Well, consider the two shapes in the image below

they have the same area (‘volume of water’), do they 'look' equivalent? Even if you can calculate the volume of intra/interspaces, trying to equalize volumes (numerically) is a very crude approach, I suspect.
I don’t know much about perception but I think you need to consider other factors, like the 'compactness' of a shape, the absolute dimensions, etc.

daniele

"I think you need to consider other factors, like the ’compactness’ of a shape, the absolute dimensions, etc."
Certainly. I sure hope the adherents of the "glass of water" model wouldn't apply the principle mechanically/mathematically (that would be scary), but adjust the space optically – but still based on the "glass of water" idea. (FWIW, that's what they taught me at school.)

Still, the metaphor is there and I wonder how much merit it has. How elastic can a glass be (esp regarding the point size question) before it stops being a glass? :->

As a PS, I just remembered a slightly different metaphor used by Jost Hochuli, who talks about the white as light shining into the lettershapes, which allows for slightly finer gradations (like he says light coming from above is stronger, which is why a sans serif "u" needs to be narrower than an "n") but still departs from the idea of equal "light" inside and around the letters.

Thanks for having quoted the light metaphor (I didn’t know about it), I like it. My hypothesis is that the liquid you have in the glass is not homogeneous and/or it can be 'activated'/'de/activated' by the boundaries of the glass.

Cool, sci-fi plasma fluid! :-)

I can look up the light metaphor and quote it more precisely if you're interested.
But I read it in a book I don't have, so I'd hit the library tomorrow.

Nina, Yes, generally, larger type needs less interletter spacing. Other things occur along with the visual mass concept (glass of water theory expanded to include perceptual bias). A big factor is proximity of nearest segment--meaning a serif on a cap V being near a Serif on a cap W. When the two are quite close, the proximity factor begins to overcome part of the visual mass of space issue. It is often impossible to have a space as those between upright adjacent letters like H and N be truly equal to adjacent opposing diagonals like V and Y unless you vastly letterspace. Using the proximity effect minimizes the perceived difference though.

ChrisL

That glass of water is not drinkable - dump it down the drain.
Proximity of parts is a central parameter.

hhp

Thanks guys. Sounds like the "glass of water" model, at the very least, isn't accounting for everything that needs to be considered.

"Yes, generally, larger type needs less interletter spacing."
Thanks for confirming this, Chris.
I still wonder why, if anyone is up for explaining that. Maybe something to do with there being absolute limits to visibility of interletter space (especially related to the proximity parameter*)?

* Which I can see now is probably included in the concept of "perceptual distance" quoted above (I read "distance" as a linear concept rather than [only] volume-based).

It could be like a visual Doppler effect where sound appears to increase in pitch as it gets closer,

ChrisL

?
Sounds mysterious! (I know what the Doppler effect is, but I don't get how the translation works.)

It is an effect of perception, not an actual effect on pitch of sound. Translated to looking at small type, our perception may make a feeling of closure to neighboring glyphs, at some point, almost a sense of gravitational pull takes effect but only in our eyes.

ChrisL

Tracking

No offense Chris, and I'm not contesting what you said, but this sounds a bit esoteric. :-)
And maybe this stuff is just that – but I guess I'm wondering if there are any attempts to find out how this funny effect of perception* comes about.

* By the way, I always thought the Doppler effect is an actual effect on pitch of sound, in the way that sound waves emanating from a moving object in the direction of that movement (relative to the observer) are "squooshed", making for a different received frequency – not just the brain playing tricks? I may be mistaken though.

I post "tracking" when I want to follow a thread so it "bookmarks" it in my account settings. RSS seems to not work on my current computer. Funny miscommunication given the content of this thread eh? None taken.

Christopher, I was talking to ChrisL. I only got confused about your "Tracking" posts the first twenty times or so I saw them. :->

@ A,

That made me laugh. Thank you.

