Do you know this one?

Dyson, M.C. and Haselgrove, M. (2001) 'The influence of reading speed and line length on the effectiveness of reading from screen.' International Journal of Human-Computer Studies, 54, 585–612

I don't know whether this answers your specific question, though.

Is there a link to that study, Tim? I'd like to read it.

To David:
Interesting concept. I suspect that line length would be effected by all caps text. A line of Industria all caps of 70 characters is going to be much harder to read than one in Times. I suspect that leading can also extend the potential length of a line ... if you have additional leading equal to the typesize, as opposed to something set solid.

Unfortunately, these are just guesses, and not substantiated by any studies (that I know of).

I always assumed that when people talk about line lengths and text faces what they really mean is a non-condensed typeface in the 8 – 12 point range. But I really think that rules like this are given because people like books that just give them easy to follow rules instead of telling them to actually think about the problem, create many potential solutions, test them, refine, repeat. It’s just like those rules about not using transitional/modern type for text and never mixing sans faces.

David, the old rules for typewritten copy follow this too: if you leave about an inch for the page margins (A4), you will fit 63 pica or 76 elite characters, the rule of thumb used to be 66, I think. Whether this is a pragmatic coincidence or research-based number, I do not know. These rules and the number of lines (30 double spaced) were meant to leave space for editors to write corrections, rather than anything to do with reading.

Is there a link to that study, Tim? I’d like to read it.

Yes, here, on Haselgrove's site.

Michel

@ Boyer: How did you make it so that "here" link downloaded a PDF? Thank you. That's really handy.

(Although next time, you might want to put (PDF) after a link like that. I thought it was a broken link so I just kept clicking. I now have 36 copies of Dyson & Haselgrove ;)

I was taught the optimal linewidth being 10 to 12 WORDS a line for books (lenghty texts). In Dutch the average word is 6-point-some characters. Adding a space after each word (sic) results in 70 to 84 characters a line. Using this rule, linewidths would differ for each language, but I have a hunch that the presence of spaces in a line of text sets the pace for readers, which makes my version of the rule-of-thumb with words instead of characters making more sense. Or not?

. . .
Bert Vanderveen BNO

>I suspect that line length would be effected by all caps text.
Good point. So a recommendation should be for mixed case or all-caps. And I also know size is a factor in some designs between text and display, particularly on the display side of 'text and display' of designs.

Also, I understand that there is a lot of information on text, and I know that 'external' factors like leading and margin are appropriate concerns when taking composition into account. And thanks for that, but I'm mostly interested here in what can be recommended based solely on the design of the font itself. . .

Cheers!

I now have 36 copies of Dyson & Haselgrove

Sorry for that; next time I will say it is a pdf and give its size.

pdf files may be used as links the same way as html files. The format is

<a href="link">some text to click on</a>

For the link itself, I just right clicked on the link to the pdf in Haselgroves' page and pasted it between the double quotes.

Michel

> I now have 36 copies of Dyson & Haselgrove ;)

Good for virtual wallpaper. A trick I use when a link appears to fail is to look at the status line in the browser, which does indicate that the file is a pdf. (Although there would be nothing wrong with a person putting a file identification after the link, if it is not HTML.)

And thanks to Michel for the link. (I felt so lazy afterwards for not just Googling it.)

And thanks to Michel for the link. (I felt so lazy afterwards for not just Googling it.)

Welcome. I did not expect to find it accessible to everyone though (I have access to that journal).

Shortest # 1: Maurice Williams and the Zodiacs, "Stay", 1:37.
Longest #1: Guns and Roses, "November Rain", 8:57.
Average #1 song length: 3:21.

Nick:

did you ever have a chance to be in the same room with Maurice Williams & the Zodiacs when they performed "Stay"? Absolutely riveting. I'll never forget it. When that's in my brain I cannot think about words or characters per line.

powers

I now have 36 copies of Dyson & Haselgrove

That's why you should install PDF Download

Will, the only "Stay" I know was by the Hollies, and I never heard them live.

Shortest # 1: Maurice Williams and the Zodiacs, “Stay”, 1:37.

Does anyone have statistics on books written in Chinese?

Just to chime in back on the line length thoughts...

