If bold or sans-serifed versions of a typeface are considered less legible, than this is explainable from the point of view that the archetypal model was never developed with these derivatives in mind. [ Frank ]
[ perhaps off-topic ]
Frank, interesting proposition! I only have a glancing familiarity with your work, but do you have a sense of how an archetypical model for the sans-serif — or a sans-serif like Neology (that employs a new concept of style-fusion) should differ?
Frank, Luckiesh did not address the issue of sans serifs, as these were not used for text at the time, in the US at any rate. However, he did study boldness, and I think his work shows conclusively that the reading of bold is influenced by the human apparatus for visual processing and not only by the "archetypal model," as you suggest might be a possibility. The archetypal model, while very interesting and important is not the whole story.
In Luckiesh's studies, he found that bolder type was consistently more visible than less bold type. —He measured boldness independently from visibility by reflectance from the page, and visibility by the contrast method I mentioned earlier in the thread. This greater visibility I think explains why bold type is suitable as titling: its greater visibility defeats the way crowding hides words in a mass of other words. The bolder type was equally quickly read, but resulted in a higher blink rate over time, indicating lower ease of reading, or "readability" by his definition. So all in all his studies on boldness explain why it is less suitable for text, but more suitable for display.
Again, this is all with reference to the human visual processing apparatus, rather than archetypes of the original designs. Presumably, these same kinds of advantages and disadvantages of bold type would hold for non-latin scripts, for example.
Peter: ‘but do you have a sense of how an archetypical model for the sans-serif […]’
I will try to explain this briefly (although the latter is not my strongest point ;-).
The fitting Jenson applied on his roman type shows an optimal combination of balanced white space and a regular stem interval. Also Griffo’s type for the Hypnerotomachia Poliphili shows the same balance (see below).
The distance between the stems is dictated by the white spaces in the letters, which are all related to each other. The length of the serifs helps to preserve the space between the letters. Or, the other way around, the serifs work as wedges and help to force the letters in the rhythmic system. Jan Tschichold briefly mentions the stem interval in Treasury of alphabets and lettering (Ware, Hertfordshire, 1985 p.34): ‘The old lettering masters followed the rule that all the basic strokes of a word should be spaced at approximately equal distance. This rule is disregarded today; lower case letters are pushed together’.
This optimal rhythmic system basically only works for text sizes, i.e., roughly 16 pica points (Jenson) and smaller*. Because of the lack of serifs, it is impossible to apply such spacing on sans serif typefaces, with exception perhaps of condensed versions, in which case there is not much space anyway. The figure above shows the serif version of DTL Haarlemmer. The space between the perpendicular letters is optically equal to the space inside the letters. The serifs make it impossible to tighten the spacing more, because they would collide then.
The next figure shows the sans-serif version of DTL Haarlemmer. Although the equilibration of the spacing is obvious, the stem interval of the serif version could not be maintained. The stem interval in sans serifs with proportions related to the archetypes is by definition disturbed.
* The archetypal model was purely meant for these sizes.
…this optimal rhythmic system basically only works for text sizes, i.e., roughly 16 pica points (Jenson) and smaller. Because of the lack of serifs, it is impossible to apply such spacing on sans serif typefaces…
So you are saying it is physically impossible to find an archetypal system that facilitates an optimal combination of balanced white space and a regular stem interval for sans serif typefaces because of the lack of serifs?
In a presentation last year to members of the Configural Processing Consortium at their annual meeting here in Toronto I said: in psychophysical terms, the optical-grammatical or gestalt concerns of typographers and type-designers translate as a concern for proper salience and distinctive cue value (when it comes to individual strokes and counters), robust categorical perception (when it comes to letters), and cohesive equilibrium (when it comes to words).
When balanced white space and a regular stem interval combine optimally, there is cohesive equilibrium.
However, types with thin horizontal stems and serifs create a problematic “fence post” effect, because the white space is so poorly defined. That effects all three concerns you mention, all of which need to be addressed for optimal readability.
This may be why Bodoni Bold was traditionally (early and mid 20th century) preffered as the text weight.
I have for long felt that sans cannot be optimally spaced, for reasons similar to what you say. It also favors sans that are based on the oval rather than circle (somewhat condensed) as more readable. I'd love to see studies on Sans Serifs with blink rate, as Luckiesh did, to see whether this measure of readability shows a difference.
Nick: ‘However, types with thin horizontal stems and serifs create a problematic “fence post” effect, because the white space is so poorly defined. ’
These types deviate from the archetypal model. Bodoni’s flexible-pointed-pen-based type and that of his French predecessors refer to the structure of the Renaissance precursors, but are essentially different. Serifs represent the contrast-sort, contrast-low and contrast, and Bodoni’s model is outside that of the broad-nib based archetypal model. In the image below the horizontal serif would be an extrapolation from top left.
Flexible-pointed pen letters and broad-nib ones share a similar skeleton, but in case of the first the contrast-flow can be applied to any arbitrary skeleton and that is definitely not case with the broad nib. The appliance of the broad nib using a certain nib-angle (not the angle of the pen-holder) in relation to the ‘baseline’, results in a fixation of the contrast-flow; for instance the arches of the Humanistic minuscule are connected with their thinnest part (intersection point) to the stems. This standardization automatically implies that the skeleton itself is the result of the movement made using a certain nib/vector angle.
