Do you have any citations for this claim? I can only find scholarly sources from the 1980s.
If it is the hardest to read, why is it used in print? What distinguishes print from electronic media that means that justification works in one place but not the other?
I would be particularly interested to read anything recent that considers legibility in print, the advent of high pixel density displays, and the impact of good justification such as that offered by microtype.
I'm slightly biased here: I believe that justification has got a bad reputation due to how badly web browsers (especially Chrome which doesn't even support hyphenation) handle it.
> What distinguishes print from electronic media that means that justification works in one place but not the other?
Browser are pretty bad typesetting engines. Word breaking doesn't work very well, they don't and can't do protrusion or expansion. The low resolution of displays means you microadjustments don't really work, because they at most slightly shift contrast from one anti-aliased edge to the opposing edge.
Justified text works very well if you're able to competently apply all those techniques (e.g. MS Word and similar document editors didn't do any of that for a long time).
Bad justification results in rivers, uneven spacing, broken edges etc.
Justification was used because people thought equal width columns looked better. Columns of text are generally unnessary in the digital world because you can have one infinitely long column.
In Grid Systems in Graphic Design (1981) the author argued against justification because he thought having spaces of different lengths between words created a poor rhythm (I will find the exact quote tomorrow).
But most publications using justification initially didn't use columns, e.g. books.
I do think multi-column layouts are sorely underutilised on the web. Modern websites make a mockery of desktop wide-screen displays where they use about a quarter of the screen and fill the rest with adverts.
You can use justified text, but if and only if you also use hyphenation. This way, you won't have varying widths of whitespace between each word. But hyphenation on the web is... not satisfying.
It is used in print to pack in the most letters into a single line, because the layout is fixed and words can be broken. It was simply economic to do so. Today we don't have that constraint and with tons of displays we can't be sure how the text will look fully justified.
Browsers are indeed bad at it, but we don't need fully justified text in the first place. For me personally, I prefer varying line lengths because I can easily keep track where the previous line was.
I was hoping for rigorous studies rather than opinions as I have plenty of those myself :)
Thank you for taking the time to respond, though. I had not heard the columns-as-line-separators argument before, but I'm not sure I buy it as most books are justified but do not use multiple columns.
> with tons of displays we can't be sure how the text will look fully justified.
This argument could be applied to ragged text, too. Are you saying that you can't be sure how it looks and therefore you can't justify it? The answer then seems to simply be "compute the layout on the client", as is done anyway.
There is none I'm aware of, but there are some quotes though :) I think we have justified text because of historic reason, it doesn't make much use on the web since it's a totally new medium.
> Are you saying that you can't be sure how it looks and therefore you can't justify it?
No, I'm saying it's unpredictable so we should avoid using it, while left aligned text is the default that works best in any circumstance.
”It is used in print to pack in the most letters into a single line”
Huh? I would think that, for every fully justified line there’s a left-justified one with the same content that is, at worst, equally wide, but typically less wide.
I think full justification traditionally is done for aesthetics, possibly with a bit of “look what we can do” added (it takes a better typesetter to set a line fully justified than it does to set a left-justified one with equal-sized spacing, even when both cases use kerning)
