Sentence Capitals vs ALL CAPS

Michael Hernan's picture

Nina asked: Hmm… so what are 'Sentence Capitals'? :) Here

Perhaps I can elucidate...

Sentence Capitals:

BTW Forgive me if I am going over old ground - but would be interested if anyone else has investigated this (intentionally).

When looking at set type, I quite often find, that the capital letter that starts the sentence in relation to the rest of a word is bold in appearance. Because I am not convinced this is necessary for set text I am investigating the idea of specially optically-balancing the capital letters to the weight of the following lowercase glyphs.

There is a precedent in having bolder capitals. This is clearly seen in Scotch Roman. Also - there might be (a remote chance) that there is some element of legacy suggesting this 'style' where capitals from an existing face were used with a slightly lighter lowercase set?

Optical Consideration:
I am aware that due to the nature of a capital letter having both a less complicated structure (they are more open due to their relative size to l/c) and elongated stroke-lengths naturally lightens its form which needs to be compensated. I would still expect Capital strokes to be thicker than lowercase however to a lesser degree than what is currently considered 'normal'

Capitals tend to be designed with other capitals so the design ecosystem is not the same as real world use. i.e. I suggest that more consideration is given to how the capital 'exists' with the lowercase rather than its CAPITAL glyph relatives.

However - What seems to happen is that these lighter capitals that have been designed to work as an initial capital in a word don't look so good when set ALL CAPS.

Natural compromise:
So what we see in fonts for the most part (but not Scotch Roman) is that a compromise has been allowed for so that the caps looks great when set ALL caps and OK when set with lowercase.

My thinking is that this balance can be readdressed so that the Capital letter designed to compliment with lowercase gets the priority styling consideration and that if necessary a secondary set of capitals for ALL CAPS can be supplied.

I hope soon to supply some 'proof of concept' but would welcome points of view and other examples by those who care to give this some consideration...


Nick Shinn's picture

All caps: that would be in the OpenType "titling" feature. It's the original definition of the feature, and Adobe Garamond was the first digital face to implement it, c.1990. I don't know of any other faces that have separate titling capitals, offhand.

BTW, Goudy's Kennerley (1911) was designed because he found the text colour of Caslon (IIRC) too spotty, due to its prominent capitals, or so he says in his book.

I don't think it would be a good idea for the letter glyphs to change from normal caps when "all caps" is applied, that's not what people expect to happen.
I mainly use that feature for changing to a set of figures that harmonizes with capitals, and raised punctuation.

Thomas Phinney's picture

Actually, all caps would be the OpenType "case" feature. InDesign and others support it, and invoke it automatically when text is formatted as all caps. But I agree with Nick that a huge change would be disturbing.

Titling may or may not be all caps, but is always intended for larger sizes. It's more of an optical size feature than an uppercase feature.



Nick Shinn's picture

Thanks Thomas, I didn't express myself clearly.

To clarify:
The "all caps" setting that Michael is describing is best implemented by the "titl" feature.

People expect that the "All Caps" menu command (which applies the "case" feature) will not produce different glyphs than if they had just typed with the caps key held down.

For OpenType features, refer to:

Michael Hernan's picture

Above: Uniform – Medium with unadjusted Capital U
Below: Uniform – Medium with adjusted Capital U to compliment the lowercase.

Opentype Implementation

Yes, I think what I am suggesting would be implemented by 'case' feature.

From the user perspective this is best done with the 'case' feature accessed [InDesign] by 'Character' palette chosing 'All Caps' from the drop-down menu.

The result would change all the capitals to a slightly darker design where all the capitals work well together OR if the weight of the caps is fine just swapping out Capitals that were modified to get the harmonious result with the lowercase.

In more detail the mechanics of the relationship of Capitals to lowercase and why I think Capitals become darker in the design process:

Above: Uniform – Light with unadjusted Capital I and O.
Below: Uniform – Light with adjusted Capital I and O to compliment the lowercase.