Nina, the sound of a car engine at the source is a given pitch and does not change. Sound moving through an air mass with dense atmosphere can alter what hits your ear some distance away but does not truly change the source of the sound (the car engine). Granted, I am making a metaphoric comparison, not a scientific one--so far, none exists. The ratio of letterspace to glyph does not change in reality but it appears to. Since the eye is always the final arbiter, we have to appease then not a measuring device. There may come a day when science has a more measurable reason for the difference and can predict the proper distance of change but it is just easier to make your change visually. Take a font designed for text reading and another for display from the same family and scale them to the same size. You will see the difference but you may not find a ratio of change that suits all families of type.

ChrisL

"the sound of a car engine at the source is a given pitch and does not change."
Of course not at the source, but in this case, in the space between the source and the ear; I'd put that in the realm of physics, not perceptional psychology. Whereas with the issue of spaces between small lettershapes, I wouldn't expect some light rays to physically go astray on the way to the eye – but rather something strange to happen inside the eye, or the brain? But that's conjecture, and maybe also semantics rather than anything else.

"Since the eye is always the final arbiter, we have to appease then not a measuring device."
Oh that point is totally clear. And I wasn't looking for numbers (how much to adjust it by, or something). I'm just very curious to learn how perception works, maybe overly so. If you say science doesn't know yet, that's an answer.

Nina,

My sound analogy probably was less helpful than confusing :-)

We actually only anecdotally know that it works--even preferentially if you count legibility studies where "increased letter spacing" seems to be preferred by subjects of studies at reading text sizes. I don't remember the names of the studies any more and am sadly only left with what the perceived conclusion was. We are all curious to find out more about how perception works. There have been numerous studies but the why and how to part still does not seem clear enough to help a type designer much more than trial amd error. Oh, well, sorry I could not be of much help.

ChrisL

I might have said this before but I always thought a surface area calculator would be a useful plug-in. Not to be relied on, of course.

"And I wasn’t looking for numbers (how much to adjust it by, or something)"
Following, Briem’s page on spacing you always have the type you are employing to guide you. So, if you are composing type, and you want to see what tracking works, and how the glasses of water are equalized, set a line like iiiniii in the face you are spacing, or IIIHIII if caps, (the i's and l's have 'no water inside' and the n or H show you the how the average inside space relates). With tracking you can adjust this text sample so that the space between i's and the counter of the n are even, or as I and many others prefer, with slightly less space outside than in. Then, you can apply this tracking to broader text.

"I post “tracking” when I want to follow a thread so it “bookmarks” it in my account settings."
...but like so many things we are all free to do, if we all did this...

Cheers!

> But it seems that the inter-letter space needs to be adjusted more for smaller sizes than the intra-letter space usually is

Nina, can you explain what you base this statement on? I'm just interested in understanding your frame of reference. I want to make sure I'm reading you right. You're saying that it seems that for smaller sizes the space inside letters is usually adjusted more than the space between letters. Did I get that right?

Are you thinking about a comparison of text to small text, or are you comparing display to text? (I think these are definitely two different paradigms, so I want to understand which you're talking about.)

And are you looking at/thinking about metal type (which was inherently optically scaled) or current "opticals" like what Adobe and others have done?

-- Kent.

The glass of water thing in Briem is seriously misleading if you take it too seriously. Looking visually equal and actually being equal in area are two different things. I've tested it by measuring inter-letter areas between nn and oo on well-spaced type. They're not mathematically equal, even though they look nicely balanced.

As well as visual equality of inter-letter space, making the outer space proportional to inner counter space, keeping adjacent parts from visually crowding, keeping a (not exact) regular rhythm of verticals and white, and even adjusting for weight of strokes--left vs right side of M--are all things to consider and weigh. And they are sometimes in conflict with one another.

I certainly don't claim to know the ideal way to balance all these. Then again there may not be any ideal, but various compromises.

Oh, and it does change with optical size. Further, the tightening of display vs text is probably best not simply tracking tighter. The oo and ll may work a little differently--at least that's Slimbach's view.

With tiny letters, the whitespace is tiny too...so that the eye needs to be much more acute if the letterspacing isn't increased (due to clotting). So there's an inverse relationship, but I don't imagine that it's a linear one. And I imagine that the intra-letter space doesn't change so much because the letterforms themselves are still recognisable as long as they're not crunched up next to other ones? However, I think also that the x-height tends to increase as your point size goes down too?

