Well, since Sabon was designed to work (look) about the same when set with a Linotype linecaster, a Monotype (metal) machine, or hand-set foundry type, I'd say, no, it is not a better Sabon.

Charles — are you saying it's a *worse* Sabon. :)

Or that I am out of line making the comparison at all?

I'm not really saying either -- though in some ways, comparing the two is like comparing Roger Federer with Rod Laver. Neither font is quite in that exalted class perhaps, but you get the idea.

For me, Sabon was one of the few really good fonts available for the Linotron 202 to use for book-length text. It wasn't used so much with the earlier Linotype linecaster technology. To my eye, Janson was the "really good" linecaster font, but Janson with the 202 was quite poor. And for whatever reason, PostScript Sabon doesn't quite succeed.

I haven't used Garamond Premier Pro yet. None of our customers has requested it, and I am not looking forward to fixing all the kerning needed for book work, so haven't specified it in one of my designs. I do wish it has a set of long descenders; the foundry font seems to have even shorter descenders than the older Adobe Garamond.

Why would you need long descenders for book design in particular? I'm just curious if they aid in legibility, or if it's a stylistic decision.

And what kind of kerning fixes for book work are you talking about? This is also an interesting question. Is there something in general that typically needs tightening up?

As to the long descenders -- my belief is they've always been used for book work, even back in the linecaster days. Short descenders were for newspapers & magazines. But there were always more linotype machines sold to newspapers, so the *standard* fonts were for those customers and had short descenders. Makes commercial sense.

Kerning fixes are as much loosening as tightening. They include single to double quotes; kerning for superiors used in note calls, both with themselves and with the punctuation they most often follow (quotedblright, period, comma) and a few letters. I also don't happen to like much foundry kerning with spaces -- where an f+ space has a positive kern, but is a smaller value than the negative kern with space + quotedblleft, so the *sum* is that the quotes beginning a world are closer to that "f" in the preceeding word than to the word where the quote begins. Stuff like that. Better to make a terminal "f" & write a contextual substitution than to kern it with a wordspace. Etc. etc.

Also, single quote marks (raised & turned commas) are often too tight from the foundry (for my eye), and are used for other purposes than signaling a quote or apostrophe (e.g., glottal stop, pharyngeal frictive, etc.). And period & commas are used with number strings as well as sentences, & on & on.

I've rarely claimed that foundry kerning is outright wrong; there are different points of view as to how things should look, even for bookwork -- and bookwork is a small, specialized market, I am told. Our books have a certain look, and that is part of what our customers pay for. I would note that our small shop -- six people -- has turned out more books that got in to the AAUP Book Show than any other shop, from 1991 to date. With that, I can conclude that whatever we're doing, it does no harm.

BTW, we did a book a while back where I didn't do the work to the font & the comp had to do a series of search & replaces. It takes me about 25 hours to rework the roman & italic. It took the comp 4 hours to do the series of S&Rs. That means the seventh time we use a font family I've worked over ahead of time, we're money ahead.

FWIW

Wow. That's interesting. If you happen to get some free time, yeah right, it would be really cool if you could put together either a tutorial about what you're talking about, or just a series of examples that others could work from.

I'm new to kerning and spacing myself, so I want to learn all that I need to make professionally spaced type.

Thanks!

As to the original question about Adobe Premier Pro vs. Adobe Garamond vs. Sabon, I think the difference that is being noted is that Adobe Garamond has been significantly "smoothed down" from the other two. It has less "bite" to it, and it is noticable when set as a complete page.

One of my college newspaper designs used Garamond Pro for body copy, and one time I did switch it out for Premier Pro, just to see if anyone noticed. They did, but not a single person in the room could figure out what was different :)

Charles, have you ever considered collaborating with a foundry on font development?
We may have to compromise over the length of slashes, but please contact me offline if you're interested.

Once upon a time ... I hope I get this right ... Robert Slimbach went to the Plantin Moretus museum and took thousands of photos of the Garamond types (and I'd imagine other good stuff). He then went back and started work on what became Adobe Garamond. Still seen today as an excellent typeface for (from?) its time. Fast forward through until a couple years ago. Technology has advanced. His skills have increased ten-fold. He revisits his research and out pops Garamond Premier Pro. GPP is not a revival of Adobe Garamond. Garamond Premier Pro is what, I believe, Robert originally hoped to do with Adobe Garamond but for whatever reason is, instead, what we have today. Optical sizes for size appropriate usage.

