Critique on Hebrew design wanted

I’d like to hear advice from colleagues who are more familiar with the hebrew script than I am.
This is my first (ever) attempt on that script.


  • Hi @Andreas Stötzner, I should preface this by making it clear I'm not particularly familiar with the Hebrew script, so it's possible things I see a potential inconsistencies are either intended or else rendering issues.
    • The shaped right sides of the /Qof and /Lamed are different with the /Qof appearing to terminate above the baseline.
    • The top right junctions of the /He and Chet have different forms
    • The straight vertical strokes terminate at the base in a variety of styles in different glyphs (compare, for example, the /Final Chaf with the /Resh
  • Uwe WaldmannUwe Waldmann Posts: 8
    edited January 11
    Hi @Steve Gardner, I have to start with the same disclaimer as you (I'm neither an expert on the Hebrew script nor a native speaker). That being said, I have the impression that different top right junctions for /He and /Chet are not uncommon. For example, they are different in David and in Narkis Block, and Adi Stern actually recommends to make them different in his MA dissertation.
  • thank you both for your thoughts. – I have (again) looked at several other typefaces with regard to the treatment of those parts which are equal/similar. I think in Hebrew this is a delicate matter, since a lot of details, or a lot of glyphs bear close resemblances. For the sake of consistency one leans towards making matching parts equal, on the other hand, in order to improve legibility, the differentiation of such parts may be desirable.
    As for Lamed vs. Qof, the downward length of the right part is deliberate, perhaps the difference may be more subtle. – I have also applied a number of alterations to other letters; please look at the new specimen.

  • This is a nice start, Andreas. The general impression I get is that the font is somewhat wobbly, with some letters being slightly right-leaning (alef, bet, zayin, tet, kaf, kaf sofit, mem sofit, nun sofit) and others being more straight for even slightly left-leaning. The design has many echoes of Frank-Ruehl (1910), which is an unfortunate resemblance, in my opinion, as that type has some perversely narrowed glyphs (hei, chet). Frank-Ruehl departed from the tradition of earlier metal-cast Hebrew type that was established in the early 16th century and remained more or less stable for over three hundred years. It’s difficult to explain why it became so popular, but I suspect that the gap in Hebrew typefounding in the 1930s-1940s had extended its shelf life long after the expiration date. 

    Because the 16th-century typefounders made their merubah (square) text types to be combined with the elaborate diacritics of Bible and liturgy, almost all set on separate lines above and below, they divided the em square into quarters, so that the alphabetic glyphs occupied 2, 3, and 4 quarters and the diacritics either 1 or 2 quarters. The letter shin was an exception, as it often occupied 5 quarters, but so too did the extended letterforms used for justification (these often went to six or even seven quarters). 

    Since your design is typographic (as opposed to calligraphic) and traditional in style, I think it would be a good idea to revise the design with the classical proportions and see where you want it to go after that. The restriction will allow you to become more confident in the letterforms. It would also be a good idea to start making words. If you do, you’ll see that some of your details will make it very difficult to achieve good spacing. For example, the upper arms of alef (which looks a little lost) and tsadi extend to the right beyond their feet below and this will cause you nothing but trouble.

    Good luck—I look forward to seeing where you go with it!

  • @Scott-Martin Kosofsky, thank you very much for your valuable insights. This is the sort of guidance I hoped for … ;)
    I was aware that the Frank-Rühl still has some kind of normative status for printed Hebrew, despite its more or less outdated characteristics which root in the 19th century. Therefore I also looked closely at some earlier manuscript samples where the script is more vivid and fluent. But I should definitely take earlier print samples more into account, as you recommend.
    Do you have particular works of reference in mind? I should be grateful if you could give me some hints about prints you’d recommend looking at.

    Right here I have one specimen of older lead type, it is from the Imprimerie Nationale (Paris). – What about it?

  • This is one of my Prague fonts, which I mentioned above. It’s an interpretation of types that appeared in several publications of Abraham Cohen, in 1530. The leftward slant was typical of the Ashkenazic script since the Middle Ages. The style is “spikier” than Le Bé’s, which reflects the different tools with which these letters were written by hand. The Sephardic and Middle Eastern scribes used reed pens, whereas the Ashkenazic scribes preferred the quill or stylus. 

    For years, I had wanted to make a font based on the Prague types. When Covid-19 struck, it became urgent—a now-or-never kind of thing, despite extra busy with other work. So I did, in two weights. (The text here, from the Book of Proverbs, is in the lighter version.)

