Kerning lower case and font letter spacing

[Melchior]

I have found that not a single digital font is properly spaced and kerned. Not one! I am revising some fonts that I have obtained, and have found it necessary to kern lower case letters.  All capitals should have symmetrical side-bearings, as should most lower-case letters (except f, j, and sometimes y). See the attachment. I have modified Trajanus to my taste. See how cramped the word 'justified' is in the second (unmodified) sample. The t and i almost clash, as do the f and i.  Most of the other letters are too loosely spaced.

Modified Trajanus.pdf

[Ralf Herrmann]

8 hours ago, Melchior said:

The t and i almost clash, as do the f and i. 

Well, preventing that is not necessarily an objective for spacing/kerning. Evening out the space between the letters is the goal. 

And what version did you use exactly? What is labelled “unmodified” in your PDF does indeed look weird — but checking Trajanus on MyFonts I do not get what your PDF is showing. 

download at MyFonts

 

[Melchior]

Ralf:

This holds true for every font I have looked at. The spacing is uneven. Kerning/spacing should give a perfect result, with no gaps or clashes. I used Linotype Trajanus Type 1, purchased from Linotype around 1997. I did it again, showing you the result in the attachment.  I used Open Office instead of Word this time, but results are similar. Note how the c and i in society almost clash, and the t and y do clash, in the unmodified version. The only way to achieve perfect spacing is by making nearly all the sidebars equal and identical. Only f, j and maybe y need be different. Then, all combinations of upper-case and lower case will space perfectly when foot serifs or top serifs are involved (e.g., VW, dl, Ah, dM, nm, lk). Kerning will take care of the others.

 

 

Trajanus-Specimen2.pdf

[Ralf Herrmann]

1 hour ago, Melchior said:

This holds true for every font I have looked at.

In this case, I am concluding you are the single wrong-way driver who complains about the hundreds of cars coming his way from the “wrong” direction. 😉 

Or in other words: If every single font looks wrong to you, then probably you have different understanding of how spacing should be done. And we already nailed down a specific problem, which is even more clear from your latest PDF. You did not improve the f-i spacing — you made it worse. Look at the space between the “i-f” and the “f-i” between the baseline and the x-height. Those two should be as equal as the letterforms permit. Yet, you made the right one huge, and thus, created the uneven spacing yourself. 

It’s not just a little bit uneven. The right area is almost twice as big as the left one!

Even spacing means an even flow of the stems and the spaces between them. We don’t “measure” how close serifs get for example. They can even overlap, as it is not uncommon with letters with diagonals (such as w/v). The clash – especially with copy text – is a smaller problem then the huge gap it would create without the overlap. 

download at MyFonts

And if it gets really bad, then there are ligatures. That’s why an f-i close encounter doesn’t matter much, since it would usually be replaced by an f-i ligature. 

[Melchior]

The 'area' between the letterforms is irrelevant. The distance between the closest parts matters. I read a lot, and poorly spaced letters are common. It's tiring. The combo 'rn' easily looks like 'm'. I know about ligatures, but many fonts were designed without them (Palatino was designed to set well without them.) The Garamond Pro sample you provided looks awful. The spacing should facilitate reading, not impede it. This is not an exercise in geometry, but to facilitate ease of reading. We don't read spaces, but letters. The area under the curve belongs to mathematics, not reading.

[Ralf Herrmann]

27 minutes ago, Melchior said:

The 'area' between the letterforms is irrelevant.

You must be kidding. 

Read any typography or type design book. Take any typography type design class. Ask any respected type designer. You will always get the same answer what even spacing is. And it is the one I gave you. 

27 minutes ago, Melchior said:

 This is not an exercise in geometry, but for ease of reading.

Correct. But you would have to demonstrate how even areas of space impede reading, not just assert it or label it “geometry”. Good luck with that. You are going up against 500 years of type design and probably 99 % of all living type designers. 

[Melchior]

Although I am rarely critical of tradition, that would not surprise me at all.

Ralf, I have read many such books. Some of them are wrong. Robert Bringhurst's book (Elements of Typographic Style) is full of nonsense.

https://medium.com/re-form/a-refutation-of-the-elements-of-typographic-style-3b18c07977f3

The clashing of letters is inexcusable, as are big gaps. Digital typesetting allows us to avoid these sins. There should be an equal space between the 'eastern' and 'western' parts of the letters (serifs) when they are close vertically (e.g., ar, VW, AB, nm, dhinmAmH). In the case of letter combinations where the serifs are not close vertically (e.g., db, dp, du, hu, iu, au, mu, nu) I draw them in a little (-12) and that seems to look better.  I have taken some pieces of text from various web sources and combined them. These have a wide variety of words (Egyptian Pharaohs and scientific names, among others). This helps find unusual combinations for kerning. See attached. It is set in Aldus (modified by me). You should find it illustrates what I am trying to accomplish.

I reject out of hand the notion of 'equal area', as it provides no benefit for reading; and this is what good typesetting should aim for. I am sure we have all seen books set in sans-serif types and been repulsed by this abomination.

Whenever I see something set in an Egyptian type, I know I'm being lied to.

 

 

 

entailing corresponding differences in their skulls.pdf

[Ralf Herrmann]

7 hours ago, Melchior said:

I reject out of hand the notion of 'equal area'

So the discussion can end here. Speaking for 99 % of our field over a course of centuries: Of course it aids reading. But since you will reject that anyway, it would be a waste of time trying to make that case. Your claim “not a single digital font is properly spaced and kerned” was a bold and therefore interesting one and we quickly narrowed down, what the cause for this view is: Your rejection of the equal area principle. This is as far as we can go. 