I don't know enough, again! :)

Could it be that there is a very simple technical explanation/reason for larger sidings in small type vs smaller in larger type? Eg the fact that there is a lot more wear and tear on the parts of metal type that are close the the edge?

I imagine a natural process having taken place where the typecasters explored this and came to the evident conclusion that there is a minimum amount of sidebearing needed to keep type in good shape (at least for a period long enough…).

Readers would of course become accustomed to this and notice discrepancies from this practice, making it kind of obligatory to continue, even after we left metal type behind.

. . .
Bert Vanderveen BNO

[This should be above ChrisL's post! I had to fix a vicious typo.]

Kent:
"You’re saying that it seems that for smaller sizes the space inside letters is usually adjusted more than the space between letters. Did I get that right?"
No – I meant it exactly the other way round! As in, the spacing needs to change more than the letterforms/counters. Sorry if I was being confusing. :-/

"can you explain what you base this statement on?"
OK, here comes the big disclaimer: Yes, I was thinking about text vs. small text. But to be honest, I wasn't strictly "looking" at anything – more drawing from what I'm used to seeing. I'll gladly go and see if I can find examples to illustrate this.

Though now I'm wondering if a lot of what I have been seeing is not actually the proper way of doing things, as I'm probably somewhat used to seeing text set at less than the intended point size, and tracked out (hell, they taught us that at school, too). But the interesting thing is, that still is more readable then than when it's not tracked out (and the counters vs. spacing ratio stays the same). And the opposite – a really wide font for small text that's spaced tight – is definitely not going to be easily readable. Which brings up the question if inter-letter spacing is by definition more of a deciding factor for readability than the space inside the letters.

***

William, thank you for this thoughtful post! FWIW, of course my interest in spacing has been triggered anew by thinking about spacing my font; so this list of criteria you posted seems most helpful in a practical way too.

Hey Ben, you might be on to something there – the concepts of clotting, and "acuteness" of the eye, sound pretty plausible to me, in the way that it might be an issue of resolution (both of the eye maybe, but more importantly paper and ink), so that the letters need to be separated from each other a bit more? (Which again would demonstrate the heightened importance of spacing over the intra-letter white for readability.)

Daniele: I'm about to hit the library and dig up that "light" quote.

…ah, so much to learn. :-)
By the way, if somebody feels I shouldn't post and ask the experts about such things before doing my own research, I can see their point. My hunger for learning is sometimes more urgent than I can read everything I'd need/want to right now; and I love these sorts of threads – but please tell me if I'm being too demanding.

Nina,
Post away. You are not too demanding at all! Your posts are intelligent and insightful!

ChrisL

And, thanks Chris! I'm glad to hear you say that.

Nina, the passage from my Typo #13 text you quote above is about design, more specifically, what I call role-architectures, or construction. “Perceptual distances between letterforms must be great enough to minimize criterial cue confusions or masking. But not be so large that cross-letter binding or constructive conjunction is inhibited.” Perceptual distance is a cognitive psychology term for how dissimilar or similar letters are. Hrant did some really nice work on this years ago.

Perceptual distances are calculated from letter confusion matrices, which are experimentally derived. For example an a and an e are more often confused than an a and an m, so the perceptual distance between a and m is greater than between a and e. Herman Bouma did some nice charting of perceptual distances that corresponds nicely with Hrant's chart of similars and modulars.

My comment on spacing follows the passage you quoted. I say: “If spacing between letterforms is not in synchronicity with the space inside enclosed forms—with the way the letterform set uses its cartesian space—there are computation costs at the visual cortex level, because the saliencies can’t compound effectively--the ‘saliency maps’ are thrown off. Configural cues become harder to assemble. Conjunctivities are weakened. The wholistic response bias compromised. So “space craft” facilitates effortless perceptual processing.