I've often thought that it makes sense to think of the fundamental measurement unit of type as THE SACCADE instead of fixed measurements like pica. The reason is that the saccade is the fundamental unit of READING.

Academics have pretty much put to rest the debate about letterwise versus wordshape recognition, in favor of the composite "parallel processing" model. But, the saccade is still the primary unit of visual apprehension of type, regardless of how it is visually processed.

Now, obviously, this isn't a precise measuring system in the sense that every reader's saccade is of different physical size, with more practiced and trained readers being able to acquire a larger saccade at higher speeds. But on the other hand, it is a very practical MINDSET since it: (1) orients the text designer to the service of reading rather than abstract & formalist aesthetics, (2) demands visual logic and reader testing to rule type decisions instead of math, (3) accounts for differences in setting various languages due to word lengths, (4) differentiates among text categories (dense hard sciences texts filled with formulas & numbers versus breezy magazine editorials) by bringing intended reading speed into the foreground of criteria.

So to get back to the line length questions, perhaps we shouldn't really ever worry or care about character counts per line as rules-of-thumb, but instead think in terms of how many saccades per line are appropriate for the text and reading experience in question. It's a reductive system, yes, but also quite practical and fast in terms of gut reaction to a text setting. The number of saccades taken per line is always a round whole number (your eye can't jump a fraction of a saccade...it either does or doesn't).

As a side note, I've often thought it would be an interesting experiment to design a typeface based on a standard workhorse that adds a small red dot just above the x-height on each line as "fixation points" for the intended saccades based on reading speed (kind of like a karaoke dot) for training / diagnostic testing purposes to the end of employing this kind of thinking among designers & students....

I think the fundamental unit of measurement should be comprehension. Eye tracking equipment can provide interesting correlational data, but measuring saccades/fixations doesn't really tell us anything about comprehension.

Measuring comprehension is going to fall outside the time and money budgets of most design projects. And comprehension is dependent on many factors outside of design. No amount of typesetting address the limits of readers vocabularies, make readers familiar with niche material, or make bad writing comprehensible. I think that many of the people who are typesetting advanced science, medical, technical and financial material lack the years of study and experience needed to comprehend the text, but it still has to be set by someone who knows how to set type.

>the saccade is still the primary unit of visual apprehension of type

>I think the fundamental unit of measurement should be comprehension.

Thanks for the good avenues of thought.

Since I'm out to help composers, who are trying to satisfy readers, only Alphabet Design is able to effect the rule-of-thumb parameterization of this value.

Cheers!

Are you asking about reading comprehension, or pleasure. And as I say this, I realize the two are not unrelated. That's a theory, of course. Moreover, I believe (more theory) that the content of the text also plays a role in optimum line length, for both ease of comprehension and aesthetics.

At some level, I come back to the notion that if all these things could be quantified, we could come up with a computer program to specify them. While eliminating a need for graphic designers is an intriguing notion -- with it's own pleasure component -- I suspect we're not going to get there.

On thing I have noticed: most texts will set better with some "number of characters" or equivalent value, and that value changes with the text.

For example, when we used TeX for book composition and I was designing, it was easy to change the type size, measure, spaceband values, whatever, and run the whole book. TeX could be set up to write off to a separate file the number of lines that did not fit within the values affecting the horizontal space you'd selected (overfull hbox). Of course, this includes spaceband values as well as measure and type size.

Grep "overfull" in that file, pipe it to another file, and the number of lines in that file would be the number of lines which would not set within the given parameters. The whole deal took less than five minutes; changing a parameter, running the book, grepping the overfull hboxes & taking a look.

Here is the interesting part: Within reason, the values that gave the best mechanical setting (fewest overfull hboxes) were usually the values that made the text the most comfortable to read. Nor was this always the smallest type/longest measure, given the obvious restriction that (1) you just wouldn't go larger or smaller than a certain type size, (or essentially the equivalent, longer/shorter than a certain measure), and (2) you shouldn't change the spaceband values much.

With these restrictions, the technique amounts to tweaking rather than designing per se, but it was interesting how often it proved useful, and for me, showed that the text itself is a factor in choosing certain parameters.

Sadly, AFAIK, you can't set up InDesign to give such a report; in effect, it always has "emergency stretch" turned on.

"grep"?