  • John SavardJohn Savard Posts: 899
    Not being a native speaker of Hebrew, the fact that I didn't see anything wrong with the original specimen was not really useful information. But there's something I will say to your second specimen.
    I think it's a mistake to make the currency symbol an exact stylistic match to the alphabet. Instead, ₪, the "new shekel" sign, is typically in a squared-off sans-serif style no matter what the style of the typeface is.

  • … What is the purpose of your type? …

    The intention is to incorporate the basic repertoire of Hebrew in Andron. I do not plan (at least not now) to provide the full programm of vocalisation and cantillation marks; I think for the purpose of special biblical and liturgical editions there is a sufficient number of high-profile fonts available already. And I’m fully aware that a full in-depth realisation requires a lot of further study and technical expertise (which I lack, in this case).
    Even though my goal in this project is rather limited I think it would be a desirable addition to the multi-script Andron suite. So, the target may be described as a) basic Hebrew for scholarly editing in which some Hebrew is required alongside e.g. Latin, Greek, Coptic; b) to cater for modern (simple) Hebrew in multiscript text scenarios. Andron is mainly used by scholars in linguistics and related subjects.

    Thanks for the references to Le Bé and your excellent font samples … it’ll take some time to digest all this!
  • Ahh, I understand. Andron is a great and vast work, so adding a basic Hebrew would make sense. Given Andron’s roots, you might do well to look at the types used by the Soncino family in the late 15th and early 16th centuries, some of which were cut by Francesco Griffo. They tend to be lighter than most of Le Bé's text fonts, making them a better match for Latin texts. I’d also recommend taking a look at Baruch Gorkin’s Venecia fonts, which are based on them, though in a rather original and engaging way.

  • John SavardJohn Savard Posts: 899
    Looking for a Hebrew font cut by Francesco Griffo, I found this page
    which shows a page from the proof of the Polyglot Bible which Aldus attempted.

  • Andreas StötznerAndreas Stötzner Posts: 622
    edited January 20
    Thank you John for the pointer to that great Griffo site, very precious.
    I’ve overhauled the entire alphabet now. I have adjusted the widths of most glyphs to one size, from which follows that a number of letters are wider now. This tendency I also spotted in various printed samples.
    This is the new stage:

  • Again, I should stress that I'm not familiar enough with the Hebrew script to comment on form, but are the differences highlighted intended?

  • … are the differences highlighted intended?
    well, yes and no. These parts should look more or less matching, like stem serifs in Latin. Which implies that there may be tiny differences, but they shouldn’t pop up before the reader’s eye. – Thanks for pointing at that, these details certainly deserve some finish treatment.

  • 1. Fei is a little too wide, even though it is equal to most of the other square letters. The right side will need more sidebearing than, say, bet. Try to keep sidebearings in mind throughout the design. 

    2. Chaf, ditto

    3. Mem Sofit looks a little narrow.

    4. Chaf Sofit and Fei Sofit would be better if they were 3/4 of the square, rather than 4/4.

    5. Gimel and Nun would be better as 3/4 letters in this design, which is a traditional one, rather than Modern (Israeli) Hebrew.

    6. Tet could be slightly narrower.

    7. The vertical stroke in Qof should move left. Although you say you do not intend to include diacritics, there’s no reason to make it difficult to place them.

    8. Read what I wrote above about overhanging arms and spacing.

    9. Bet, Mem, and Alef still lean a bit to the left.

    10, The downstroke angle of Resh, Nun Sofit, Chaf Sofit, and Fei Sofit should be more similar. I would try splitting the difference between Nun Sofit and Fei Sofit and applying the result to all of them (more or less). There's no need to be rigid about it, but you shouldn't veer off too far either.

    You need to make some words at this stage.

    Hope that helps!

  • Regarding what I wrote above about the Mem Sofit appearing narrow and the sidebearings of the square letters, I meant to add that the Mem Sofit, being the only “fully closed” letter with two vertical sides, will be the one that determines the right sidebearings and, therefore, the widths of the other square letters. Keep in mind, though, that this was the practice of the metal typefounding tradition, in which the medium demanded obedience if the letters were to accommodate the diacritics. While it does not have to be followed literally in digital type, it still is the general guide for traditional designs. To do otherwise can look “incorrect.”