Sabon is a lovely face. But it leaves me cold. It is too crisp and cold. Especially compared to Adobe Garamond and, even more so, Garamond Premier Pro.

Maybe crisp is the term I was looking for? Adobe Garamond certainly wouldn't be described as "crisp."

But being "authentic" doesn't always mean better, and I'm sure that AG is better for some uses that GPP.

I wonder what circumstances would call for one over the other?

I am still getting used to using GPP having used AG for so long. But with new project I don't hesitate to use Garamond Premier Pro because I know all of my (optical) needs are met.

Tyfa*, Sabon, Scala Serif ... probably more ... all of these faces are overly sharp and better for advertising and magazines (when crisp can help with the shout) and letterpress (when they soften up a bit).

*Although I'm totally guilty of using Tyfa for a series of books. But, now when I see it it is just too sharp on the page.

Tyfa*, Sabon, Scala Serif ... probably more ... all of these faces are overly sharp and better for advertising and magazines (when crisp can help with the shout) and letterpress (when they soften up a bit).

I think I know what you mean; I think of it more as "anemic" than "sharp," but probably the same root cause.

Bit of a story. We have set the South Atlantic Quarterly since the early 1980s. It first went from Times Roman (unknown designer) to a redesign by Mary Mendell using Pilgrim. You could get Pilgrim from Linotype for the 202, & I always felt it was Gill's best text font. Linotron 202 Pilgrim needed lots of kerning, but that's a different story.

When we were ending our use of the 202 & moving to PostScript fonts in the early 1990s, there was no PostScript Pilgrim. What to do? Mary redesigned the journal again using Scala, which kept much of the flavor of Pilgrim. This was before printers were accepting PDF files, and as we used TeX, they weren't about to get application files from us. We ran out the journal on repro paper & pasted it up the old-fashioned way.

Finally came PDF files & direct-to-plate printing. All of a sudden, the type looked anemic. It also looked too sharp. I scanned in samples of the printed journal, both repro-negative-plate and direct-to-plate, and compared them side by side. It didn't take too much work to get back the look of the repro-negative-plate, which must have been what Martin Majoor was looking at anyway, since there was no direct to plate printing when he cut the type.

So which is the "right" Scala? Maybe both; it is just a different look depending on different needs. It is impossible for a foundry to anticipate the uses their fonts will be given.

Interestingly, nobody mentioned the fact that Jan Tschichold's Sabon was based on actual Garamond types... Which was new at the time, since most of the available Garamond types, then, were drawn after a specimen which was wrongly attributed to Garamond.

I know this doesn't add anything to the conversation, but I thought it was nice to remind it to anyone who doesn't know it yet, if any.

dr

Charles — these are fascinating comments and, like DanGayle, I feel I'm learning a lot. The idea of creating a special terminal f and making it contextual is a great idea, and would solve a lot of problems with digital setting. Can you mention one of your books, maybe one where you used either Bembo or some Garamond, where you feel that the setting is particularly good? I'd like to have a look.

I also agree with you, here and elsewhere, about the short ascenders and descenders of GPPro. It was, in fact, this that made me make the comparison with PS Sabon. The subjective impression is of a heavy, stumpy font with not much elegance but a lot of readability — a serviceable font. It is being used quite a bit by Oxford University Press at the moment and it looks ok without looking "beautiful".

I'll stick my neck out here: I think that much that has been said about whether a particular font is a historically accurate revival has to do with focussing on too few points of comparison e.g the axis and the characteristics of the serifs. But in the small face sizes used in books these are micro features of the letterform. The macro features of a letterform — the things which just jump out at you from the page — are the x-height, the relation of lowercase to the capitals, and the lengths of the ascenders and descenders, and these are allowed to vary **wildly** between the *historically accurate* revivals of Renaissance fonts. Sabon looks, on the page, completely unlike AGPro — if both are deemed historically accurate then something has gone very wrong with our ideas of historical accuracy.

