I've opended severla typefaces in Typetool and none of them had a thinner stem in the r. Is my eye acting weird, or am I making a mistake here?

oldnick's picture

am I making a mistake here?

Yes: in English, the adverb meaning "excessively" is spelled "too"; otherwise, the stem of the r seems fine.

blank's picture

It looks heavier because the curves going into the top of the stem are slightly heavier than anything similar happening elsewhere in the font. If you slightly lighten the left side of the notch in the top of r it should appear to have the same weight as the top of p. But this is a very minor detail, you’re probably the only person who will see it, and compensating for it might just create a light spot, so be careful if you try to change it. This kind of thing is pops up here in there in most type, and it isn’t necessarily wrong or undesirable, so don’t overthink it.

1996type's picture

I just noticed the same thing in t, a and f, so it probably has to do something with the bend at the end of the straight line.

blank's picture

Because of the intersections strokes/stems in a, e, f, t, f, œ and æ normally need to be a little thinner, and significantly thinner in heavy type. I don’t see it as being much of a problem here because your horizontal crossbars are rather light and don’t connect, which eliminates the extra heaviness that normally occurs. If you open up a well-constructed font and start measuring stems you’ll find this stuff all over the place.

1996type's picture

Thanks guys. Maybe I'm exaggerating a bit here, but I really want this typeface to be perfect in balance, since it's mainly a learning project. Could it be that the note with the (ugly) arrow next to it should be moved up?

1996type's picture

never mind. I've tried it and it doesn't make a difference.
@ Dunwich: Are you suggesting to make the intersections/stems thinner or not? Does 'this stuff' mean 'small differences in stem thickness for optical illusions'?

blank's picture

Are you suggesting to make the intersections/stems thinner or not?

Try it and see. I can’t see your font in print, so I don’t know how it really looks.

Does 'this stuff' mean 'small differences in stem thickness for optical illusions'?

Yes. Type design is the art of tricking eyes and brains into seeing what you want it to see as opposed to what it has evolved to see.

brianskywalker's picture

For me, working on my font project, I ran into these things all the time... I would find a letter that looked too dark or light, even though it was the same as others mathematically, and then go into other fonts and realize they have accounted for what's apparently a common optical problem.

Also, DTF, when you say, "Type design is the art of tricking eyes and brains into seeing what you want it to see as opposed to what it has evolved to see." Wouldn't "Type design is the art of tricking eyes and brains into seeing what's really there as opposed to what it has learned to see." make more sense?

Tomi from Suomi's picture

You should think about those spaces in-between and between those glyphs.

And then you should think about those glyphs: your 'o' is too square when compared to 'c' 'a' and 'g', which has way too square lower part. And have you compared your 'a' and 'e'? Hmmm… And your 'z' is way too narrow.

blank's picture

ouldn't "Type design is the art of tricking eyes and brains into seeing what's really there as opposed to what it has learned to see." make more sense?

No. If people have to see what’s really there they have to stop reading and start thinking about the letters. What you want them to do is not see all of the weird stuff you have to do to make a font work and just read.

Rob O. Font's picture

Are you sure the contour direction of the r is correct?

1996type's picture

@brian: thanks. I'll have acloser look at stem thickness in the future. btw: I'm using myriad pro to compare it to. Is that a good choice?

@Tomi:
- "You should think about those spaces in-between and between those glyphs." Your talking about spacing and the whitespace inside a glyph, right? If so, I haven't paid much attention to spacing yet.
- "your 'o' is too square when compared to 'c' 'a' and 'g', which has way too square lower part." The left side of the c is identical to the left side of the o, but I'll make both a bit rounder. The lower part of the g doesn't look squarish to me at all.
- "And have you compared your 'a' and 'e'? Hmmm… And your 'z' is way too narrow." ofcoarse I've compared the a and e, but obviously not in the right way. What am I missing? My z is indeed quite narrow, but it doesn't look to dark to me and I like the way it looks.

@dberlow: yes I am.

1996type's picture

thanks for your comments so far everybody. It's already been of great help.

brianskywalker's picture

Myriad is a good font to compare with, but I wouldn't limit myself to a single font to learn from. Also try Franklin Gothic, News Gothic, or Calibre - or all of them.

1996type's picture

Here's an update. I've used centro sans as a comparison. The stems of r, f and t are 5 (pnts?) thinner. Also many other glyps have been changed. I hope this is an improvement.

brianskywalker's picture

I think that's an improvement. Some quick things: the j dot needs to be larger than the i dot. Right now it looks too small. The z is also a bit pointy, and you might consider joining the diagonal strokes of the k to be more like <. You might try moving the break in the g to the right a little. The d, b, and p look a little too open. Looking good otherwise. Make the uppercase!

nina's picture

"the j dot needs to be larger than the i dot"
? Why should that be?

That "r" looks potentially too thin to me now, but it's really hard to say by staring at a bitmap image.