Nowadays letters are often treated as skeleton forms, on which a certain contrast-flow is applied. In the article Hans Eduard Meier’s Syntax-Antiqua in Fine Print On Type (London 1989, pp.22–25 [p.22]) Sumner Stone wrote: ‘It seems doubtful that Renaissance scribes thought of their letterforms as anything but organic units, but the abstractions to a skeleton form do capture the essence of the letters […] The concept of an essential linear form is not unknown in the lettering pedagogy of this century. It is mentioned by Edward Johnston in Formal Penmanship, and was used extensively by the Austrian lettering teacher Rudolf Von Larisch and his student Friedrich Neugebauer. Father Catich also used it in his teaching of letterforms.’
The image above shows the appliance of nib angles of respectively fifteen and thirty degrees on a skeleton form. The fifteen-degrees angle results in a shift of the intersection point away from the stem. The resulting cluttering stem-arch connection has a destructive effect on the shape of the counter.
The derived heart lines in the image above show that the shapes of the letters are the result of the applied nib angle and not vice versa. A change of the nib angle while maintaining the same construction results in different heart lines.
William: ‘[…] I'd love to see studies on Sans Serifs with blink rate, as Luckiesh did […]’
If the outcome is that sans-serif typefaces are harder to read, what does this tell us exactly? That sans serifs are in an absolute way less legible, or that for readers who are conditioned with the archetypal model any deviation is a hampering factor? After all, a child’s mind is blank before it is conditioned. In One Second of Reading in Visible Language, Volume VI, Number 4, (1972 pp.291–320 [p.310] Philip B. Gough writes: ‘The Reader converts characters into systematic phonemes; the child must learn to do so. The Reader knows the rules that relate one set of abstract entities to another; the child does not. The Reader is a decoder; the child must become one. […] what is necessary for the child to learn to read is that he be provided with a set of pairs of messages known to be equivalent, one in ciphertext (writing) and one in plaintext (speech). This implies that a child can learn to combine any set of abstract entities, i.e., graphemes, with certain phonemes: it is just a matter of conditioning.
In While You’re Reading (New York 2007, p.93) Gerard Unger suggests that the reason that the letterforms have undergone very little change since Jenson and Griffo, is probably because these had already largely crystallized and were adapted to ‘the ergonomic needs of the readers’. Considering the facts that Jenson’s roman distinctively deviates from the Humanistic minuscule and was developed in a relatively short period, it seems just as plausible that this archetype defined the ergonomic needs for the reader. The reason that Griffo and Garamont copied (and refined) Jenson’s model, could have been an optical preference. The selection could also have been determined by the fact that Jenson’s model was highly standardized and therefore easier to (re)produce.
" After all, a child’s mind is blank before it is conditioned." This is John Locke's idea of the "tabula rasa," the blank slate. I can understand why you think that everything *must* be due to learning if that is your premise. However, it is about 300 years out of date, and has been soundly refuted, in detail, by modern science. We are born with a complex, structured visual apparatus, among other inborn features of our perceptual, language, and thinking capacities, and even individual personalities. The inborn structure is revealed, for example, in the optical illusions shared by human visual perception—and these are in fact important in type design, as you know. Here is some discussion of the history of the idea of the tabula rasa, with a few links to the voluminous refutation of it: http://en.wikipedia.org/wiki/Tabula_rasa#Science
Frank, I was hoping to see a demonstration of what happens to the optimal combination of balanced white space and a regular stem interval in the bodonis. Does it produce a picket-fence effect (too much regularity) there, whereas it doesn't in older seriffed types?
Peter: ‘[…] what happens to the optimal combination of balanced white space and a regular stem interval in the bodonis.’
I’ll keep it short. ;-) IMHO' one can’t separate the rhythmical pattern from the shape and function of the broad-nibbed-related Renaissance serifs. That’s what I tried to explain above. If one removes the weight from the horizontal strokes then stuff is less ‘glued’ in horizontal direction.
If one deviates from the original model, one can expect that things will work differently. I think that most optical rules are purely relative to a certain model and that is why for Hebrew, Arabic, Indic, CJK, et cetera, scripts the type designer has to be conditioned accordingly. The underlying patterns of these models are the result of evolution, changes in taste, and (the moments in history of) technical innovations. What is considered to be harmonic, rhythmic, and esthetical in type is merely the result of conditioning, i.e., cultural habituation, of their creators, i.e., type designers, their appliers, i.e., typographers, and their users, i.e., readers. If harmony and rhythm would be absolute matters, there wouldn’t be so many differences between scripts.
Well, yes and no. In an absolute way it’s possible, because the reduction of weight is in both the serifs as in the arches in the Bodonis. Relative to the archetypal models the effect is different, because the vertical strokes become more isolated, which results in the spotty effect.
"If harmony and rhythm would be absolute matters, there wouldn’t be so many differences between scripts." But there are huge commonalities in different scripts, reflecting the fact that there are some inborn factors involved. This is similar to grammar. There are many languages, with very different structures. But all have subjects and verbs, and words for subordination of one clause to another. What is going on is a set of common inborn constraints, and the evolution of different structures within those constraints.