I see one factor relating to spacing, and three that relate to scaling for small sizes. For spacing, the blacks must be in step and the whites in synch in spatial frequency terms; for small sizes, perceptual discrimination affordance to the role-unit and evoked-form level must be intact, the ink-spread factor must be accommodated and the dependence of vision on the white must be kept in mind.

Bear in mind that ‘in step’ and ‘in synch’ are not mathematically quantifiable. In step is not equidistant and in synch is not equivalent in area. I try to think in terms of rhythmic cohesion with the whites, and narrow but not exact phase alignment with the blacks. Fourier transforms, I believe, can gauge this. The method Berlow describes is a way toward achieving it.

Perceptual discrimination affordance is the visual cortex being able to make out what something, for example a letter-part, is. There are size, crowding, and acuity thresholds here. Making out what a letter-part is means resolving marks into shapes the visual cortex knows. I call this process of resolving ‘quantization.’ Quantization underlies categorical perception, the psychological term for ‘seeing as.’ Seeing a d as a d for instance.

In relation to the dependence of vision on the white, bear in mind that vision uses reflected light, and the black does not reflect, so if the white is overwhelmed by the black, the fundaments of readability — perceptual discrimination affordance, quantization, categorical perception and rapid automatic visual wordform resolution — all suffer.

Topic Deviation: On InDesign, you have the option of 'optical' kerning as opposed to 'metrics' kerning. I realise kerning is not the same as tracking, but should this option be able to add extra tracking to smaller sizes?

And how does 'optical kerning' measure the amount of space between letters? Is it shortest linear distance between glyphs or does it calculate the area somehow magically?

(Sorry if this isn't quite relevant.)

Nina -- No, it's not you being confusing, it's me having a hard time keeping inter- and intra- straight in my mind. I don't know why this pair always trips me up. Sorry.

Okay, I wanted to make sure you were talking about the text realm, because in this case I have some visual material that you may find pertinent and interesting. And I suspected that you were drawing your conclusions from relatively recent practice. Which is still somewhat removed from the standard practices from the metal era.

I'll post a visual a little later when I get a moment to prep the scans.

-- K.

Nina, thanks, that post is the result of my effort over some time to answer the question you asked, a question I also had. I still don't have any definitive answer, but I feel that at least I better understand the dimensions of the problem.

>so if the white is overwhelmed by the black, the fundaments of readability — perceptual discrimination affordance, quantization, categorical perception and rapid automatic visual wordform resolution — all suffer.

Very interesting, Peter, this might be why bold works in display but is less readable in text. In larger sizes the white has enough 'presence' for the eye to do its job, but not so much at small.

Bendy, in this recent thread Kent Lew brilliantly analyzed the damage done by 'Optical Spacing'.

> The glass of water thing in Briem is seriously
> misleading if you take it too seriously.

Yes, please take it frivolously. Basically, not at all.

> the tightening of display vs text is probably best not
> simply tracking tighter. The oo and ll may work a
> little differently

Certainly the best example is the "r":
the spacing of its right side is largely
independent of tracking!

> at least that’s Slimbach’s view.

Are you "reverse-engineering" here,
or actually citing something he's said?

> I think also that the x-height tends to
> increase as your point size goes down too?

Indeed - and this increases the letterspacing needs further.

> before doing my own research

Hey, this is your own research!

hhp

Peter, thank you for stepping in and saving me from completely misreading the very passage I quoted! :-/
For the record, I thought your text was very well written, and quite understandable for a novice like me – I should have paid more attention. (I always find it hard, in a new field, to find out which terms are "cuewords" that have entire theories attached to them, and which ones can be understood more-or-less by the basic meaning of the words. Here [in "perceptual distance"] I assumed the latter, probably also seduced by the following paragraph, which is about spacing, like you said.)

I find your post extremely illuminating – thank you very much for explaining!
This is a healthy dose of food for thought, which will need some time (and re-reading) to sink in.

"Perceptual distance is a cognitive psychology term for how dissimilar or similar letters are"
Ah, I see! I've come across this concept (mostly, actually, in Hrant's work, like his paper in "Graphic Design & Reading" [now I sure hope I'm not mixing up things again]).

