NEGATIVE SPACE - Letter Spacing.

[nhallam]

This topic was imported from the Typophile platform

I am currently writing a paper that is discussing Massimo Vignelli's quote:

"We think typography is black and white. Typography is really white, its not even black. Its the space between the blacks that really makes it."

I'm looking or any previous writings or information about the topic that I could reference in my paper.

I would have thought that someone would have written about the topic, but I have yet to find any serious writings.

If you know of anything, please post it up!

N

[Dunwich Type]

What have you read so far? I’m confused as to how you aren’t finding anything, because the concepts of counterforms and spacing coms up in just about every typography textbook, most calligraphy manuals, most significant books on type design, and in interviews with type designers.

[nhallam]

J,

Sorry, i should have been clearer. I have of course read a lot in type text books etc. but i am looking for research that explores the importance of good letter spacing vs. non. I am putting forward the idea that the figure ground relationship between the positive and negative ground in type design and in book/poster design, is a vital aspect to the stressfulness and clarity of the intended communication.

As you know a lot has been written about "how to" correctly do this, but im trying to determine "why" it is necessary.

Why is it, for example, that Helvetica is often judged as a more aesthetically appealing typeface than Arial, even when looked at by people who know nothing about type. Being so similar i believe that there must be a sub-conscious psychological reason for this.

[Ricardo Cordoba]

I think that the figure/ground relationship of type is one thing, and the quality of Helvetica vs. Arial is another. The former does not necessarily affect the latter. See "The Scourge of Arial" for more info on Helvetica vs. Arial, and the reasons (not necessarily aesthetic or related to negative space).

For general discussions of the white space between black type, see:

Counterpunch, by Fred Smeijers

Letters of Credit, by Walter Tracy

...and also these online resources:

http://briem.ismennt.is/

http://www.typeworkshop.com/index.php?id1=type-basics

[Manlio Napoli]

See also Detail In Typography by Jost Hochuli.

[ebensorkin]

It sounds like you are suggesting that the "white*" space has a similar role in (illustration?), poster design text, and micro typography. I think this is mistaken. The thing about the white space is that as far as I can see there are many phenomena occurring at once: you have overshoots which is a kind of optical illusion , and the other similar illusions which I won't enumerate here, scale specific illusions**, and then interactions the specific qualities of the design such as regularity ( or not) of forms, Horizontality, and so on. It is a rich mix.

And you might also want to look at the phenomena of "crowding". Do a search for Pelli and crowding. The article "The uncrowded window of object recognition" by Denis G Pelli & Katharine A Tillman in Nature Nueroscience might be the best place to begin.

But back to the why, I think that the reason you don't find much about why is that it is on some level a scientific / optical / neurological question about perception. As designers we try to become sensitive to what is happening & what works and thereby develop an intuitive sense of what's happening. We can observe, name and compare common typographic phenomena - this is probably the case with Tim's book for instance, and is the case when we look at a different phenomena like overshoots. However without insulting that ability which helps us a great deal (quite the reverse in fact!), I think it is safe to say that this kind of activity isn't so much explaining why but what. To get at why you would need to do scientific experiments. And again, this is not to say that you can't get a great deal done by analyzing what. And indeed you can start to develop some theories about why from this and even test them in a soft sense in your own designs.

In terms of Helvetica and Arial, I think that given enough time and attention you will eventually see why one is better drawn than the other in general. But you will also discover that there is not "one" Helvetica, and by this I don't mean Neu Helvetica or different weights, I mean that there are actually different versions. You will also discover that Arial is good at things that Helvetica isn't - some specific screen applications, and in specific rendering environments. As far as it's relevance to your stated topic I have to agree with Ricardo.

*since it may simply be a the contrasting color

** If you look at what works best for type design at various sizes ( and you might want to get Tim Ahrens' new book https://typography.guru/forums/topic/66793-forwarding ) you will find that the designs for various sizes differ. For instance the counters for letters that will be very small need to get larger and the spaces between letters do too. And the features need to be more beefy/crude/large/obvious. Also, experimenting a little you can find that a badly drawn font which is well spaced will outperform a one which is better but which has spacing which is worse. To me this fact alone explains the Massimo Vignelli quote. Ironically it is the size specific quality of the spacing ( look at how close the i & l are set ) that makes Helvetica nice when it is set large and not so nice set small. It is that lack of robustness in the design of its spacing, among other things; that makes Helvetica a very poor choice for signage systems for instance.