This commonality across grammars is also similar for type—for rhythm, and boldness, for example. Extended text in any language is immediately recognizable as text, even by those who can't read it. For one thing it is in lines, with space in between. Peter has studied some of the pictures of text with Fourier transforms on them. I don't know for how many scripts this has been done, but I'm sure you will get very similar kinds of quasi-regular patterns. And on stroke weight, for example in Chinese when you have more strokes within the square, they are lightened in stressed styles. There is a strong tendency toward medium weight for text usage in all scripts that I know of.
Do you really want to insist on the tabula rasa? You will be contradicting masses of facts from psychological research. The "picket fence effect" is one good example of optical illusions demonstrating the existence of inborn characteristics of our perceptual apparatus—constraints relevant to type design. Everyone, no matter what script they read, experiences "dazzle" in a pattern of fairly thick, even alternating black and white stripes. There's no physical reason this should be so. Thus we know it's an artifact of our inborn visual processing apparatus.
Here are two sets of fourier transforms done on blocks of text. In the first, the full transforms are on the right, just to the left of the font name. The blocks of text are on the left. The middle image zeroes in on the area just above and just below the horizontal axis of the full transforms.
These transforms record the phasal regularity or alignment of visual information. In most well set types there is a high degree of information at or around the mean distance between the vertical strokes. That is what the horizontal snippets focus on. Characteristically, the phase alignment is narrow, but not exact in roman fonts, and tighter is some fonts than others. Also, there are no strong competing harmonics in the horizontal plane.
With wide spacing this characteristic phasal structure begins to break down. This is shown in the second set.
The phasal structure of the arabic is entirely different. I did some once years ago but lost them in a hard drive crash.
Thanks Peter for showing! It’s quite a pity that you lost the Arabic ones.
William: ‘[…] This commonality across grammars is also similar for type—for rhythm, and boldness, for example.’
Then it must my personal problem that there are many scripts of which I don’t understand one iota of their harmony and rhythm. Sometimes I look at (for me) exotic scripts as a monkey at the guts of a broken watch.
A couple of years ago I invited a Korean musician with a master degree from the Royal Conservatory in The Hague for a talk on harmony and rhythm in type that I gave at Leiden University. She explained (beautifully sung) the actual lack in Korean traditional music of what is considered to be harmony in Western music. The younger South-Korean generations don’t understand anything of this anymore, because they are conditioned with Western music.
Bill and Peter, I find it impossible to communicate to you what I think is so wrong about foisting the scientific conception of “readability” on type culture.
I’ve conducted an experiment in readability (or perhaps I should say legibility), the results of which are quite “if it looks like a duck, walks, quacks, etc.” And yet you insist on critiquing my conclusion because it has not been lab tested, and loading this discussion with scientific terminology.
I know this is the Technique and Theory section, but this is also the Type Drawers site.
Unless I am mistaken, neither of you are scientists, or have conducted the kind of scientific experiments you are keen on. And your experience of type design is extremely limited.
Could we please be a little less theoretical, and a little more empirical? I challenge you to put your knowledge, skills and good taste to practical use in scientific and/or type design experiments, concerning the issues which so interest you.
Design, which by another name is drawing, and consists of it, is the fount and body of painting, sculpture and architecture, and of every other kind of art, and the root of all sciences. Let whoever may have attained to so much as to have the power of drawing know that he holds a great treasure.—Michelangelo
Nick, TypeDrawers advertises itself as “a discussion forum for professionals and enthusiasts in the fields of typeface design, lettering, and typography.” I am a Registered Graphic Designer (RGD) with the Association of Graphic Designers of Ontario and an enthusiast in the fields of typeface design and typography. I don't draw type. I have recently co-written a scientific paper which is now under review.
I read back through my posts to see what may have led you to believe I am trying to foist a scientific conception of readability on type culture. Perhaps it is this passage:
I think that what your tests show is that normal everyday judgments of readability by qualified practitioners […] are insensitive to the differences in readability […] of your variations when even colour […] is kept constant. I’m not sure this is a sufficient basis for saying “readability is the same.”
What I meant by this is that perhaps there are small but real and functionally significant differences in readability in your samples that normal everyday judgments of readability — sensitive as they are — might be insensitive to. I didn't mean to imply that normal everyday judgments of readability by qualified practitioners are insensitive to real differences in readability, or that the differences in readability that typographers and type designers are able to detect aren‘t real.
I tried to point to issues of the efficiency of cortical integration routines in the visual cortex as an example. I'd say these kinds of efficiencies are factors in ease of reading. It’s not the whole story. There are other factors. Factors that have to do with the eye moving along the line efficiently, and easily navigable text organization.
Irregular spacing probably makes cortical integration of visual information in the visual cortex less smooth. Perhaps the irregularities introduced by Neology don't. It could easily be that they don’t. We don't know. It’s not a big deal, at least not in short stretches of reading, but it might be there. It might be nice to know.
You have a broad concept of readability which includes more than what I or Luckiesh might include, but it includes what I or Luckiesh might include. I can compile a few of texts which reveal Luckiesh‘s concept if you think that might be of use.
I think you want to include what Ovink calls atmospheric values. I can’t tell if you want to extend it to include a consideration of the comprehensibility of the reading material as well or not.
Certainly you weren't using it in those more inclusive senses in your opening remarks when you said you couldn’t detect a difference in readability between the three samples. There you expressly say yourself you were focussing on the “immediate decoding of text, which is the concern of readability.” This is precisely what Luckiesh and Moss were looking at. Cortical integration routines play a fundamental role in the immediate decoding of the text.