That's a new term to me, Charles. Explain?

powers

Will, go to the command line and type “man grep”.

Will, the only “Stay” I know was by the Hollies, and I never heard them live.

It's the same song, Nick. The Hollies were doing their cover of the Zodiacs. Jackson Browne also covered it on his Running On Empty album, with David Lindley hilariously doing a line in falsetto. (The liner notes read, "The Zodiacs don't mind.")

You're giving me a hard time, right Will?

grep:

http://en.wikipedia.org/wiki/Grep

Charles said: At some level, I come back to the notion that if all these things could be quantified, we could come up with a computer program to specify them. While eliminating a need for graphic designers is an intriguing notion...

I'm not talking about reading comprehension or pleasure. It's only hard if you try to say the value you give to a single parameter must "design to a specific aesthetic" of a display type, or to "design to a specific comprehension level" as a text type, much less design for a specific aesthetic as a text type, or the most obscure which is to design for a very particular comprehension in display type.

These things are beyond the scope of this, particularly when applied to web type. Because coming networked choices in font selection will be presenting uneducated users unfamiliar choices "we" already know answers to... taking the 'duh' out of some of some of graphic design, is what I'm asking.

But in the mean time, while you all were trying to read, I've done it.

Cheers!

Charles:

Nope: I'm not giving you hard time. I had indeed never heard the term before reading your post. I did look at the wikipedia article moments after I posted the query here. But I wanted you or someone else from the trade also to respond as a check against that source.

But now I now and thanks for expanding my vocabulary by another word. Even though it is one I'm not likely to use very often if ever.

powers

Nick:

http://www.youtube.com/watch?v=o1Z_hskvz1M

But it is nothing like being 18 in 1964 and hearing them at a dance in a hall in North Carolina. They were really on that night.

powers

Rules of thumb are just that – averages or typicals based on common assumptions. Obviously, each font is going be read and comprehended in a unique way. The practiced eye of an experienced and properly trained designer is the only mechanism that should be used to define characters per line in actual practice. Common sense trumps statistical studies every time.

James Tocco
Designer, Detroit, Michigan USA

To what James Tocco wrote, I'd like to add this:

Experience, training and common sense do more than trump statistical studies. They very often confirm statistical studies. As an apprentice compositor I was trained by blue-collar men and women who had never been in the same room with a statistical study of typography. Nor had they ever read any legibility/readability studies. They passed on two things to me: what they had learned from others like them, and what they had absorbed by working on book designs cooked up by masters of that craft. We talked about line length, leading, type size (nominal and visual), word space, and all the other factors noted above. I learned from them how to make simple layouts and set the type in metal. Many years later decided I could try to design books.

In the interim I read some of those reading studies (with my eyelids propped open by toothpicks and good shot of No-Doze). Virtually everything I read in those books I had learned on the shop floor. & had a lot more fun absorbing.

powers

Amusingly, the line length of the Dyson & Haselgrove article itself is, er, somewhat longer than that optimum... Wonder if anyone at the publisher noticed that?
_______________________________________________
Ever since I chose to block pop-ups, my toaster's stopped working.

<< Maurice Williams and the Zodiacs, “Stay”, 1:37. >>

SO:
I've been on hold with the clinic, trying to make an appointment. & just what song is on the Muzak? Of course, "Stay." What the hell else am I supposed to do, not stay? Hang up and call back for more perma-hold? "STAY" indeed.

powers

Display uses and sizes bring other things in to play. The actual text--what it means, and all the stuff around it like images and graphics, can make us stop and ponder. A heading by its nature implies that continuous reading is not going to happen.

My untested theory would be that shorter lines that break the text at meaningful places read better that the usual text length.
Content is more relevant here.

ChrisL

David, I heard Bill Hill, recently of Microsoft, say at TypeCon a few years ago that the reason for the 2 1/2 alphabet maximum rule for single columns is that after that angle, the reader has to move his or her head, instead of just moving the eyes, and more easily loses place when dropping from line to line.

It seems that's Hill's own theory, to me plausible, based on eye physiology. There have been also some studies done at Reading, on line length.

For the fray: Visual angle…

Thanks Bill Bill. I changed the variable to 'text length', because it's much much clearer and I'm thus avoiding discussing it further in the presence of the statistically threatened.

Cheers!