  • John SavardJohn Savard Posts: 899
    edited January 21
    I have adjusted the widths of most glyphs to one size, from which follows that a number of letters are wider now. This tendency I also spotted in various printed samples.
    As it happens, from a site preserving information about Monotype matrices, I saw that their version of Hebrew, for whatever reason, had letters in only two widths, full width and one-half width.
    However, I do not know if this is a general convention of Hebrew lettering, or just a characteristic of the one particular typeface (certainly, it made it more feasible to apply points to the letters by mechanical means).
    And, of course, for Biblical writing, in addition to full-width and half-width, there are special wide versions of some characters at a multiple of the base width.
    EDIT: Ah, I see that this was already addressed in the post by Scott-Martin Kozofsky above.
  • Thank you, John Savard, for mentioning metal Monotype Hebrew. You are correct in saying that the composition matrices had only two widths. This system was developed at the Lanston (American) Monotype Co., not by the British Monotype Corp. Ltd., which had tried and failed earlier, though later made Hebrew types using the American system. The guiding light for this effort was a Philadelphia printer, typesetter, and publisher named Maurice Jacobs (1896–1984), whose company become the principal typesetters for “exotic” languages and scripts in the U.S. They claimed to have the capability to edit and compose type in 165 languages. 

    The two-width system for Hebrew was a severe compromise dictated by the capacity of the composition matrix case. The issue is not so different from making monospaced Latin sans serif fonts. And yet it worked well enough for its moment. The Monotype Hebrew types with diacritics could not be used to set the Hebrew Bible, as they lacked the ±30 cantillation marks, but it could be used successfully for setting liturgical texts, which have only the vocalization “vowels.” The character set can be seen below. The glyphs with superior (cholom) and medial dots (dagesh) can also be found in foundry-cast Hebrews.

    For anyone interested in the history of the development of these types, the papers of Maurice Jacobs can be found in the archives of the American Jewish Historical Society (part of the Center for Jewish History), in New York City.

  • Here we go again. – I’m in favour of the possibility to handle equal character widths not that rigidly.
    The down-right part of Tsadi seems tricky to me, if I further extend it outwards it becomes a very dark spot. On the other hand, I think I detected a slight inward tendency of this part in other typefaces.
    This the updated alphabet:
    … and a bit of text (taken from Wikipedia):

  • John HudsonJohn Hudson Posts: 2,250
    The word space is a little too narrow for Hebrew. Consider adding a kern lookup in the hebr script tag that just increases the width of the /space/ glyph.
  • John HudsonJohn Hudson Posts: 2,250
    ע (ayin) is a bit unstable and appears to fall to the right.
  • Much better, Andreas! There are still a few thing you should look at: 

    • Regularizing the vertical stroke angle of nun sofit (should be like chaf sofit.

    • The weight of the vertical stroke of tsadi sofit.

    • The slant of the nun (a little more left).

    • It’s good for the zayin to have a somewhat different angle from the vav, but this one looks confused.

    • I agree with John about the ayin, though I think the word space is nearly correct‚ just a little more. You'll need that kern lookup especially for glyphs that are very open on the left side.

    Obviously, you have some spacing and kerning to deal with. It would be help to make sure your test words are in the correct direction! Overall, though, this should make a useful addition to Andron. I think you've got the height right, which I prefer to see a little higher than the small caps, but well under the caps.

  • John SavardJohn Savard Posts: 899
    edited January 25
    The word space is a little too narrow for Hebrew.

    You do suggest a technical way to deal with this; once you mentioned it, I saw that you were right, but since Andron supports many other languages, it wasn't clear to me that it could be dealt with, except by making the Hebrew characters for a given point size smaller - which obviously has its own problems.
    Also, it occurs to me that a good starting place to answer some of the basic questions about what certain parts of certain letters should look like... would be a work on Hebrew script calligraphy. That would be a resource for determining what each portion of a letter is intended to be, I would think.
  • John HudsonJohn Hudson Posts: 2,250
    but since Andron supports many other languages, it wasn't clear to me that it could be dealt with
    The most robust technical method I have found to vary the wordspace for different scripts is to use a dist or kern feature lookup that adjusts the width of the /space/ glyph in the GPOS stage. I have seen it done also using variant /space/ glyphs in the GSUB stage, but that means adding a bunch of extra glyphs to the font, whereas my method works directly on the generic /space/ glyph.

    I say most robust, but that doesn’t necessarily mean very robust. Software deals with spacing of words in a variety of ways, and may not always roll the space character into glyph runs with adjacent script characters, or may not even paint the /space/ glyph at all.

  • Thank you all for your comments, they are highly appreciated. I’ll see to make the best of it.

Sign In or Register to comment.