[Dunwich Type]

Try reading the section about spacing in Gerrit Noordzij’s The Stroke. I’m pretty sure Cyrus Highsmith talks about it a little in this interview. IIRC Mike Parker discusses it some more in the extra features of the Helvetica DVD, if this is a big paper it might be worth trying to interview him for his thoughts on the topic.

[Chris Dean]

@ nhallam:

Sorkin made some good points. In order to answer the "why" you will need to conduct controlled experiments with a directed research question, a clearly articulated hypothesis and independent and dependent variables. To conduct research at this level, you will probably need about two years of undergraduate psychology specifically focused on cognition. Psych 2000 (or the equivalent of the primary 2nd year psychology course at your institution), a few labs, and an intro stats course should be a good start. And your supervisor will more than likely need to be research scientist, not a designer.

Sorkin mentioned:

Pelli, D.G., Tillman, K. A. (2008) The uncrowded window of object recognition. Nature Neuroscience. 11(10) 1129–1136

This is a well respected journal with a high impact factor. Check out the references. There is a lot of good stuff there, but it's pretty narrow-cast science. Pelli and Tillman (2008) is a review article and is a little more digestible. You may find the reference somewhat obscure, and you'll probably need to get someone who knows stats to help you interpret the results. I usually do as I still haven't got as much math under my belt as I would like. I believe I had to go through the database and journals at my school to get this, which usually requires paid access. My school pays about $3,000,000 a year to have access to all of this knowledge (pier-reviewed journals &c) so you may need to call in a favour. However, I often have great success simply contacting the author directly and asking if they wouldn't mind sharing their research with you.

Larson also has a nice piece of work online:

The science of word recognition.

One of the earlier players in this game:

Bouma, H. (1973). Visual Interference in the Parafoveal Recognition of Initial and Final Letters of Words, Vision Research, 13, 762-782.

A lot of research in this area is still in dispute however. It's fairly hot at the moment. To date, the answers to our "why" questions are primarily supported by convention, aesthetics and intuition, and not objective measures of human performance and empirical data. There are two pretty strong camps on this area so be mindful of both. Often, threads of this nature end up in a "art vs science" debate. Myself, I favour science.

[nepenthe]

G Noordzij's LetterLetter contains some articles about pos-neg space relations, for example "The dimensions of the mental image and its origins" and "The puzzle".

[enne_son]

To my mind, in reading, the entire word is figure. The ground is the surrounding space the word stands out against. One of the great oversights of perceptual psychology, it seems to me, is the failure to understand that within the word as bounded map, both the black and the white are information contributing to visual word-form resolution. Psychology sees the black but has nothing to say about the white.

Underlying this is the principle that the rods and cones in the retina are responsive to reflected light.

[Chris Dean]

Fovial and parafovial physiology may also be worthy of investigation. I'll see if I can dig up a study on Monday.

[J.Montalbano]

Why would anyone take anything Massimo Vignelli says seriously! Have you ever looked at his line of men's clothes? The clothes that he claimed would revolutionize the way men dressed for business. As far as anyone can tell Massimo is the only one to ever wear those clothes!

[John Hudson]

Whenever people start talking about ‘designing the white’ or about how the white is more important than the black, my bogosity meter starts to twitch. It sounds really cool to say one ‘designs the white not the black’, but what does it mean? What is the process? How does this process differ from designing the black?

Peter: both the black and the white are information contributing to visual word-form resolution

I'm not convinced of that or, at least, not convinced that the kind of information contributed by the white is either similar to or as important as that contributed by the black. Letters are positive shapes occupying negative space. The white is space more or less occupied by a given letter shape. Yes, the letter shape creates internal and inter-letter space-shapes, but those shapes a) vary much less than the letter shapes so provide relatively little cue value and b) are not independently featural, their size and shape is defined by the letter weight and style. Letter shapes of different weights and styles may still retain obvious structural and featural similarities, but the white spaces within and around them will not.