Thomas Phinney remarked on some thread (sorry, I've forgotten where that was now) that AGPro was the more historically accurate of the two Slimbach Garamond creations. Looking at a page set in both, I agree with him.

best

El.

Sorry, to clarify, where I say "It is being used quite a bit…" I mean Sabon, not GPPro.

Can you mention one of your books, maybe one where you used either Bembo or some Garamond, where you feel that the setting is particularly good

I can't think of a title offhand, but wife recently designed, & her company set a book called An Introduction to Manuscript Studies that used my Bembo. The paper isn't ideal (wasn't her choice), but given the many plates it was a reasonable compromise. Anyway, it uses Bembo with a terminal "f."

If I can remember, I'll try to post a sample showing the sidebearings when I go into work tomorrow.

There are other uses for such a "f," BTW. What I usually do is shorten up the terminal and increase the right sidebearing, which is just what an "f" needs when followed by a vowel with an umlaut. If you finesse it just right, it will work with vowels that carry a grave accent -- except for an accented "i."

Or you can go whole hog & make up a number of "f's" -- one for a terminal, & others for a following vowel carrying an umlaut, grave, etc. occur. I don't write code anywhere near as well as a lot of the people who post on the forums, & that probably limits what I think of.

My terminal "f's" are just that, and are switched only when white space follows, not a period or comma. I don't kern punctuation that tight. But if you do, you just write a class & switch it for that. For the f-grave vowels, I tend to make extra ligatures, but someone clever could probably do it all with fewer glyphs.

Apart from its modified proportions, GPP isn't "historically accurate" because it's too slick. It doesn't parse the irregularities and bounce of letterpress printing. And it's got kerning.

But it's not meant for recreating facsimiles of Renaissance documents, is it?

Poliphilus was better able to mimic the documents from which it was derived than Bembo, but Bembo was created to improve upon mere similitude.

The idea of "historical accuracy" as a feature or benefit of a typeface is just so much marketing BS, unless it isn't ashamed to be a little rough around the edges.

Take your pick, do you want to listen to Beethoven on a fortepiano or a pianoforte?

It has beeen possible for several years, with OpenType contextual alternates, to make fonts which duplicate the inconsistencies of old letterpress printing, but the thinking seems to be that people don't want that kind of Garamond, or that MvB Celestia Antiqua is best, a most obvious distress rather than something that just looks a bit off the pace.

Funnily enough, as Goudy recounted concerning his creation of Garamont in 1919, many of the folk at Lanston Monotype didn't have much faith in his wonky interpretation, but it sold like hot cakes. It's still my favourite Garamond--in magazines from the 1920s.

Ultimately, I don't believe historical accuracy is a meaningful concept, it's all interpretation.

"The task of renovating or recreating a design from old impressions is the most difficult of all. The effect of impressing upon damp paper, of worn type, and of the spread of ink, have to be reckoned with; and great skill is needed if, while removing blurred outlines, the subtleties of the original engraving are not to be lost." --Stanley Morison, A Tally of Type

Yes, "historical accuracy" makes it seem as though you should able to linearly order the approximations to accuracy — and this is obviously completely wrong.

But if one thinks of nearness along several parameters for the features that are to be approximated then you can speak about accuracy along the parameters that you deem most significant. There is nothing wrong with that idea, that I can see. But of course there is then room for endless disagreement as to whether the features deemed most significant really are the most significant, and you might say that such disagreement is subjective.

but wife recently designed, & her company set a book called An Introduction to Manuscript Studies that used my Bembo.

Thanks for that Charles. I can see from the web that the cover of the book is mightily impressive. Can I ask: have you ever done any books of poetry that are worth looking at?

> Sabon looks, on the page, completely unlike AGPro — if both are deemed historically accurate then something has gone very wrong with our ideas of historical accuracy.

Eluard -- I don't believe that Sabon was intended to be "historically accurate." Tschichold likely had some goals of historical similitude, but I'm sure these were quite secondary to all the technological constraints of the brief.

With regard to the vertical proportions of Sabon (vis-a-vis "original" Garamond), designs in the 20th century, whether for foundry metal or for machine composition, conformed to a few standardized alignments. This means that for any given point size, the baseline was set as a specified, standardized location on the body, effectively limiting the proportion of ascent and descent. The relative x-height was still discretionary, but the descender length was essentially fixed.