I can’t force you to go there, and I won’t pursue it further. I’d just rather not be made to feel I’m barking up the wrong tree, raising things that are entirely beside the point for all practical purposes, or doing something destructive to type culture.
Nick: ‘Could we please be a little less theoretical, and a little more empirical?’
I think this is a valid question considering the fact that roman and italic type are not the result of Renaissance legibility research, but the products of craftsmen/artists who invented, organized and executed a complex and intelligent production system that comprised besides the design aspect, the cutting of punches, the striking and justification of matrices, and the casting of type. One can perform research that explains why the resulting patterns work for eyes and brains, and this certainly helps to explain matters such as legibility. But one can also simply state that if the patterns wouldn't have worked as such, they would not have been applied.
It seems to have become popular among type designers to refer more and more to the way the eyes and brains work, and to make reference to for instance publications by Stanislas Dehaene. Again, IMHO this helps to explain why the patterns work, but based on my research I believe that not only optical matters defined the structures of roman and italic type (but if the results would have been optically disappointing, the Renaissance punch cutters would not have applied them, I reckon).
BTW, the nowadays-famous printed books from the Renaissance were also not always considered of a quality equal to handwritten ones. Morison notes that in spite of Jenson’s almost divinely assisted craftsmanship, fine writing nevertheless was elsewhere so highly esteemed that even his printing failed to please many contemporary collectors of books. The bibliophiles of Florence insisted that printing was so inferior to the manuscripts as to be unworthy to their libraries.*
*Stanley Morison, Four centuries of fine printing (London, 1949) p.19
Just a little childish spanner in the works, merely for your amusement:
For emaxlpe, it deson’t mttaer in waht oredr the ltteers in a wrod aepapr, the olny iprmoatnt tihng is taht the frist and lsat ltteer are in the rghit pcale. The rset can be a toatl mses and you can sitll raed it wouthit pobelrm.
Nick, as I said, I find the font mixing you did here interesting, and something that could be taken further in experimental work.
However,, your presumption here in the discussion is to me breathtaking.
Not only do you presume to rule on what scientists can and cannot research fruitfully, but you presume that Peter is not doing scientific research on reading, which he is in fact doing scientific research. You presume that I have already not done type design under the influence of scientific research, which I have. The concept of readability as ease of reading extended text—as opposed to speed and accuracy of deciphering individual letters or words, or speed of reading—comes directly from Luckiesh. I didn't know where the concept had come from when I designed Williams Caslon text, but it was a main guiding principle for my eye and hand. I chose to do the revival because I had found metal Linotype Caslon Old Face unusually readable in Luckiesh's sense.
I'm glad to say that some people have found my fonts very readable. I know because they have told me so, including about the Small size, not yet in retail, but in use in the listings of The New Yorker. I have explained my process at length, including being guided partly by the standard of readability in my essays Reviving Caslon on ilovetypography.com . After that, when I was asked to do an article for Printing History, I decided to look into the history and identified the origin of the concept in the work of Luckiesh, as explored in depth in the article I did with Peter, linked above. My work is empirical to the extent that people can look at the font used in print and judge for themselves how readable it is. It can also be tested more systematically, which I would be happy to see.
Jan, those kinds of scrambling and substitution have been studied, but would probably be more appropriate in another thread.
Frank, I agree with you when you write "not only optical matters defined the structures of roman and italic type (but if the results would have been optically disappointing, the Renaissance punch cutters would not have applied them, I reckon)." To me both the questions of the constraints of visual processing and of the nature and of the archetypal influence of the original printing fonts—which you have done interesting research on—are worthy of study.
Bill, I didn’t make any presumptions. Read what I said. I’m just suggesting more walk less talk. Instead of going on and on about experiments done 75 years ago, why don’t you conduct some lab tests today? You presented your credentials, so I think it’s fair comment to note that you have only published one typeface, a revival of a design made in the 18th century. You talk about art and science getting along together, and yet you pooh-pooh post-modernism, which has been the state of art for several decades. And you disclaim to be an advocate of neuroculture, which just doesn’t make sense. You are well qualified, in my opinion, to become constructively involved in type design experiments involving lab testing. Lots of type designers inform their work with knowledge of readability research, but to the best of my knowledge nobody has incorporated such testing into the font development process, just as type design has not been used in the design of neurological experiments. Why not go there?
I found here a bibliography that was assembled for the purpose of investigating large print typefaces, but in fact contains much of the published literature concerning the "scientific investigation" of readability, including the works of Luckiesh and Tinker. http://www.tiresias.org/index.htm
The organization that published it, Tireisas.org, is concerned with digital accessibility. It offers a series of fonts that it claims will enhance legibility on screen. The fonts were designed by Chris Sharville with John Gill and claim to be the result of extensive scientific research. http://www.tiresias.org/fonts/index.htm
My empirical evaluation of the fonts yielded this result: they stink and do nothing whatsoever to enhance legibility. The entire project appears to be a con.
A meaningless aside: Far more interesting is the character of Tiresias, him(her)self. After hitting a pair of copulating snakes with a stick, Hera turned Tiresias into a woman. Later, after having resumed his masculinity, he stumbled across Hera (some say it was Athena) bathing naked and was blinded by her. I suspect it was then that he became a researcher into matters typographic. _____
Bill, in all of this thread I've been hoping for something to convince me that there's a future to this line of inquiry. What do you believe are the specific elements and conditions of type design and typography that would become the subjects for scientific investigation? And, if you could make a list of them, how would you consider their interactions?
