Scaling Up Lowercase

beejay's picture

A couple threads have mentioned scaling
the lowercase to 106 or 110 percent
(in relation to the uppercase).

Where can I find out more information about this practice, the pros and cons, etc.



eomine's picture

Hmm... I don't see the point of doing that (if I understood your post correctly).

Why would you want to scale up the lowercase (keeping the uppercase unscaled)?

beejay's picture

Eduardo, sorry, I was kinda inarticulate...should have said
scaling up the uppercase weight.

From Randy's Friday Fonts....

"The big question I'm facing before moving on is: are the caps too dark for the lower case? Right now the caps are 112% heavier than the lower case. Is there a rule of thumb on this? "

And then somewhere else, don't know where, I read
something about scaling.

eomine's picture

Oh, now I get it.
I think it's very simple. Uppercase need to be slightly heavier than lowercase. Otherwise, since Uc letters have more white space around them, they (Uc) would look lighter than lc.
But it should be just a slight extra weight. No more than 110%, I think. Of course, this percentage depends on the typeface too, some typeface may need more Uc extra weight.
A priori, there's no negative consequences. It's just a necessary optical adjustment.

.00's picture


beejay's picture

Thanks Eduardo and James.

Do you know any specific sources where I can
learn more about this?

I am still searching Google...but Carter's
stuff at doesn't seem to mention it.

However, the UC stem width in Georgia is 114%
of the LC stem width.(UC I, lc i)

In ITC New Baskerville, it is approx. 120%.

A third face, from a respectable foundry,
shows the UC and lowercase exactly the same.

plain old Helvetica -- the difference is roughly 111%.


hrant's picture

> A third face, from a respectable foundry,
> shows the UC and lowercase exactly the same.

What face?


beejay's picture

Further...after a hiatus, i'm back into
type-design-obsession mode. After lots of
methodical Monoline stuff, i jumped into
thicks/thins but finally
with a shape-based
method sted of a mimic method.

The key was getting the bowls on the 8 and the
0 established, and then taking the thick/thins
from there... but I've been doing it
minus the overshoot.

Kinda like the Benguiat thing.

I call it Nudge 80...a safe way off the
Grid Bros. Monoline Express!


beejay's picture

>> What face?


hrant's picture

"Respectable"? Awww, shucks.

Look how small my cap height is compared to the lc x-height. The question is, do the caps look too light in actual text? I actually remember somebody complaining about that to me, but being one of the very few people who's seen Patria at 2400 dpi, I'm not sure they need to be heavier... And note that my caps are very tightly spaced. Looser spacing requires darker forms.

So they might indeed be too light, but maybe the more important thing I'd note is that your ideal weight increase depends on relative proportions, and spacing.


eomine's picture

I don't think you'll find something about it on web. I think I read something on this subject right here, in these forums, but I'm not sure...

I think the variation in your numbers is mostly due to each typeface needing a particular adjustment. The important is that, in the end, you should achieve a good (even) text color.

eomine's picture

Hrant has good points, but, wasn't BJ just kidding? :-)

beejay's picture

I was kidding about Patria.

But this subject deserves some
examination and discussion, surely.

I tried some other fonts and it seems
clear that some type designers make
the optical adjustment and some don't.

I'm going to do some further research.


hrant's picture

> wasn't BJ just kidding?

Maybe about the "respectable" bit, but my cap weight increase is indeed only about 103%.

BJ, if there's another font like that, please tell me what it is!


beejay's picture

A font near 103%?...okay.

I have the Carter/Bitstream stuff
that came with Clip Art Implosion.
+ all the ITC stuff that came
with Illustrator as a good base of
respectable stuff from respectable

I'll leave the Display out for now.

I will measure the relative stem widths of 20-25 faces
and post the this time next week.

It'd be cool if someone else could do the same
with some of their fonts.


aquatoad's picture

Hi BJ,

Thanks for starting up this thread. I haven't come to a difinative answer on this yet. I scaled back the caps in my Friday Fonts thread. They went from 112% to 106%. What i wanted was for them to optically look the same. As I got darker in weight, the difference went up slightly. I think the Black weight is 109%. The light weight I recently posted will be around 105%.

Here's what I did with my lunchhour:
Adobe Garamond
Bauer Bodoni
ITC Officina Serif
Neue Helvetica

(Rounded to the nearest integer)

cap weight comparison

Perhaps we will see a trend that serifs have a larger difference? Bembo need to go on a diet! Actually heavy caps has always been a beef for me about Bembo.