Prior to the adoption of Standard Alignment, different fonts of the same size set on the same line were not guaranteed of having a common baseline.

-- K.

Surely some foundries bucked the standard and produced types with long descenders?
Snooty book types, I would guess, not bred to mix with oι πολλοί.

I'm told modern Greeks pronounce that "ee poliee," and the ancients "oi poloi."

But anyway, here is a screen shot of the terminal f. If I were really doing it up brown, I'd make a special ligature -- move the grave to the right some, while not moving, or not so much -- the "a". But this passes, I think, and the terminal works as intended.

Synonyms:
Commoners, great unwashed, huddled masses, masses, multitude, plebians, proletariat, rabble, rank and file, riffraff, the common people, the herd, the many, the masses, the working class

New phrase to add to the memory banks. Thanks Nick!

Thomas Phinney remarked on some thread (sorry, I’ve forgotten where that was now) that AGPro was the more historically accurate of the two Slimbach Garamond creations. Looking at a page set in both, I agree with him.

I'm quite certain I would have said the opposite. I don't know whether I used the phrase "historically accurate," but I'll certainly grant that whether or not I said that, the more precise phrase would have been the the letterforms are closer to Garamond's originals. Heck, Rob even based the different optical sizes on specific sizes of Garamond's original types.

I find the forms of GPP more idiosyncratic and less carefully balanced than AGP. (I could write a whole mini-essay on balance in letterforms and how one's ideas on the topic might evolve as one's skills improve....)

Cheers,

T

Yes, do! Don't hold back on secrets from "the inside". A mini-treatise on the Slimbach Way would be nice :)

Nick --

Yes, of course you're right. In foundry type, some designs came to be cast on what was called Art Line, which was a different standard alignment than Standard Line. But a standard nonetheless, and so it does fix the proportion allotted to the descenders.

For Linotype machine casting, they got around it by requiring the long-descender versions to be cast upon a larger line (i.e., leading was mandatory in order to cast a line including these sorts). So, for instance, a 12-point design might be created on 12-pt alignment, but then the descenders would extend beyond the 12-point body. But rather than call these 13-point designs (which would have required a different alignment), they are 12-point fonts which must be cast on 13-point body. A little confusing, yes? Took Dwiggins a couple explanations from Griffith to get how to design for this.

My point to Eluard was that various standards fixed descenders in certain ways and this may account for why ascender/descender proportions might have been "allowed to vary **wildly** between the *historically accurate* revivals of Renaissance fonts."

-- K.

Thanks for the clarification Kent.

Did the relative position of the standard baseline vary with size?
It seems to me that the smaller the type, the lower the line.

I’m quite certain I would have said the opposite. I don’t know whether I used the phrase “historically accurate,” but I’ll certainly grant that whether or not I said that, the more precise phrase would have been the the letterforms are closer to Garamond’s originals.

Dammit, I knew that I'd be forced to find that comment of Thomas's. Here it is, it is from 2007:

"Personally, I don’t feel Garamond Premier renders Adobe Garamond obsolete. The former is a more restrained and modernized interpretation, and the latter is a more directly authentic revival. There’s room for both, and more besides."

So "directly authentic" rather than "historically accurate, but apart from that…

Nick -- The short answer is Yes. I think I have a comparative graph or something somewhere. It might take some digging, but if I can find it I'll try to post something.

-- K.

Oops, I simply reversed "former" and "latter" in that original comment back in 2007. Sorry about that. But trust me, it makes *way* more sense if you flip it around. Sorry I didn't catch it at the time.

T

Nick --

Here is the diagram I was thinking of. This is taken from Letters of Credit, p. 49, where Walter Tracy discusses the pros and cons of the standard. (And as you can see, he took it from a Stephenson & Blake catalog.)

As you can see, the relative placement of the baseline was not a simple proportion. Another part of the concept of Standard Line (also called Point Line) was that, not only would different fonts of the same size base-align when set alongside one another, but different sizes could be aligned easily using leads of incremental points.