As for me, I think the matter is hugely complicated, with so many subtle variables that it defies a strictly scientific basis for its evaluation. One of the variables is whether or not a person is interested in the reading material that is part of the test, which, in itself, makes any particular result unreproduceable, a central disqualification to any claim to science.
"I think the matter is hugely complicated, with so many subtle variables that it defies a strictly scientific basis for its evaluation." Scott-Martin, all events have many subtle variables. Basic technique in scientific experiments consists of trying to keep fixed all but the factor you want to study—known as "controlling" the other variables.
Apparently this is a difficult concept to grasp, as I keep hearing this objection. In fact experimentally investigating phenomena with many variables, though always challenging, is done routinely and successfully in science. In a famous example, Dr. Semmelweiss proposed that only one variable was responsible for childbirth fever. He thought there was some agent, unseen, that was unknowingly carried from one woman to another by the doctors. He tested this systematically, by changing only one thing in care of the birthing women: having all those who examined a woman who died afterwards clean their hands with chlorinated lime, before going on to care for any other woman. Systematically varying only this factor dramatically reduced the incidence of childbirth fever.
However, nobody believed Semmelweiss. After all there were "so many subtle variables" involved. "At the time, diseases were attributed to many different and unrelated causes," so it was inconceivable that one variable could be responsible, and studied in this way. Semmilweiss was dismissed as a doctor, and regarded a crazy by his peers, and his wife. He started denouncing his fellow doctors as murders for failing to listen to him and heed clear evidence. In response, he was thrown into a mental institution. He was beaten to death shortly afterwards by guards, in 1865. The story is here: http://semmelweis.org/about/dr-semmelweis-biography/
Now Semmilweiss is regarded as a hero of science, who was vindicated by Louis Pasteur's germ theory of disease, and as a pioneer of antiseptic procedures, which have saved the lives of millions.
Next, I'll answer directly about controlling variables in testing readability.
On testing for readability, Luckiesh, who invented the concept of readability as ease of reading extended text, was quite systematic about his controls. For example, in testing for increase in blink rate over time, his set up controlled for distance to the text, lighting levels, content of the text, time period of reading, and so on. When he studied boldness, Linotype set for him text in four different weights of the slab serif Memphis in the same text at the same size, measure and leading. It is true that the bolder text was wider, so this could be faulted, but it could also be tested for independently in other tests—a common procedure, when you can't control a variable completely that you think may be important.
Correct experimental controls was a big issue between Luckiesh and Tinker. As we explain in the article, we think Luckiesh had the better of the argument. However, as we also point out, Luckiesh for convenience used shorter time periods in his later experiments, and we do question whether this might have introduced "confounding factors"—variables that causally affect the result, and haven't been controlled for.
For this, among other reasons we think Luckiesh's work needs to be repeated and extended.
Are there “so many subtle variables” or so may subtle factors and a few variables — readablity, visibility, legibility?
Qua experiment, Neologic controls one variable factor — colour — in typography. I think this approach to studying readability hasn't been tried. I think it's an interesting approach: controlling the factor you want to look at to study the variable — readability — you are interested in.
Comments
Frank, interesting proposition! I only have a glancing familiarity with your work, but do you have a sense of how an archetypical model for the sans-serif — or a sans-serif like Neology (that employs a new concept of style-fusion) should differ?
In Luckiesh's studies, he found that bolder type was consistently more visible than less bold type. —He measured boldness independently from visibility by reflectance from the page, and visibility by the contrast method I mentioned earlier in the thread. This greater visibility I think explains why bold type is suitable as titling: its greater visibility defeats the way crowding hides words in a mass of other words. The bolder type was equally quickly read, but resulted in a higher blink rate over time, indicating lower ease of reading, or "readability" by his definition. So all in all his studies on boldness explain why it is less suitable for text, but more suitable for display.
Again, this is all with reference to the human visual processing apparatus, rather than archetypes of the original designs. Presumably, these same kinds of advantages and disadvantages of bold type would hold for non-latin scripts, for example.
I will try to explain this briefly (although the latter is not my strongest point ;-).
The fitting Jenson applied on his roman type shows an optimal combination of balanced white space and a regular stem interval. Also Griffo’s type for the Hypnerotomachia Poliphili shows the same balance (see below).
The distance between the stems is dictated by the white spaces in the letters, which are all related to each other. The length of the serifs helps to preserve the space between the letters. Or, the other way around, the serifs work as wedges and help to force the letters in the rhythmic system. Jan Tschichold briefly mentions the stem interval in Treasury of alphabets and lettering (Ware, Hertfordshire, 1985 p.34): ‘The old lettering masters followed the rule that all the basic strokes of a word should be spaced at approximately equal distance. This rule is disregarded today; lower case letters are pushed together’.
This optimal rhythmic system basically only works for text sizes, i.e., roughly 16 pica points (Jenson) and smaller*. Because of the lack of serifs, it is impossible to apply such spacing on sans serif typefaces, with exception perhaps of condensed versions, in which case there is not much space anyway.