"For spacing, the blacks must be in step and the whites in synch in spatial frequency terms"
Ahha! So to think in terms of linear distance for the black, and area for the white – a little illumination in itself!

"if the white is overwhelmed by the black, the fundaments of readability […] all suffer"
… but would this hold true for intra- and inter-letter space/white equally, or is it more crucial for the latter?

***

Ben, there was a recent thread (I think it was titled "Opinions on Meridien for books") that involved a somewhat elaborate discussion on InDesign's optical spacing – now I'm trying to find it again…

***

[I need to think/type more quickly.]

"Hey, this is your own research!"
I love this place.

Kent, no worries – I thought I might have mixed up the "inter" and "intra" myself.
I'm looking forward to visuals! Meanwhile, I'll be typing up that "light" metaphor that Daniele wanted to hear. Gonna be curious what the jury says on that.

**

Re the glass of water: If this is so dauntingly limited, how can this continue to be a popular metaphor that is even taught to design students? Is that like they teach you the atom model at school in a way that is scientifically wrong, but at least they think it's understandable enough for students?
What Peter said above about how to gauge the black and the white in terms of distance and area; and William's points about what needs to be considered in spacing; sure, that is more involved, but also a lot more tangible, and real, than imaginary glasses of imaginary fluid being filled into… counters. :-{

**

Ben, for the record: the thread I was referring to is the one linked to by William Berkson above.

If this is so dauntingly limited, how can this continue to be a popular metaphor that is even taught to design students?

Most of what design students are taught is dauntingly limited. But in many cases it’s the best starting point we’ve got, and it’s up to the students to let their imaginations and judgement take them beyond the limits of pedagogy.

>Are you “reverse-engineering” here,
or actually citing something he’s said?

See page 53 of this Adobe booklet: www.adobe.com/devnet/font/pdfs/5091.Design_MM_Fonts.pdf . I was talking to Lucas de Groot after his TDC presentation (with Miriam Bantjes--what an all star event!) and he referred to this paper as a hidden gem and said that "Slimbach himself" wrote most of it. The illustration also uses Minion.

As promised, here's Jost Hochuli's alternative to water, light. It's actually not half as exciting as I thought I remembered – I should have quoted it a few hours ago, considering how far this thread has taken us from when I first mentioned this!

This is from "Detail in Typography" (1989); my library only has the German edition, so here's my wobbly and definitely unquoteable translation:
[First he goes on talking about the importance of making everything evenly grey with no "holes"]… "Even though the spaces between the lettershapes are not [mathematically] equal, a well spaced word looks even; so this can not be about equal space in terms of areas of space. If we replace the term "space" by the term "light", everything becomes a lot easier, and we can do without unclear terminology…
The light – the lightness of the letters' environment, the ground – flows into the counters and the spaces between the lettres from above and below; the light that comes from above is more active than the light from below. This brings about the fact that the letter "n" in a sans serif needs to be a bit wider than the "u" in the same font if they are meant to appear to be the same width optically. In much the same way, the area between "I" and "A" needs to be less than that between "I" and "V" (assuming the same angles on the "A" and the "V"…). This phenomenon cannot be accounted for with the theory of equal areas of space, but very much so with the theory of equal light."

I should add this is a tiny book for students/novices. I haven't read anything "bigger" by him.

"(assuming the same angles on the “A” and the “V”…)"

I have been making the angle of the A more open than the V to overcome the darkness that comes with the addition of the crossbar in the A.--Not a huge amount though.

ChrisL

Nina --

For your consideration (and without any analysis or conclusions from me), here are examples of fitting sequences taken from the Advance Proofs for two different sizes of Dwiggins's Falcon, scaled to equivalence. The upper sample is 10 point; the one beneath is 6 point.

 
An Advance Proof was a standardized final proof that was made by Mergenthaler Linotype for each font of new type before the matrices were passed for stock and released to the trade. One portion of the proof was devoted to checking the fitting intervals and base alignments and consisted of each character in the alphabet alternating between a "square" and a "round" letter (i.e., H & O for capitals and m & o for lowercase).