D.B. Updike explains this in his discussion of Standard Line on p. 35 of Printing Types:

"The convenience of being able to align different sizes of type by the use of 1-point leads or their multiples is no doubt appreciated by the compositor; but to make this possible, types below 18-point are arranged, so far as 'line' is concerned, in but three groups: in the first are 5 and 6-point types; in the second, 7 to 10-point types; and in the third, 11 to 16-point."

Both Tracy and Updike bemoan the prevailing late-19th-century taste that established the set baseline so that descenders were given such short shrift. Updike again:

"Taking the second group for illustration, the 'descenders' of g, j, p, q, and y can be no longer in a 10-point than in a 7-point face; and as they are none too long in 7-point they become much too short in 10-point."

Later, into the 20th century, as tastes reverted back to longer descenders for book faces, various workarounds to the shortcomings of Standard Line were developed; but the standard remained a technical parameter that influenced the vertical proportions of type up to the photo era.

-- Kent.

Kent — this is very helpful, thank you. It explains a lot.

Kent, that makes a lot more sense when you explain it then it did when I read it in Letters of Credit.

This discussion makes me wonder, though, why digital revivals have kept up the tradition of short descenders. Is it just tradition now, or do the standard settings of apps like Word somehow enforce it?

Apps like MS Word use the font's bounding box to determine default line spacing, but I wouldn't consider that a cause of short descenders, particularly.

Digital revivals of typefaces that were originally designed on standard line tend to inherit those limitations. Revivals of older typefaces don't necessarily.

Looking at the proportions of Adobe Garamond Pro, Garamond Premier Pro Regular and Garamond Premier Pro Display, I see that Rob's designs optimized for 11-12 point would slightly exceed the "standard line" and GPP Display's descenders would exceed the standard line depth by a good 40% or so.

I guess typefaces like GPP would need to have been cut on "art line" instead. That was what they called it, right? Anybody have a handy chart for "art line" like the one Kent posted for "standard line"? I don't see any details on the "art line" in a quick perusal of "Letters of Credit."

Cheers,

T

I've never encountered a similar chart for Art Line, but one could probably be derived from an ATF specimen -- ATF Garamond for instance. I might create one, if I can spare an hour or so. No promises.

-- K.

Is it just tradition now, or do the standard settings of apps like Word somehow enforce it?

If one is discussing revivals, they are by nature conservative, so the designer is unlikely to go out on a limb with unusually long descenders.

The new historicist faces of the 1920s had more extreme proportions, but Bernhard Modern and Koch Antiqua had teeny descenders!

Today, digitally unconstrained, even when designing an original old style face, I haven't made long descenders. Frankly, I find it difficult to make the "g" and "y" look comfortable when low slung. I look at Perpetua, in which those characters are, IMO, a bit too attenuated, and think, long descenders, why bother?

Another question: why make all descenders the same depth? Some types have a "g" or "y" which is nowhere near as low as "p" or "q". Same thing with "f" -- it can be made quite a bit taller than "l", "k", etc.

Sorry it isn't Garamond, which would be more on-topic, but here is what I did for descenders with Bembo. Reworked on the left, foundry of the right -- again, remember this is for text-size setting only, so the glyphs have been re-weighted a bit over the foundry release.

Ah, I found the remake I did for Adobe Garamond, back in the Type 1 version. Compared to AG Pro, which is pretty similar, I believe. The re-worked descenders on the left, foundry on the right, but letter by letter this time. Looking at it, the "g" needs some help. I have no doubt Mr. Slimbach would have done a much better job, which is why I really wish the foundry would include them. FWIW

Thanks Charles — in both cases I prefer your mods, particularly in the second case I think — though that may just be because the weights are more directly comparable. I take it you slotted your mods into the expert set.

Eluard,

No, not an expert set. With type 1 fonts, we used TeX running in a DOS box, and so were not limited by the Windows OS. I made up fonts having all appropriate glyphs in them -- in many ways, our Type 1 fonts looked like current OT fonts, except they were name based. We would write a different encoding vector when we needed other glyphs -- actually, you could, on the fly, make up temporary fonts & have two or three fonts from my database font -- one having small caps, one having tabular old style figures, one having proportional oldstyle figures, etc.