As far as the Ed Benguiat LC o = UC stem, here are the grades for the roman weights:

AGaramond: cap too narrow
Bembo: cap too wide
BBadoni: cap too wide
Officina Serif: exact

NHelv: cap slightly wide
Futura: cap slightly wide
Univers: exact

So much for my theory about bolder weights having more % difference between cap and lc!


beejay's picture

Randy - very interesting!...Maybe an authority will come in
and explain whether there is a method to the madness or
whether it's all in the optical correction. (eyeballing)

Hrant...103% for Patria. When did you scale the Uppercase...
Was it after looking at high-res proofs...or was it on screen.
Or was it in the sketch stage?

Did you "nudge" the bezier points .03 and decide that that
looked right...What I mean is, was it trial and error or what...

Not to barrage you with questions, but you are The Answer Man.

Also, upthread, I mentioned Carter/Bitstream stuff...
Carter said (in Porky's Designing Typefaces book)
that during his time at Bistream, the only typeface he
drew was Charter. My mistake.


hrant's picture

My "master" weight (from which the light and dark extremes were made) had lc stems of 90 ems. I made the caps 95, but when I made linos realized it was a bit much. I try to avoid 1-em mods, so I went down 2 ems, and it looked fine. 93/90 = 1.0333...


aquatoad's picture


Have you made any headway on your 20-25 fonts? I'd like to see a broader sampling, and more contemporary designs (last 10years) to see how they stack up.

I've decided my personal preference is between 103 - 110%. Anything more seems dark to me. Although I must admit my test is flawed. I'm comparing horizontal width of the stem of the h with the stem of the H. This doesn't neccesarily fully take into account the font's modulation. For example, Adobe Garamond. The roman weight is 111%, but text settings don't look dark. I think this is because on the "width of lc 'o' = UC stem" test the caps looked narrow. This tells me the over all color is even, dispite the "heavy" %

Hope that makes some sense?


beejay's picture


I was thinking that yours was a good start ;)

I was hoping that Kent or John Hudson might
shine some light this way...

I did check out Lux Sans and I think that one
was 106%.

maybe more tonight.


kentlew's picture

bj --

My ears were burning. I think I have a file somewhere
where I was compiled various measurements, proportions,
and attributes from a variety of typefaces. I don't recall
if I calculated the cap stem as a proportion of the lc stem
in the file. I've certainly looked at it on an ad hoc basis
for my own edification. I'll see if I can dig that up later
tonight and get some data for you.

-- Kent.

beejay's picture

Here are a couple of quick studies...

Ellington and Georgia.

You can zoom in with control-click.

caps.swf (4.8 k)

the thing I think we are looking for is the *how*.

The method or process used by various typographers
seems different, and in some cases, the scaling does
not exist.

I'm also wondering whether this is something typically
taught in typography courses...


beejay's picture

And here is the basic method that I started using a month
ago for non-monoline faces. I omitted a tiny bit of the
overshoots and took
my thicks/thins from there.

nudge.swf (1.5 k)

Miss Tiffany's picture

*thinking out loud*

Couldn't it be that instead of an exact percentage, it is somewhat similar to "what came first the chicken or the egg". For instance, all these designers don't have a specific percentage, they design to taste or optical taste anyway. It may be that they have a general percentage that has happened, and maybe they start there, but how could you have an exact percentage?

I'm not disagreeing, these thoughts just occurred to me. Or am I completely wrong?

beejay's picture

I agree Tiffany. Definitely. There's no *ideal* percentage.

I'm just wondering *how* they arrived at the
solutions...the percentage itself is relatively irrelevant*
but it's the process.
At some point, the professional typographers
learn the process. Or they have a
different method, like the Benguiat method
that James uses.

When all the type tutorials that I've ever seen
neglect to mention this type design basic

beejay's picture

revised the above post...didn't make sense

kentlew's picture

bj --

Here is the data calculated from that file. I thought
that I had measured more faces, but these were the
only entries with the stem data. The first figure is the
Hstem/nstem and the second is Othick/othick.

Janson Text1.2221.253
Miller Text Roman1.1651.198
Minion Regular1.0771.066
Requiem Display1.1821.125
Requiem Fine1.2031.138
Requiem Text1.1731.098
Whitman Roman1.0831.088

hawk's picture


bj, what type tutorials do you have? work with?