The figure above shows the serif version of DTL Haarlemmer. The space between the perpendicular letters is optically equal to the space inside the letters. The serifs make it impossible to tighten the spacing more, because they would collide then.
The next figure shows the sans-serif version of DTL Haarlemmer. Although the equilibration of the spacing is obvious, the stem interval of the serif version could not be maintained. The stem interval in sans serifs with proportions related to the archetypes is by definition disturbed.
* The archetypal model was purely meant for these sizes.
To me that’s very telling.
When balanced white space and a regular stem interval combine optimally, there is cohesive equilibrium.
This may be why Bodoni Bold was traditionally (early and mid 20th century) preffered as the text weight.
These types deviate from the archetypal model. Bodoni’s flexible-pointed-pen-based type and that of his French predecessors refer to the structure of the Renaissance precursors, but are essentially different. Serifs represent the contrast-sort, contrast-low and contrast, and Bodoni’s model is outside that of the broad-nib based archetypal model. In the image below the horizontal serif would be an extrapolation from top left.
Flexible-pointed pen letters and broad-nib ones share a similar skeleton, but in case of the first the contrast-flow can be applied to any arbitrary skeleton and that is definitely not case with the broad nib. The appliance of the broad nib using a certain nib-angle (not the angle of the pen-holder) in relation to the ‘baseline’, results in a fixation of the contrast-flow; for instance the arches of the Humanistic minuscule are connected with their thinnest part (intersection point) to the stems. This standardization automatically implies that the skeleton itself is the result of the movement made using a certain nib/vector angle.
Nowadays letters are often treated as skeleton forms, on which a certain contrast-flow is applied. In the article Hans Eduard Meier’s Syntax-Antiqua in Fine Print On Type (London 1989, pp.22–25 [p.22]) Sumner Stone wrote: ‘It seems doubtful that Renaissance scribes thought of their letterforms as anything but organic units, but the abstractions to a skeleton form do capture the essence of the letters […] The concept of an essential linear form is not unknown in the lettering pedagogy of this century. It is mentioned by Edward Johnston in Formal Penmanship, and was used extensively by the Austrian lettering teacher Rudolf Von Larisch and his student Friedrich Neugebauer. Father Catich also used it in his teaching of letterforms.’
The image above shows the appliance of nib angles of respectively fifteen and thirty degrees on a skeleton form. The fifteen-degrees angle results in a shift of the intersection point away from the stem. The resulting cluttering stem-arch connection has a destructive effect on the shape of the counter.
The derived heart lines in the image above show that the shapes of the letters are the result of the applied nib angle and not vice versa. A change of the nib angle while maintaining the same construction results in different heart lines.
William: ‘[…] I'd love to see studies on Sans Serifs with blink rate, as Luckiesh did […]’
If the outcome is that sans-serif typefaces are harder to read, what does this tell us exactly? That sans serifs are in an absolute way less legible, or that for readers who are conditioned with the archetypal model any deviation is a hampering factor? After all, a child’s mind is blank before it is conditioned. In One Second of Reading in Visible Language, Volume VI, Number 4, (1972 pp.291–320 [p.310] Philip B. Gough writes: ‘The Reader converts characters into systematic phonemes; the child must learn to do so. The Reader knows the rules that relate one set of abstract entities to another; the child does not. The Reader is a decoder; the child must become one. […] what is necessary for the child to learn to read is that he be provided with a set of pairs of messages known to be equivalent, one in ciphertext (writing) and one in plaintext (speech). This implies that a child can learn to combine any set of abstract entities, i.e., graphemes, with certain phonemes: it is just a matter of conditioning.
In While You’re Reading (New York 2007, p.93) Gerard Unger suggests that the reason that the letterforms have undergone very little change since Jenson and Griffo, is probably because these had already largely crystallized and were adapted to ‘the ergonomic needs of the readers’. Considering the facts that Jenson’s roman distinctively deviates from the Humanistic minuscule and was developed in a relatively short period, it seems just as plausible that this archetype defined the ergonomic needs for the reader. The reason that Griffo and Garamont copied (and refined) Jenson’s model, could have been an optical preference. The selection could also have been determined by the fact that Jenson’s model was highly standardized and therefore easier to (re)produce.
I’ll keep it short. ;-) IMHO' one can’t separate the rhythmical pattern from the shape and function of the broad-nibbed-related Renaissance serifs. That’s what I tried to explain above. If one removes the weight from the horizontal strokes then stuff is less ‘glued’ in horizontal direction.
If one deviates from the original model, one can expect that things will work differently. I think that most optical rules are purely relative to a certain model and that is why for Hebrew, Arabic, Indic, CJK, et cetera, scripts the type designer has to be conditioned accordingly. The underlying patterns of these models are the result of evolution, changes in taste, and (the moments in history of) technical innovations. What is considered to be harmonic, rhythmic, and esthetical in type is merely the result of conditioning, i.e., cultural habituation, of their creators, i.e., type designers, their appliers, i.e., typographers, and their users, i.e., readers. If harmony and rhythm would be absolute matters, there wouldn’t be so many differences between scripts.
Can an optimal combination of balanced white space and a regular stem interval be achieved in the bodonis, whereas in the sans serif form it can’t?