1996type's picture

@brian: I think the j dot is fine. I've never seen a typeface with a larger dot on the j. It's probabaly just a rendering issue. The z is a bit pointy indeed, but I like it. I might just make it curve a little bit at the ends. I can't make the k more like < because than the gap looks ugly, and I don't think it looks ugly now. "You might try moving the break in the g to the right a little." What do you mean by 'the break'? "The d, b, and p look a little too open." I don't really get this. You think the gap should be smaller or something else? "Looking good otherwise. Make the uppercase!" Thanks! I have already made uppercase, as you can see in the critique section, but I'm completely redrawing them. I'll make an upload including uppercase and numerals in a few days, I hope.

Thanks alot for your help!

@nina: "That "r" looks potentially too thin to me now, but it's really hard to say by staring at a bitmap image." True, I'll add a pdf for the next update and have a look at some more prints myself. Thanks for your help.

brianskywalker's picture

Optically, the dot of the j usually needs to be slightly larger to account for the descender. When looking at a j with an i directly next to it, where the dots are the same size, the j appears to have a smaller dot, even though they are exactly the same. It's mentioned in Karen Cheng's Designing Type, I believe.

Unfortunately, finding fonts that I have which do this seems to have been a bit difficult. Some fonts that do: IM Fell French Canon, High Tower Text, Monotype Garamond, Adobe Caslon Pro. Surprisingly, some fonts actually do the opposite, this is especially true of italics. Example: Monotype Garamond Italic.

In my example below, I also have included fonts that I had thought would have accounted for this, but didn't: Century Old Style, Minion Pro, Myriad Pro, FF Scala, Hoefler Text, Futura.

This is a very interesting study, I think. Although you may not need to change the size....

Regarding the d, b, and p being too open, I meant that, in comparison to the o, they look a bit too wide, or the counters look too large.

nina's picture

Interesting. I noticed it in Caslon too. OTOH in this case, we're looking at a rather display-ish sans – where things would be more regularized generally than in a classic text face. I haven't yet found any examples in this genre, although I admit I haven't been looking very hard.
BTW, looking at the horizontal position of the tittle in the "i" would make me reconsider using Adobe Caslon as a tittle reference. :-)

brianskywalker's picture

Actually, the only sans I could find that compensated for that optical illusion was Arial - the dot of the j was slightly taller than the dot of the i. But it was almost unnoticeable.

The positioning on Adobe Caslon certainly is interesting, it seems to center on the serif instead of the stem.

1996type's picture

@brian: Wow! you really took your time for this. I'll make a test print to see if the tittle on the j needs to be a bit larger. It does look to small to me right now, but it might just be a rendering issue.

@nina: Caslon is a very old, but still succesfull typeface, so it must be good. ADOBE Caslon is often said to miss the true characteristics of Caslon, though I don't see it. Caslon in general -I think there are over ten different productions of Caslon.- looks out of balance to me. I know a true textface often looks weird at display sizes, but Caslon still doesn't look good to me at text sizes. For example: I think the top serif in the i looks too much different from the bottom serifs, which are (almost) slab serifs.

1996type's picture

Many thanks for a very fun lesson so far!

Tomi from Suomi's picture

For the logic of the design, shouldn't your 'r' be like this?

1996type's picture

Your a far better and more experienced designer than I am, but I can't change something only because you say it's better without actually agreeing with it myself. There are many typefaces where the r is simply a cut off n, but this doesn't look right to me. This may not sound very logical, but I think the r should be more like a cut off p without descender. To be honest I haven't tried your r in a word, so I'll give it a go, but I think I'll hold on to my own. Furthermore, I've read and fully agree with this: http://www.typeworkshop.com/index.php?id1=type-basics&id2=&id3=&id4=&id5...

The way they explain it, clearly wont work for my n. My conclusion is that the r should be seen as a glyph to be designed the way it looks best, disregarding the 'logic' of the design. If you still want to convince me otherwise, please do so. I'm always willing to learn. I'll upload a comparison of the two r's in a word today.

1996type's picture

Here's a comparison. 'Your' r does give it a slightly more 'display' feeling, but I think I'll stick with the original and make 'your' r an alternate option. Don't mind the spacing.

Rob O. Font's picture

b.skywalker> When looking at a j with an i directly next to it, where the dots are the same size, the j appears to have a smaller dot, even though they are exactly the same.

I see no such effect as you state.

I believe you show serif designs with irregularities of dot placement in the service of "oldness", not serving optical size correction.

The fact that it's hard to find in sans is because sans is usually designed for "newness", and since there is no optical correction to do, the dots end up the same.

1996type's picture

@dberlow: I do see the effect brian states, and I'll correct it. It might be true to a certain extend that some of the details in serif typefaces are due to irregularities in the service of 'oldness', but I clearly see in my typeface that the dot of the j optically looks smaller.

Bert Vanderveen's picture

Another issue that clouds the view is the fact that in the abc-representation your r is next to s, which has a somewhat excessive overshoot (the amount of intrusion below the baseline, necessary to balance the curved parts) which could be a pixelation thing. To me it now looks like the stem of the r is too short, which is not true of course.
Maybe you should use a row like nrirmrhr etc. first to settle the stem width problem and then look into the relation of the r with curved glyphs is a row like rorsurer etc.