From my study and experimentation - I see that the 'capital O' has a different relationship to a string of lowercase letters than a 'Capital I'. If after adjusting (independently) the the weights of the O and I in relation to a lowercase string of letters I then comparing them side by side again - it is seen that the I is lighter than the O. (See below)

[Note for illustration below: I actually see that the I is lighter (more than I would expect) than the O in the original so I will have to repeat this with another font design to confirm my idea.]

Above: Uniform – Light with unadjusted Capital U, I and O.
Below: Uniform – Light with adjusted Capital U, I and O to compliment the lowercase. Brought back together I expect to see irregularities in weighting across the Capital set of Glyphs.

As a designer - this doesn't seem logical and so the natural tendency is to darken up the I and then even perhaps further darken all the capitals so that they all look darker (and not just the straight ones). This is certainly a trap if your original intention is to have an even look to your set text.

This therefore suggests that a set of capitals exist within the font file that are presented when the user selects 'All Caps'. The point being that if the user sets a title using the Shift key - the Title wont be either be optimised typographically (as is the normal use for 'All Caps'/'case' feature - as Nick mentioned) or aesthetically correct (my theory), as the letters presented by a Shift key action are in fact an extension of the lowercase and not true capitals... but Sentence Capitals.

Nick Shinn's picture

This won't work, because people will use the keyboard shortcut/All Caps command to capitalize a phrase in running U&lc text.
That's what I do, and I assume that many others do as well.
With that scenario, your "Sentence Capitals" would really stand out, against the normal capitals elsewhere.
You can't expect people to use small caps, and some will want to use All Caps for a couple of words after a drop cap, &c.&c.

Have you researched user behavior?

Michael Hernan's picture

We can assume user behaviour (at this point) to be to get approximate results using the minimum of effort.

We can also assume that OpenType features are not used very much - but the features are there for those who want to take advantage of them.

We would all be interested in seeing figures or a breakdown in OpenType usage in sectors of publishing output. This is not something I personally see any benefit in doing and would be happy to work on my two assumptions above.

Regarding an obscure feature not instantly understood even by OpenType guidelines - there maybe some users who would perhaps want to achieve such a typographic effect, but this as you allude to would at the cost of knowing to do something beyond what is usual practice and actually understanding the benefit of doing it.

I am fully aware there might only be a user group of one that might demand such sensitivity, but that is all I would need to continue considering this. Though I am concerning myself with what is a very small avenue of innovation - it is an innovation nonetheless, and I see this along with other small innovations or ideas as the reason for my working on any font. I find it unnatural to accept the status quo over working though a concept or innovation.

Perhaps after working out though investigation as to if there *is* any benefit of Sentance Capitals (as a concept) would I then work on the practicality (from a user perspective) if it generally might be accepted or used. Perhaps even to devising a strategy by which the user is rewarded by their extra effort?

I really don't want to get bogged down in OpenType functionality or user scenarios though discussion so far has been enlightening and shown me what I was originally considering (Titling Alternatives) to be incorrect. [Thanks]

In this case, as you can see, the font itself does not lend itself to drop caps or All Caps anyhow. I see the default being Caps that work with the lowercase. Another design/model - Different story.

Nick Shinn's picture

"All caps" -- and keyboard shortcuts such as "command shift K" -- has been on the menu of a great many layout apps these past 20 years, since well before "case" was piggy-backed onto it. So it's not just an obscure OpenType deelybob.

oldnick's picture

I find it unnatural to accept the status quo over working though a concept or innovation.

I suppose you could call that the Garden Party approach to design: you can't please everybody, so you got to please yourself...

Michael Hernan's picture

Continuing down the Rabbit Hole...

Interestingly Williamson in 'Methods of book design' alludes to the light weight nature of the capitals of Juliana (1958). Knowing that Juliana is firstly a book face I think that Sem Hartz had the sensitivity to look at the capitals as an extension of the lowercase and not a bolt-on necessity which so easily is the case when getting a typeface to market.