So, this offers an excellent opportunity for us to peer into the kinds of adjustment that were being applied in terms of spacing and proportions, at least in the Linotype approach (ca. 1950s). Griffith & co. prided themselves on carefully adjusted individual drawings for each character of each size, not mere geometric or photographic "blowing up" or "blowing down."

BTW, Griffith is quoted as expressing that he felt the Falcon series represented "an almost perfect proportional curve to the eye." (See Walter Tracy's Letters of Credit, page 53 for further discussion of this.)

It just dawned on me that if one thinks the "light" theory through, Chris, even without a crossbar the "A" would have to be wider than the "V". :-/
Simplification, I detest thee.

Simplification is a myth in type design. It takes a great deal more work to make it look simple than to make it look complicated.

ChrisL

William, am I glad I asked!
That looks like a major hidden gem alright - thank you.

BTW, looking through that quickly I saw a graphic that
Raph posted to Typophile recently*, except he referred to
it as "Adobe Tech Note 5091" which made it sound boring. :-)

* http://typophile.com/node/54815

hhp

Here, for the full effect, is a collage of comparisons from the full set of Falcon Advance Proofs I have, in descending point size -- 12, 11, 10, 9, . . . 7, and 6. I apologize: for some reason, I don't have the 8-point proof.

 
[Sorry for the delay, which separated my posts. Seems we've got a few subthreads going simultaneously.]

I teach the glass of water, because it is the best way to explain how basic sidebearing width works, and for students to get their fonts to space nicely.
I use O, I, and V in Futura as examples.
(Admittedly, this is somewhat device-dependant, as Karsten has noted.)

However, in theory I believe that the perceptual process of reading is multiplex, and that they eye is aware of not only the area of positive and negative space, but the weight of strokes and, as Hrant notes, proximity, or the distance between elements. There is also the issue of scalar versus absolute relationships, the quality of sharpness and the shape of curves. And there are of course visual dimensions of design that we are not aware of and cannot express.

The reader's eye sees all these different raw visual qualities, and the brain synthesizes meaning.

Chris, I meant simplification of theory, not in design.
(What is it with our misunderstandings?! :-) )

Kent: Wow, thank you – this is great! I will need to (and gladly will) look at it in detail, and figure out what it can tell me with regard to the questions I had/have.
I won't be able to keep up the current pace of this thread with that, though.

Hrant: that thread looks like it could actually be more interesting than I first thought it would be. :->

Nick, I just realized that spacing at my school was also practiced with sans serif caps with classical "Roman" proportions (though we drew those ourselves first) at huge sizes – which is a pretty different animal (I'd expect) than spacing a text font, so maybe some simplification should be allowed for… apart from the fact that most of my classmates already were annoyed by type. :-/

"The reader’s eye sees all these different raw visual qualities, and the brain synthesizes meaning."
Beautifully said. I find this exact process to be infinitely enticing (and especially so when it includes actually designing what the reader sees) precisely because it's so complex, not really understood yet, and because it's so close on the border between research/science and design; a tectonic fault between two worlds that is calling for bridges to be built, from both sides.

Thank you guys, this is a great discussion. I think my head will explode soon. And my type class is only coming up this weekend! :-)

For giggles, I'v superimposed an inverted Myriad Pro A onto the letter V to see the differences in the optical adjustments made by Robert Slimbach.

The A is certainly wider, and notice the notches cut to make the counter more "open". I'm curious why the same notches weren't needed in the V?

"I’m curious why the same notches weren’t needed in the V?"

The V does not have an enclosed counter to battle with. Also, the V joins at bottom along with the W which it is more often visually compared.

ChrisL

That belongs in a gallery, Dan.

Thanks for translating the light theory, Nina. I’ll be sure to read the English edition of the book. And Kent, those scans are great!

On the subject, what are people’s feelings on using fine optical adjustments to control spacing such as in the letter A above? Is it acceptable to shape the counter that way from the start, or should that sort of thing be considered trapping and only done to deal with ink spread?