Underlying this is the principle that the rods and cones in the retina are responsive to reflected light.

And yet we do not obtain more information from more bright things: we obtain information from contrast which enables us to discern and recognise shapes. When I see a raven fly across a bright sky, the shape of the sky around the raven isn't where I'm getting the information from to recognise the bird: it's the bird that is the shape, and the sky is only providing a field of sufficient contrast to see the shape.

[enne_son]

John, I think designing the white means being attuned to the white as shape in the process of designing, and it means designing so the whites are all in the same key, to use a musical metaphor.

In some words, like ‘look’ the wide open counters are the most salient feature. This is even true for ultra bolds. I can't believe it isn't used. Some recent tests appear to confirm this. Considered from the point of view of role-units, counters vary at least as much and are as rich in cue-value as stems and curves.

What I'm still wondering about is if the white uses the same spatial frequency scale as the black. It could be that it uses a coarser scale. And if it does, what does this indicate in process terms. Is information at the density or granularity of the salient whites resolved first?

It’s convenient that the raven is black. If it were an osprey, you'd be using the white as information.

[Ricardo Cordoba]

During the 1990s, I went to Alberto Breccia's comics workshop for a brief three months, and he had a great exercise for beginners: to train you to "see" and appreciate the white space, he would have you pick a black and white photo from a collection he had. Then, you would have to draw the picture on a colored piece of paper -- red, green, blue, etc. -- and fill in both the white and black areas. You would end up with a high-contrast drawing, and having to paint in the white areas made you appreciate that the white space was indeed part of a dynamic between it and the black. The white was a shape, just as much as the black was.

[William Berkson]

I think it was Frutiger who first said that he designs the whites, rather than the blacks. I think that's a huge exaggeration to make the point that whites are important, which is true.

I think there are a lot of different effects at play with whites. One of the important ones is the phenomenon which I learned from Kevin Larson is called "Mach Bands".

Here from the Wikipedia article of that name is an example:

The white Mach Band is the whiter band you see to the left of the black. (There is also a blacker one to the right.] It is so obvious that you will be surprised that it disappears when you cover the black on the screen with a piece of white paper.

It is also noted by G. Noordzij, though he mistook the cause. The cause, as the article says, lies not in optics, but in the way our brain processes contrasts.

This is one of the things that Helvetica and Frutiger in different ways are designed to take advantage of. The closed shapes of the Helvetica bold e I think result in two Mach Bands enhancing each other in the 'gap', causing the letter to look bright and scintillating. The smaller eye of the e in bold weights also makes the Mach bands more vibrant. Helvetica really does take advantage of this.

Frutiger does it in a different way. If you compare the e's in Frutiger and Myriad, you will see that the lower terminal in Frutiger swells lightly, producing a sharp angle in the black and a more vibrant Mach Band effect than in Myriad, which tapers and is much less vibrant and assertive, even though the difference in the outlines is very small.

[Sorry I don't have the time to copy and post the pictures]

The way Mach Bands are handled is an important feature of whites, but only one. Also it is not necessarily desirable. In a text face, you may well want less shimmer, and more quietness.

There are a lot of other issues, and I think it's important to recognize that there isn't just one effect involved.

[Chris Dean]

@ enne_son,

"Some recent tests appear to confirm this."

Citation(s)?

[ebensorkin]

I would like to again raise a distinction between between John's model and a notanic model in which the forms are not opposed but linked. Perhaps rather than linked I should say inseparable. Take the example of the bowl. A bowl is one because of the empty space made inside it. It would be silly say the most important part of a bowl is the empty part of the bowl or the body of the bowl. If they are a bowl they are not opposed. The opposite is true. I do want to say that despite my preference for thing about B&W in terms of notan that I don't think that this means that black & white have to be the same, equal, or equivalent. Far from it. In fact I would be surprised if that was so. Inseparability does not preclude difference. But at the same time as far as I know we don't have substantial research based reasons to characterize either one's role. If I am mistaken I would love to be corrected. Obviously type is not bowls or physical in the same way and type isn't one shape. But despite all these differences it seems to me that the reality that opposition is nonsense carries through. I realize that this analogy may sound like Taoist propaganda to some and that is okay. I accept that.