Also, the capitals in Juliana adopt appropriate width. Each letter seems to have a sense of them-selves. This is clearly seen with the capital letters E, F L and P.

Linotype Juliana, 1958 by Sem Hartz

oldnick's picture

I think that Sem Hartz had the sensitivity to look at the capitals as an extension of the lowercase and not a bolt-on necessity which so easily is the case when getting a typeface to market

This is clearly a case of putting the Descartes before the horse ("I think, therefore I tend to over-think."). With the Roman alphabet, the uppercase letters came first, and the lowercase letters became a bolt-on convenience. You have developed an elegant theory which solves a problem which doesn't seem to exist anywhere else but in your mind, and—as Nick Shinn suggested with his question "Have you researched user behavior?"—probably isn't terribly practical to apply.

Michael Hernan's picture

Come on guys - this is not a thought experiment - Neither is it a human experiment.
I want to focus on the visual aesthetic problem. I am merely using speculation (the idea that someone else actually thought about this problem before) to suggest a path of action and to build up evidence of similar considerations of how capitals relate to lowercase.

Please argue the substance of what I am saying and not the way I present and formulate ideas.

However that said, oldnick - Though what you say about the garden party and cart before the horse is funny, wouldn't you agree that there was a time when the Roman Capitals and the minuscule letters came together – that there might have been a period of scepticism while the problem worked itself out? It may not have seemed too practical at the time either? Perhaps you can share some of these early examples? This is actually fascinating. Also what do you think about what is being suggested here anyway?

Computer/Human behaviour can wait until I work through this visual problem first.

To do as you suggest - Now that would be like putting the cart before the horse.

William Berkson's picture

As I understand it, the history is somewhat different. The Roman caps came first, but the usual written caps for Latin don't seem to have been the Roman Imperial caps that we know from inscriptions, which are now held to have been first painted with a brush, then incised. They are broad and upright, whereas the Rustica hand is more slanted, with a different stress. The various Roman hands evolved through scribal use into something like our lower case, known as the Carolingian Minuscule. The Carolingian Minuscule was only married to the Imperial caps in the 15th century, at the time type came into being. Then the Carolingian Minuscule was made more upright and retrofitted with more symmetrical serifs, to match the Imperial caps more.

It is a perennial question how bold to make Caps. Historically, Jenson had taller caps and Griffo (as in Bembo) made them shorter to harmonize better with the lower case. But still remaining bolder. To maintain even color (apparent same density of black) with their larger size and different structure, the caps need to be bolder. So contrary to Michael, I don't think that there is a decisive argument for lighter caps to match the lower case.

But you can do it however you want, and if it works visually with a given design, it works.

For myself, I think that you never really overcome the fact that the Caps and lower case were originally different hands, and have quite different structures, except for a few letters. And the difference is used to mark the beginning of sentences and also proper names. In German also nouns. So my philosophy is not to try to minimize the differences between lower and upper case, but to use the contrast for the different purpose of the caps in relation to lower case.

However, I recognize that there are lots of legitimate ways to deal with the issue.

As far as the design of the Caps, I think I remember reading Jonathan Hoefler arguing that more evenly wide caps work best to match the lower case, whereas varied width—eg. Trajan works best in all caps. So that would argue for different designs for all caps. But this again seems to be more of a choice than a decisive argument.

The thing that does seem to be a stronger argument, in my view anyway, is that tighter spacing of Caps works best for harmony with lower case, and in all caps wider spacing looks best as a rule. This was traditionally taken care of in metal by spacing out all caps, and now by tracking out all caps in digital. It can also be done automatically through the "cpsp" feature in open type. However, it seems that in titles custom kerning is generally needed. There maybe you could do more to make the default better, but it seems that even size variation affects kerning, and that custom kerning is needed anyway for best looking titles.

Syndicate content Syndicate content