Just my 2 cts. (I am very impressed with what you have done)

brianskywalker's picture

1996type, how are you following the example on typeworkshop.com by making a cut off p?? The version of the r Tomi from Soumi presented follows that much better. Typeworkshop didn't recommend not making your r from an n, they recommended rotating the whitespace above the join of the n, and following that for the whitespace above the join of the r. The idea, of course, wasn't to follow this extremely strictly, but was to demonstrate a way to make a more balanced r.

Like this:

- Start with an an
- Chop it off
- Fix the end of the r
- rotate the shape of the whitespace
And you now have a balanced lowercase r!

The final example show that isn't the only way to make a fairly balanced r (this form is used in many classical serifs, and in some sans serifs), although the fourth example is slightly more balanced in this case. What you use, in my opinion, should depend on the design of the typeface. While you do sometimes need to sacrifice the logic of a design to make a more readable letter, sometimes you also need to sacrifice certain things about a letter to make a cohesive design. In this case, the letter r will not lose any readability, and in your design, this new r is more balanced than your previous design.

1996type's picture

@bert: The overshoot in the s is indeed to much. I'll have another look at it.
@brian: "how are you following the example on typeworkshop.com by making a cut off p??" Ok I have to admit that usually the r is more like a cut off n than p, but the 'normal' r looks much more like a cut off p from expletus than an n from expletus. "and in your design, this new r is more balanced than your previous design." I doubt that, and it's certainly much less legible. I don't believe (disregarding the 'logic' of the design) that the original r looks less balanced than the one tomi suggested and it's more legible. You may think otherwise and your welcome to try to concinvince me, but don't try to convince me with things about what is 'logical' or 'usual'. I think the main point of typeworkshop.com is that the arm of the r should meet the stem at a lower spot than the n which is what I agree with. Tomi's suggestion doesn't do this, and the original does.

I do realise I was very unclear in my previous comment, but my opinion remains unchanged.

1996type's picture

Here's the update! Uppercase, lowercase, numerals, and more. To judge it properly I suggest you download and print the pdf from here: http://www.typophile.com/node/72368

Hope you like it!

brianskywalker's picture

It's looking very good.

I should note that the reason that's recommended on typeworkshop is to make the letter more even, and help account for the large gap to the right. It's not to add legibility. Most fonts don't join much lower (slightly lower is usually necessary optically, and much lower helps the whitespace) and are very legible. But joining lower definitely helps with whitespace, and it was my decision for Neuton to do it the way recommended on TWS. But in some cases a high join is much better, it all depends on the font.

Scala, Fago, IM Fell, and some of Fred Smeijers' fonts, have an r which actually joins higher than the n (unfortunately I don't have any of Smeijers' fonts at the moment). Most fonts actually don't change the join at all, and many join lower, some very exaggerated like Chaparral.

PabloImpallari's picture

Nice way of comparing fonts Brian!

1996type's picture

Thanks Brian. I have made the (joint?) a little thinner where it meeets the stem, but kept the shape the same. I'll upload it soon. Neuton looks good, even though it's just two glyphs.

brianskywalker's picture

Thanks.

For some links on Neuton:
http://typophile.com/node/68311 Original thread
http://typophile.com/node/74121 The italics
http://code.google.com/webfonts/family?family=Neuton#specimen more recent version of the regular

Looking forward to your improved r btw.

William Berkson's picture

Brian, I think the reason for the larger dot on the j in Adobe Caslon is that the head serif on the "j" is larger than the head serif on the "i". The larger dot is to match the larger head serif.

The larger head serif I think is, in turn, to deal with the different shape of the "j"—the white space opened up by absence of foot serifs, and the bulb to the left. This is partly an "old style" thing, I think--and can be a good one.

In my Williams Caslon Text, I made the j dot also larger, for the same reasons, but only slightly, as my j has only a slightly larger head serif, though quite a different shape. Also my feeling was that when the difference is too marked it looks a little irregular and antique, which I didn't want.

In a sans, there is no reason for a larger dot, as there is no head serif.

brianskywalker's picture

Interesting, that does make sense, to an extent. Although, I distinctly remember something about this in Cheng's Designing Type? Anyone have that book? (I got it out on interlibrary loan a few months ago)

nina's picture

I have the book. I'm not seeing this under either serif or sans serif "i"/"j",
but I may have overlooked something.
Anyway, just because it's in a book doesn't make it true :-)
And Cheng's book isn't free from criticism anyway from what I gather.

William: Thanks for that explanation!

brianskywalker's picture

> just because it's in a book doesn't make it true :-)

You got a point there! :)

1996type's picture

This all sounds very logical, but the dot on the j in expletus does look smaller to my eye than the dot on the i, so I've corrected it.

Nick Shinn's picture


Gill and Century have always struck me as having a most pronounced difference between the structure of r and n.

brianskywalker's picture

@1996type, I see the j dot being smaller also, but from what I can tell, it's not usually necessary to make an adjustment; although in certain weights or font style, I think it my become unnecessary entirely - for instance in an ultra bold.

Syndicate content Syndicate content