David hamuel

beejay's picture

The tutorials are all pretty basic.
Chank. Larabie. etc.

Those are a starting point for most newbies.

Gerald Lange referred to one in another
thread a few days ago. (briem or something?)

That one seemed pretty good, but no
mention of scaling.

Early on, I developed a bunch of methods
that used the particular strenghts of
Illustrator. The methods are named
on the splash page here:

Jared B. is putting
together a tutorial, but I don't know the
status of that.

New typographers enter our fray yearly, so
hopefully, we can put together something that
is a little bit more comprehensive than what
is out there.


thanks for the info Kent.

kentlew's picture

bj --

As you've said, there isn't a fixed percentage. It's an
*optical* compensation, so you have to *look* and see
what it needs. Some designs might need more, some
might not. It may also depend a little upon cultural trends,
and also perhaps printing technology (see below).

If you had asked me I probably would have told you
that I made the Whitman capital stem 110% of the
lowercase. But I see now that I didn't. Go figure.

In writing about the cutting of Linotype Monticello
(a revival of Binny & Ronaldson's Pica Roman No. 1,
aka Oxford) C.H. Griffith wrote that he was surprised
at how heavy the capitals appeared in contemporary
proofs made with type freshly cast from the original
matrices, compared to specimens from the original era.
The theory he developed was that 19th century printing
technique tended to build up more color on the lowercase
than the capitals and the typefounder then added more
weight to the capitals to compensate and get them to
print more in harmony with the lowercase. But 20th
century technique did not have the same effect and so
the extra weight in the capitals was more pronounced
in his contemporary proofs. The result is that he slimmed
down the weight in the capitals in order to get an
equivalent harmony in Linotype Monticello.

I seem to remember Matthew saying that when he did
his recent digital revival of Monticello, he chose to slim
the weight of the capitals even further, relative to the
lowercase, because the capitals of 19th-century style
types (even as adjusted by Griffith in the 50s) still
tended to look too heavy to the modern eye.

I say, Trust your eyes. And if you don't trust your eyes
yet, then train them by looking closely at other typefaces
and developing your own judgement about which look
balanced in the capitals and which do not.

-- K.

beejay's picture

btw, we are on semi-hiatus at Apollo 26...
the site hasn't been updated since December 2002.

thus, in the reverb section,
"January of this year" refers to 2002.

beejay's picture

Sorry Kent, we cross posted.

Indeed, thanks for the input.

This is the kind of information that
is great to build from...thanks.

hrant's picture

> C.H. Griffith wrote that he was surprised at how heavy the capitals appeared in contemporary proofs made with type freshly cast from the original matrices, compared to specimens from the original era

Is it possible this has to do with gain (which was greater in the past), since a certain [absolute] amount of gain applied to both lc and UC bodies would actually bring them closer in relative weight? (This would also affect stroke contrast, in fact to an even greater extent.)


BTW, it also happens that caps are sometimes made intentionally much darker than the lc. Usually this happens in very small sizes and/or darker weights, where all-caps setting might benefit from extra punch. I remember David Berlow saying something to this effect on the ATypI list a while back.


jfp's picture

These kind of analyse remember me the Commission Jaugeon who done the most horrible typeface called Romain du Roi, particularly the italic -- corrected visually by the punchcutter, who never tell what he have done to them! Others done similar with geometric analyse of Roman inscriptions with compas and rules, funny!

There is no strict rules for such things, my Parisine, a bit like Gill Sans seems to have quite heavy caps, just because I want them like that (for some reasons that I know well), others, such Stone or Minion show pretty light caps, "et alors?" Its matter of taste, "culture" and so on.

If you start a new typeface from a compilation of everything found on various typefaces, you will get probably the most boring typeface ever. no?

Type design is more feeling than maths.

kentlew's picture

Hrant, I should have known you'd pick up on the Griffith story. Although I think I know what you mean in general about absolute gain, I'm not sure I really followed what you were specifically stating above. But no matter. This veers from the topic of this thread, but what the hell -- here's the whole quote:

As the

hrant's picture

> Type design is more feeling than maths.

Yes, but it's also a set of "procedures" (different for each of us), and I think BJ is after those, not a formula. The numbers are only clues as to what the procedures are, but often they're the only thing we have.

> I'm not sure I really followed what you were specifically stating above.