This commonality across grammars is also similar for type—for rhythm, and boldness, for example. Extended text in any language is immediately recognizable as text, even by those who can't read it. For one thing it is in lines, with space in between. Peter has studied some of the pictures of text with Fourier transforms on them. I don't know for how many scripts this has been done, but I'm sure you will get very similar kinds of quasi-regular patterns. And on stroke weight, for example in Chinese when you have more strokes within the square, they are lightened in stressed styles. There is a strong tendency toward medium weight for text usage in all scripts that I know of.
Do you really want to insist on the tabula rasa? You will be contradicting masses of facts from psychological research. The "picket fence effect" is one good example of optical illusions demonstrating the existence of inborn characteristics of our perceptual apparatus—constraints relevant to type design. Everyone, no matter what script they read, experiences "dazzle" in a pattern of fairly thick, even alternating black and white stripes. There's no physical reason this should be so. Thus we know it's an artifact of our inborn visual processing apparatus.
These transforms record the phasal regularity or alignment of visual information. In most well set types there is a high degree of information at or around the mean distance between the vertical strokes. That is what the horizontal snippets focus on. Characteristically, the phase alignment is narrow, but not exact in roman fonts, and tighter is some fonts than others. Also, there are no strong competing harmonics in the horizontal plane.
With wide spacing this characteristic phasal structure begins to break down. This is shown in the second set.
The phasal structure of the arabic is entirely different. I did some once years ago but lost them in a hard drive crash.
William: ‘[…] This commonality across grammars is also similar for type—for rhythm, and boldness, for example.’
Then it must my personal problem that there are many scripts of which I don’t understand one iota of their harmony and rhythm. Sometimes I look at (for me) exotic scripts as a monkey at the guts of a broken watch.
A couple of years ago I invited a Korean musician with a master degree from the Royal Conservatory in The Hague for a talk on harmony and rhythm in type that I gave at Leiden University. She explained (beautifully sung) the actual lack in Korean traditional music of what is considered to be harmony in Western music. The younger South-Korean generations don’t understand anything of this anymore, because they are conditioned with Western music.
I also did Hebrew, Greek and Chinese. I will put it all together in a script comparison file in the next few days and make it available.
I’ve conducted an experiment in readability (or perhaps I should say legibility), the results of which are quite “if it looks like a duck, walks, quacks, etc.” And yet you insist on critiquing my conclusion because it has not been lab tested, and loading this discussion with scientific terminology.
I know this is the Technique and Theory section, but this is also the Type Drawers site.
Unless I am mistaken, neither of you are scientists, or have conducted the kind of scientific experiments you are keen on. And your experience of type design is extremely limited.
Could we please be a little less theoretical, and a little more empirical? I challenge you to put your knowledge, skills and good taste to practical use in scientific and/or type design experiments, concerning the issues which so interest you.
I read back through my posts to see what may have led you to believe I am trying to foist a scientific conception of readability on type culture. Perhaps it is this passage: What I meant by this is that perhaps there are small but real and functionally significant differences in readability in your samples that normal everyday judgments of readability — sensitive as they are — might be insensitive to. I didn't mean to imply that normal everyday judgments of readability by qualified practitioners are insensitive to real differences in readability, or that the differences in readability that typographers and type designers are able to detect aren‘t real.
I tried to point to issues of the efficiency of cortical integration routines in the visual cortex as an example. I'd say these kinds of efficiencies are factors in ease of reading. It’s not the whole story. There are other factors. Factors that have to do with the eye moving along the line efficiently, and easily navigable text organization.
Irregular spacing probably makes cortical integration of visual information in the visual cortex less smooth. Perhaps the irregularities introduced by Neology don't. It could easily be that they don’t. We don't know. It’s not a big deal, at least not in short stretches of reading, but it might be there. It might be nice to know.
You have a broad concept of readability which includes more than what I or Luckiesh might include, but it includes what I or Luckiesh might include. I can compile a few of texts which reveal Luckiesh‘s concept if you think that might be of use.
I think you want to include what Ovink calls atmospheric values. I can’t tell if you want to extend it to include a consideration of the comprehensibility of the reading material as well or not.
Certainly you weren't using it in those more inclusive senses in your opening remarks when you said you couldn’t detect a difference in readability between the three samples. There you expressly say yourself you were focussing on the “immediate decoding of text, which is the concern of readability.” This is precisely what Luckiesh and Moss were looking at. Cortical integration routines play a fundamental role in the immediate decoding of the text.
I can’t force you to go there, and I won’t pursue it further. I’d just rather not be made to feel I’m barking up the wrong tree, raising things that are entirely beside the point for all practical purposes, or doing something destructive to type culture.
I think this is a valid question considering the fact that roman and italic type are not the result of Renaissance legibility research, but the products of craftsmen/artists who invented, organized and executed a complex and intelligent production system that comprised besides the design aspect, the cutting of punches, the striking and justification of matrices, and the casting of type. One can perform research that explains why the resulting patterns work for eyes and brains, and this certainly helps to explain matters such as legibility. But one can also simply state that if the patterns wouldn't have worked as such, they would not have been applied.
It seems to have become popular among type designers to refer more and more to the way the eyes and brains work, and to make reference to for instance publications by Stanislas Dehaene. Again, IMHO this helps to explain why the patterns work, but based on my research I believe that not only optical matters defined the structures of roman and italic type (but if the results would have been optically disappointing, the Renaissance punch cutters would not have applied them, I reckon).