[Melchior]

Yes, I do reject it. It has no empirical basis. We don't read spaces and compare their areas, consciously or subconsciously. We read letters, especially the tops of letters, as studies have shown.

Regardless, it is a fact that no font is properly spaced with regard letter combinations such as VW, dF, dAhinlmk, which should all be equally spaced. Please look at the attached file, which compares my modified Aldus with Palatino Linotype. Note the combination Se, which is too tight in Palatino. Again, observe in Palatino the sequence sty in the word Dynasty: the t is too close to the y and too far from the s, and the same thing occurs in the sequence its in the word deposits. The letters ao in Pharaoh are too loose, as are the letters de. Bear in mind that Aldus is a little narrower than Palatino Linotype. Lest we forget the italics, please look at the other attachment (Phar2) to see how the letters rw in Underworld are nearly touching in Palatino.

 

Phar.pdf

Phar2.pdf

[Ralf Herrmann]

8 hours ago, Melchior said:

Regardless, it is a fact that no font is properly spaced with regard letter combinations such as VW, dF, dAhinlmk …

No, it is not a “fact”. They are properly spaced in regards to the approach their type designers used. Since you came up with your own approach, that probably no other punch cutter, type designer, or typographer has ever used, of course they will not meet your expectation, but that doesn’t mean that you can define for a fact what proper spacing is or isn’t. That would be incredibly arrogant. 

And we don’t need more PDFs. You want to space letterform distances. We get it. 
And since you reject the reasoning for the current practice of spacing anyway, you can at least learn from this discussion, that equal area spacing is the reason why most fonts behave the way they do. Not sure where else we would go from here. 

[Melchior]

Ralf:

Contrary to your stated belief about the historical standards of spacing, it seems that current digital spacing practice differs from what has gone before. I own many books printed in the past before offset printing and digital fonts came into being. Take a look in the Century Dictionary from around 1889. It may have been composed on a Linotype or Monotype machine, or hand set (more likely). I don't see in these pages the setting features that you claim existed prior to the appearance of digital fonts. I have opened several fonts in FontForge to examine the side-bearings, and have found all manner of differences. There does not seem to be any pattern that I can discern. For instance, the side-bearings for D in Aldus Roman are 29(L) and 37(R), whereas in URW Palladio the figures are 30(L) and 43(R). For t, the figures in Aldus Roman are 26(L) and 26(R) whereas in URW Palladio the figures are 27(L) and 13(R)! I have set the words Mining in Cincinnati in Aldus, my modified Aldus, and URW Palladio, and have observed spacing discrepancies in several combinations (ng, at, and ni) with respect to other combinations. These are too loose. I have no reason to believe that this is deliberate or in any way related to the 'equal area' theory.

http://www.global-language.com/CENTURY/

[Ralf Herrmann]

12 hours ago, Melchior said:

 it seems that current digital spacing practice differs from what has gone before.

Yes, because letterpress fonts have physical limitations that digital fonts do not have. That doesn’t mean that the underlying goals and principles are necessarily different. 

Quote

the 'equal area' theory.

It is not a theory. It is common knowledge in our field, which, as I said before, you can study in books, type design classes or learn about from the vast majority of type designers living today. I said this before and you just shrugged it off with, quite frankly, the silly argument of pointing to a blog post criticizing a single book. First of all, one person’s opinion about a single book doesn’t make that book wrong, nor does it make other books wrong as well. That’s a logical fallacy. It’s like saying global warming isn’t real, because you found a blog post criticizing a certain scientific study. 
Second of all, I did not point you to those sources as a means to demonstrate that this spacing practice is the “right way” to do it. I pointed you to those sources to demonstrate that this is how spacing is understood and practices in our field. So it is actually completely irrelevant whether such books are wrong about certain things or not.

[Kevin Thompson]

Melchior said in an early post that he was using Type 1 (Postscript) format fonts from the mid-90s with Microsoft Word and Apache OpenOffice. Depending on the version of Word and the version of the operating system he’s using, support for such an old type format may be partial or nonexistent with his computer/software setup.

Documentation for both Word and OpenOffice make it clear that support for Type 1 fonts is partial at best, and I suspect that the font metrics have fallen victim to the software.

[Riccardo Sartori]

1 hour ago, Kevin Thompson said:

I suspect that the font metrics have fallen victim to the software.

It doesn’t seem the alleged problem.

20 hours ago, Melchior said:

I have opened several fonts in FontForge to examine the side-bearings, and have found all manner of differences.

 

[Melchior]

10 hours ago, Kevin Thompson said:

Melchior said in an early post that he was using Type 1 (Postscript) format fonts from the mid-90s with Microsoft Word and Apache OpenOffice. Depending on the version of Word and the version of the operating system he’s using, support for such an old type format may be partial or nonexistent with his computer/software setup.

Documentation for both Word and OpenOffice make it clear that support for Type 1 fonts is partial at best, and I suspect that the font metrics have fallen victim to the software.

Not at all. I have compared True-type and Type 1 and Open Type fonts on the same programs, and there is no issue with the font formats as such. I have Word 2002 and Open Office, and both show the same kerning with all three formats. My computer has Windows 10 installed. I set some text in two versions of Palatino (an old Type 1 and the latest Palatino Linotype ttf) in the attached pdf document. As you can see, they are identical. The ttf is on the top, the Type 1 is the on the bottom.

Palatino 1 & 2.pdf

[Melchior]

I have made further refinements to Trajanus. I also increased the 'space' to 290 from 260.

HARPERS BAZAAR HAIRCUT2.pdf