What I was saying is that if you add (the equivalent of) X em units of weight (due to gain) to the lc and the (thicker) UC, the two become relatively closer in weight.

>> "a tendency to build up color on the lower-case, whereas the capitals with generous counters and side-bearing were not similarly affected"

Does this make sense to you?


beejay's picture

Yeah, what Hrant said.

Not math. Looking at these faces is for learning.
Randy and I just had our interest aroused
a little bit, that's all.

Intriguing to me: Process, procedure, method, craft, guidelines,
tricks, tips, tools, etc. etc.

The toolbox starts wide open and empty
and stays open, with perpetual vacancy,
till we die, if we are lucky.


beejay's picture

David - the site mentioned above is:

Gunnlaugur SE Briem of Iceland.

hawk's picture


thank you. ya. i know this guy.

do you have "type by design" - Stephen Moye? do you know Mortimer Leach?

your question is more then just a question. so... how to justly proportion a typefce? are these proportions subjective? fixed? is there a quantitative method?

David Hamuel

beejay's picture

David - I'm not sure, seriously, if these are rhetorical
questions or not. If you studied with Neville Brody,
I'm assuming your questions are indeed rhetorical
and that you are toying with me.


It's all good. I can handle it.

Feel free to email me. We're both in
the LA area. I'd love to show you some stuff
I've been working on and vice versa.


hawk's picture


hey, i'm not that "old" - Mr. Brody is....something....let's see...around 50. i'm not.'s not important.

no. i don't think these are rhetorical questions.

Mortimer Leach - he, for example, drew curves - 1/3 wider than the straight stems. you can study his artwork. and compare it.

you can agree with his "method"/method - or not. and always ask - why 1/3 wider?

is that fixed? quantitative? subjective?

a good study is Bodoni (1926) and Sheldon (1947) - same cap height, not the same x-height - Sheldon is more-more generous with x-height. and , of course, not the same ascenders & descenders.

so --- is that fixed? quantitative?

that's my opinion. maybe i'm wrong. or maybe there's no maybe. maybe.


what are you doing? what kind of artwork?

David Hamuel

beejay's picture

david - sorry, I misinterpreted the tone of your previous post. Aaaargh!

No, you're right, they are not rhetorical questions.

I don't think there's a *definitive* answer.
Just like the initial question of the thread...there is no
single answer to the scaling question.

I guess Kent said it best, Trust your eyes.

As far as proportions...the age-old proportions
work pretty well.

If you have a link for Mortimer Leach, that'd be great.
I'll check out his work.

as far as stuff, I was talking about type.

need to go to bed.



eomine's picture

>Mortimer Leach - he, for example, drew curves -
>1/3 wider than the straight stems. [...]
>you can agree with his "method"/method - or not.
>and always ask - why 1/3 wider?
>is that fixed? quantitative? subjective?

See, in this case (if I understand your questions
correctly), the more we try to achieve "fixed" values
(proportions, weights, etc), the more we get arbitrary
("subjective") ones.

I don't know much about Leach's work, so I may be
wrong, but his method sounds highly arbitrary to me,
because it doesn't seem to care about other factors
that might affect optical compensation. For example,
heavy typefaces would need larger (round-shapes)
compensation than thin ones.

>a good study is Bodoni and Sheldon [...]
>so --- is that fixed? quantitative?

I think you kinda answered your own question. ;-)
It's not fixed, each typeface has its very own
proportions (horizontal, vertical, weights, contrasts,
etc). And all those little details make them what they

>[...] I don't think these are rhetorical questions.

They're kinda rhetorical, I think. Like Bj said, there's no
definitive answer. To me, in this case, the questioning
*is* the answer itself. ;-)

hrant's picture

> 1/3

To me this is a direct effect of misguided Modernism,
where small round numbers are given religious status.


hawk's picture

Hrant, why ......misguided Modernism"?

Eduardo, "'s not fixed, each typeface has its own proportions...." - ah ha. ok. in that case - you don't need to build a system/"system" about "scaling up lowercase".

BJ, i was very very very busy. still. however - i was trying to see your artwork - your website - but i saw only black screen.

David Hamuel

hrant's picture

> why ......misguided Modernism"?

The belief that small integer numbers make for better fonts is a typical Modernist fallacy. The entire movement stems from a naive desire for and attempt at control.


Syndicate content Syndicate content