We even had something like OT features, but these were run using a program we called PreTeX on the TeX (ASCII) files. The program was font-specific (really, it had a place where you could plug in an extra set of routines for specific fonts), and so get roman to italic kerning, etc -- things we can't yet get with standard OpenType features. But the downside was that PreTeX put hard codes in the files, so when you had to make an XML file later, there was a lot to strip out. And (our) TeX's color management was a nightmare for anything but grayscale images. Finally, "our" implementation of TeX was the creation of my business partner, Larry Tseng, & both he & I are getting old. We needed to move to a system where he & I weren't absolutely required, so OpenType and InDesign seemed to fit the bill.

But there is nothing in these ideas that can't be implemented in OpenType, and much of it more easily.

Nick — one place where long descenders (and ascenders) look right is in the setting of poetry. In different threads here I posted two images taken from poetry books published in Italy in 1945–6, one fronm Einaudi and the other from Mondadori. They are strikingly elegant and show the art of setting poetry in a way that complements the text (using contextual alternatives for certain double letters that occur frequently in Italian, like rr and cc). The Einaudi used Linotype Estienne and the Mondadori used Monotype Pastonchi. Both are elegant fonts with longer than usual descenders.

I can't think of too many poetry books that have been published in the digital age that look well-designed. Faber books are particularly unsuccessful, I think, but then I don't know that they were very good pre-digital. The best that I've seen have come from the U.S. (Hotel Lautréamont by John Ashbery is one of the best.)

I think the problem of setting poetry well has been much exacerbated by the stumpy digital fonts that are designed with magazines in mind. There is room for alternatives in the market place.

Hi Charles — I feel your pain over the TeX stuff! I must have spent two or three years encoding fonts so that they could be used in LaTeX, with automatic use of alternates, full ligatures, etc. Endless running of fontinstall, endless checking over the script that was to encode them, pouring over the afm files to see where the glyphs are. rerunning for tweaking the kerning, etc. Now, I feel that I'm just too old for that stuff and want to run XeLaTeX on Opentype fonts and have everything nice and simple (even if I miss out on the customizability!)

Thomas -- Here you go. I reconstructed an estimation of Art Line. I assumed that some of the same underlying considerations were in play regarding mixing sizes on a line. But it only worked if I allowed half-point increments.

This is my best guess and may or may not be entirely correct.

I didn't have the same range of samples to work with (I have a proof of several sizes of ATF Garamond l.c. alphabet). In an effort to make the chart roughly comparable to the Point Line diagram, I inserted "blanks" for samples I didn't have (and guesstimated their alignment).

It was only when I went to resize this to match the Point Line post that I realized that the original diagram wasn't actually proportional -- that is, the 24-pt block is not twice as high as the 12-pt, and the 6-pt is more than half the height of the 12-pt, etc. So it's not the accurate comparison I'd hoped for. One would have to reconstruct the Point Line diagram from the numbers to see how the progressions really compare. (Which I'm not inclined to spend time doing ;-)

-- K.

Eluard -- The long descenders in the poetry specimen you showed are not consistent. The "g" for instance is shorter than the "p".

The problem for the type designer is that while it's possible to drop the vertical stem of "p" indefinitely, and still have a round bowl, once you lower the g beyond a certain point it begins to look condensed. Gill tackled this in Perpetua by adding the distance in the bridge of the "spectacles", avoiding condensing the "eyes" but nonetheless, as I said, the proportions still look odd to me.

I'm afraid I can't agree that long descenders are graceful, which is surely the desired epithet for classic-style poetry setting, they look disproportionate to me.

I don't think long descenders are out of the question, but you can't just tack them on, it has to be part of an overall theme.

Oswald Coooper's "Packard" design comes to mind. The style of lettering he originally developed for the Packard ads in the early 'teens had well integrated long extenders. However, when the font was derived in 1915, they were reined it, for reasons of the standard line. You can see how Oz solved the issue of the g in his original lettering, by making it somewhat monocular.

Nick — yes, I take your point about the g and about unevenness in the extenders being no bad thing. The lower part of the g is so massive that it doesn't need to be extended downwards to the same extent as the p and q. The y could also be considered separately, as the diagonal makes evenness undesirable.

I feel that I've learned a lot from this discussion. Thanks.

(ATFs predictions of the longevity of the Packard certainly didn't come true!)