BTW, the nowadays-famous printed books from the Renaissance were also not always considered of a quality equal to handwritten ones. Morison notes that in spite of Jenson’s almost divinely assisted craftsmanship, fine writing nevertheless was elsewhere so highly esteemed that even his printing failed to please many contemporary collectors of books. The bibliophiles of Florence insisted that printing was so inferior to the manuscripts as to be unworthy to their libraries.*
*Stanley Morison, Four centuries of fine printing (London, 1949) p.19
However,, your presumption here in the discussion is to me breathtaking.
Not only do you presume to rule on what scientists can and cannot research fruitfully, but you presume that Peter is not doing scientific research on reading, which he is in fact doing scientific research. You presume that I have already not done type design under the influence of scientific research, which I have. The concept of readability as ease of reading extended text—as opposed to speed and accuracy of deciphering individual letters or words, or speed of reading—comes directly from Luckiesh. I didn't know where the concept had come from when I designed Williams Caslon text, but it was a main guiding principle for my eye and hand. I chose to do the revival because I had found metal Linotype Caslon Old Face unusually readable in Luckiesh's sense.
I'm glad to say that some people have found my fonts very readable. I know because they have told me so, including about the Small size, not yet in retail, but in use in the listings of The New Yorker. I have explained my process at length, including being guided partly by the standard of readability in my essays Reviving Caslon on ilovetypography.com . After that, when I was asked to do an article for Printing History, I decided to look into the history and identified the origin of the concept in the work of Luckiesh, as explored in depth in the article I did with Peter, linked above. My work is empirical to the extent that people can look at the font used in print and judge for themselves how readable it is. It can also be tested more systematically, which I would be happy to see.
Frank, I agree with you when you write "not only optical matters defined the structures of roman and italic type (but if the results would have been optically disappointing, the Renaissance punch cutters would not have applied them, I reckon)." To me both the questions of the constraints of visual processing and of the nature and of the archetypal influence of the original printing fonts—which you have done interesting research on—are worthy of study.
http://www.tiresias.org/index.htm
The organization that published it, Tireisas.org, is concerned with digital accessibility. It offers a series of fonts that it claims will enhance legibility on screen. The fonts were designed by Chris Sharville with John Gill and claim to be the result of extensive scientific research.
http://www.tiresias.org/fonts/index.htm
My empirical evaluation of the fonts yielded this result: they stink and do nothing whatsoever to enhance legibility. The entire project appears to be a con.
The scathing review of the fonts on the website below gives flesh to my supposition:
http://screenfont.ca/fonts/today/Tiresias/
A meaningless aside: Far more interesting is the character of Tiresias, him(her)self. After hitting a pair of copulating snakes with a stick, Hera turned Tiresias into a woman. Later, after having resumed his masculinity, he stumbled across Hera (some say it was Athena) bathing naked and was blinded by her. I suspect it was then that he became a researcher into matters typographic.
_____
Bill, in all of this thread I've been hoping for something to convince me that there's a future to this line of inquiry. What do you believe are the specific elements and conditions of type design and typography that would become the subjects for scientific investigation? And, if you could make a list of them, how would you consider their interactions?
As for me, I think the matter is hugely complicated, with so many subtle variables that it defies a strictly scientific basis for its evaluation. One of the variables is whether or not a person is interested in the reading material that is part of the test, which, in itself, makes any particular result unreproduceable, a central disqualification to any claim to science.
Apparently this is a difficult concept to grasp, as I keep hearing this objection. In fact experimentally investigating phenomena with many variables, though always challenging, is done routinely and successfully in science. In a famous example, Dr. Semmelweiss proposed that only one variable was responsible for childbirth fever. He thought there was some agent, unseen, that was unknowingly carried from one woman to another by the doctors. He tested this systematically, by changing only one thing in care of the birthing women: having all those who examined a woman who died afterwards clean their hands with chlorinated lime, before going on to care for any other woman. Systematically varying only this factor dramatically reduced the incidence of childbirth fever.
However, nobody believed Semmelweiss. After all there were "so many subtle variables" involved. "At the time, diseases were attributed to many different and unrelated causes," so it was inconceivable that one variable could be responsible, and studied in this way. Semmilweiss was dismissed as a doctor, and regarded a crazy by his peers, and his wife. He started denouncing his fellow doctors as murders for failing to listen to him and heed clear evidence. In response, he was thrown into a mental institution. He was beaten to death shortly afterwards by guards, in 1865. The story is here: http://semmelweis.org/about/dr-semmelweis-biography/
Now Semmilweiss is regarded as a hero of science, who was vindicated by Louis Pasteur's germ theory of disease, and as a pioneer of antiseptic procedures, which have saved the lives of millions.
Next, I'll answer directly about controlling variables in testing readability.
Correct experimental controls was a big issue between Luckiesh and Tinker. As we explain in the article, we think Luckiesh had the better of the argument. However, as we also point out, Luckiesh for convenience used shorter time periods in his later experiments, and we do question whether this might have introduced "confounding factors"—variables that causally affect the result, and haven't been controlled for.
For this, among other reasons we think Luckiesh's work needs to be repeated and extended.
Next, I'll be specific about further research.
Qua experiment, Neologic controls one variable factor — colour — in typography. I think this approach to studying readability hasn't been tried. I think it's an interesting approach: controlling the factor you want to look at to study the variable — readability — you are interested in.