But rather than dwell on white vs black, or notan or not notan, I would suggest that the interesting way of looking at this is to think in terms of features and their clarity or ability to be detected.

The fact that they are in B&W just happens to make things less complicated. When you detect an animal in the forest - perhaps a deer, your image may be incomplete because of bushes grass or brush; or because of tree trunks or branches, but if you see enough you mentally assemble the parts into a whole deer. This might be in full and posisbly very flat color which is low contrast or it may be in near silhouette. But in all cases in we are assembling recognized features, if given enough of them, into recognition of a 'deer'.

Deer in silhouette: http://www.flickr.com/photos/harve64/2271532149/

The process of recognizing letters is a similar process. We are recognizing features ( Peter would say salient features ). And it is the job of the black & white (however characterized) not just to make the features themselves easy to detect, but also to group, or as Pelli would put it to Integrate them into a whole 'a' or 'g'.

See: Crowding is unlike ordinary masking: Distinguishing feature integration from detection http://journalofvision.org/4/12/12/

So "white's" role, inseparable from "black's" or not, is to facilitate this process.

Bill thanks for the Mach band.

Peter : In terms of tuning the white not simply as a necessary aspect of the black but a a feature, in the way that we look at thins & thicks in black strokes, stem widths, and so on - what aspects are you thinking of? I assume you mean things like wordspace, letterspaces and countershapes. Perhaps lines spaces get a look in as well. One other kind one occurs to me as well: proximal spaces such as with punctuation, but also within glyphs such as between the dot in an ! and the stroke above as well as in the openings or spaces of: a e, c 3 & 5, U & H, f & t, and so on. Do you have others you would like to bring up? Is this what you mean?

[John Hudson]

Eben: So “white’s” role, inseparable from “black’s” or not, is to facilitate this process.

Yes. By providing a contrasting ground. :)

[William Berkson]

One of the other issues in dealing with white is figure/ground ambiguity or lack of it. In the West, 'notan' seems to most often refer to illustrations in which the white is used as foreground, such as snow on a mountain in Chinese and Japanese brush painting with ink.

In Helvetica Bold or Black, and other bold or black type faces, I think the counters tend to pop out partly because as the black dominates more, so the eye may want to read the white as 'figure' rather than background. I don't know how much the effect in Helvetica Bold is due to Mach bands, and how much to figure/ground ambiguity, but I suspect both are at work.

In fonts for extended text, it seems that figure/ground ambiguity is actually a bad thing. You want the blacks to be more salient, to be 'read' and the whites not to read. One of the most interesting things to me in the testing of Clearview is that medium weights are more legible at a distance than bold ones. In text fonts, I think you have something like 1/3 black 2/3 white on average between the base line and x-height. That keeps the white as ground and black as figure, and avoids any ambiguity.

That's where I'm with John, where also my Bullshit Detector--that's what we called it in the midwest when I was growing up--starts ringing with phrase "designing the whites".

The whites are critical, but I am doubtful that their actual shape is as salient a factor as that of the blacks. For example as well as 'dark-light' (notan) being one balance in Chinese theory of ink brush painting, 'thick - thin' is another. When you alter the thick-thin balance of the black in a typeface it is very noticeable. But you are only altering the white in a counter by a few percent, where you might be doubling or tripling the thickness of a thin. And the eye responds to small absolute but big relative change of the black in a way that shows to me it's more salient, i.e. has more impact.

This is not to deny that we 'read' closed counters, etc., as Peter argues. That may be. And having the right letter spacing, and counters in proportion to the right letter spacing is no doubt important. But I am skeptical that the details of the shapes of the whites are as important as those of the blacks.

[John Hudson]

Peter: I think designing the white means being attuned to the white as shape in the process of designing, and it means designing so the whites are all in the same key, to use a musical metaphor.

In my experience, the ‘white as shape’ is a by product of the shape of the letters. As for getting the whites in the same key, I think this happens as a result of getting the letters in the same key (which increasingly I think of in terms of tuning to a consistent spatial frequency channel).

In some words, like ‘look’ the wide open counters are the most salient feature.

I think that's begging the question. Certainly when looking at the word ‘look’ the counters of the double o are the largest features, but does that make them the most salient when reading? I'd argue that the most salient features from a word recognition perspective are those that distinguish ‘look’ from ‘lock’, the closest confusable word in our language. It is not the counters that are salient, but the rings enclosing the counters.

Considered from the point of view of role-units, counters vary at least as much and are as rich in cue-value as stems and curves.

I don't think that is true, at least not for the Latin script. The number of counters is far less than the number of letters, and almost all those counters are limited to the x-height range. Of these, most are of similar size and related shape (making possible the historical use of the counterpunch). Indeed, most of them differ from each other only by virtue of the difference in the arrangement of their bounding stems and curves. Again it is the shape and weight of the letter the wholly determines the size and shape of the revealed ground.

Fred Smeijers gives the example of ‘not’ and ‘hot’ -- two words with identical internal white space --, and notes that the preponderance of letters with one or more vertical sides greatly reduces the variety of inter-letter white shapes.

What I’m still wondering about is if the white uses the same spatial frequency scale as the black.

? As I understand it, spatial frequency is a measure of transition between contrasted objects. I'm not sure that it makes sense to question whether ‘white uses the same spatial frequency scale as the black’, because what spatial frequency refers to the interaction of the white and the black.

[John Hudson]

re. Mach bands...

Bill, I fail to see the relevance of mach bands to type, except in terms of antialiased text on screen. The mach bands phenomenon appears to rely on a gradient transition between dark and light. I also note that the Wikipedia entry discusses and illustrates this in terms of dark and light, not black and white. Finally, the mach bands phenomenon is scale-dependent, as can be demonstrated by putting your illustration onscreen and then looking at it from different distances. So I don't think it is possible to say that a given typeface design exploits this phenomenon, other than in particular rendering modes at particular sizes.

[Chris Dean]

A simple experiment to run regarding counters would be to run a reading comprehension test using the counters as an independent variable. Condition 1 would have a normal typeface, condition 2 would have the same typeface with the counters filed in. You'd have to outline the entire page and manually remove the counters, or customize a typeface, which would make materials preparation a bit time consuming. I'm pretty confidant that no one has conducted a study of this nature (but I tend to forget what I read, unless I've read it three times). A quick database search (PsychINFO and Web of Science) didn't turn up anything at a glance. Counterbalancing a 2x2 within subjects design isn't that hard. Data entry and analysis would be a breeze.

Ss 1 = A1, B2
Ss 2 = A2, B1
Ss 3 = B1, A2
Ss 4 = B2, A1

A = counter
B = passage

A simple t-test should do for data analysis (keeping in mind I'm not as advanced as I would like to be in stats-land. I prefer to spend my time on experimental design. But for fun, I have taken to developing three dimensional Latin squares. I'll get there someday).

@ Sorkin, thank for the reference to Pelli et al. (2004). I wandered around the paper and ended up at a great link to the Sloan font (scroll down about three screens and there's a little blue link to download. It's free). Strangely enough, the same day I read this, I had a doctors appointment. They are moving offices so they were throwing stuff out. I noticed an old eye chart in the pile *yoink!* Check out that capital G!

(I appear to be having problem inserting images. It keeps saying "Could not copy image. Error. " Anyone?).

[eliason]

(I appear to be having problem inserting images. It keeps saying “Could not copy image. Error. ” Anyone?).
Is there a space in the filename?

[John Hudson]

Christopher: A simple experiment to run regarding counters would be to run a reading comprehension test using the counters as an independent variable. Condition 1 would have a normal typeface, condition 2 would have the same typeface with the counters filed in.

What's the purpose of such an experiment? No one is denying that the existence of counters is important, and removing them doesn't affect the debate about the way in which they are important. I maintain that they are by products of the shape of the letters: by removing them you are changing the shape of the letters. Also, letters include both open and closed counters and, arguably, any internal white space constitutes a kind of counter. Where would you draw the line on what to remove? Is the space between the arms of a k a counter?

I'm not sure if there is a reliable test for the independent significance of counters in reading, because there is no such thing as an independent counter. You can't obscure the counter without obscuring the letter shape, and we already have good research on obscuring letter shapes (such that we know what parts of letters are essential to recognition).