Serif face revived/inspired from early Aldus/Griffo work...

matthew_desmond's picture

It begins without a sample of work. As I haven't started it yet.

I have however, downloaded a full copy of Hypnerotomachia Poliphili, which I plan on basing the revival/inspiration on. This book is seriously beautiful and I think this is as good as any as a starting point.

I can't seem to get through the completion of a serif face of my own design. So, to prove to myself that I can actually do it, I'm beginning from a historical model.

Since the Aldus/Griffo types were poorly "revived" in the early 20th century with Monotype's muddy Poliphilus, I figure there could stand to be a clean copy out there that is not Bembo.

My goals for the face will be as follows (critiques of any part of this are welcomed!):

1. The face should be suitable for book text.
2. The face will follow the work of Griffo as closely as possible from the printed samples available without being too literal. (Are there surviving examples of the actual type used? Are they accessible? What other sources should I be looking into?)
3. The family will include at least the following: Regular, Italic, Bold, Bold Italic, and possibly small caps and lining figures etc.

So, let me know what you think of all of this. Please pardon the thinking out loud.

matthew_desmond's picture

I'm working out some preliminary specs before drawing:

Cap height: 700
X-height: 412
Descenders: 275
Ascent height: 820

I'm thinking the capital thick strokes will be 90.

I'm trying to take an average measurement on many different impressions of each letter to get these specs.

bieler's picture


Don't mean to dissuade you but this is an old road well traveled.

To my knowledge no punches or matrices exist. There are several good collections of the Aldine printed work (UCLA has a great collection) that would provide better information than a download. Even then, however, the problem is, you can't actually trust (in terms of authenticity) what you are seeing. See my post at

I did a four-year project concerning the Aldine materials at UCLA. Quite frankly, I don't think it possible to get a true representation of an Aldine face.


dstype's picture


I think that it could be a very good idea.
For better measurement and details, you may try read the book from
Peter Burnhill, Type Spaces, in-house norms in the typography of Aldus Manutius, Hyphen Press

William Berkson's picture

Matthew, check out this thread and related ones. (Use Typophile's search utility.) I agree with Gerald.

matthew_desmond's picture

I thank you for your comments, Garry.

I agree that it is not possible to perfectly replicate Aldine forms using printed samples. And definitely not by averaging with software, because you're averaging a source form that is not accurate.

My version would be more inspiration than revival in the end. I plan on using the samples I have as a sort of framework to base my own original forms on.

matthew_desmond's picture

top: Original
middle: My letters blurred to replicate letterpress
bottom: My letters

dan_reynolds's picture

This sounds like a great project!

In your last sample above, I think that you should revisit the angle of the strokes of the letters. They seem to be more modern (Bodoni-esque… I know, that is over doing it, but follow along…) than old-style.

For example, the bowl of the P should begin to curve downward before it does in your sample. And the top of the O (12 o'clock) should be much thicker, as should 1 o'clock, while your 10 o'clock should be thinner. In other words, you need to rotate the angle of the stress a bit to the left around the "circle" as a whole (of course an O is not a circle, but I don't know what else to call it).

Also, your serifs appear to be flat-bottomed. I think they should curve, somehow.

I love your L. But could the top serifs, or at least the top right serif, be chunkier?

The pen, the pen! You must remember the pen! (Insert best-possible Yoda voice impression here…)

matthew_desmond's picture

You're right. Before I get too far I should work some more human elements into the serif forms. I think they are too "constructed" looking as they stand now.

I'm not sure about curving the serifs. Maybe I'll do both types and get someone to do a letterpress print test to see which I like better. It would be fun to vary the amount of ink to see how that would affect the overall look.

I should probably focus on the font looking good on a laser printer though, as that's probably how it will get printed most of the time.


matthew_desmond's picture

Here's a PDF of some updates: (77kb)

Changes made include adding variation in the serifs as well as making the serifs slightly curved. I also fixed the contrast in the O to be more oldstyle.

bieler's picture


Very nice. I think, as Dan suggests, the dip in the serifs would help with the authenticity. If that is of concern.

How did you do this?


matthew_desmond's picture


I'm not sure I'm fully understanding your question...

I have been studying the images I found of the book closely.
Once I find a few good examples of a letter, I compare them and work out the proportions. There is definitely some guessing and some of my own flavor going into the design. It helps me a lot to have a model to begin with.

matthew_desmond's picture

OK, this is freaking me out. All of the Xs are backwards. The lowercase "x"s are normal. Could this have been a layout error or what? Very odd.

The question then becomes, should I keep it like this in my version?

dan_reynolds's picture

To answer your questions: Yes, and yes.

matthew_desmond's picture

Here's another sample. I'm not sure about which form is best for the W. I obviously want to use the oldest form that fits in with the rest of the font. I haven't been able to find a timeline of letterforms to compare with. Also not sure about the J. I'm planning on offering a normal X along with the backwards one. Also, I think they have possibly got the Z backwards in the book also. (175k PDF)

hrant's picture

> All of the Xs are backwards.

How very interesting. Could Griffo have so consciously been going against so-called "ductal logic"? That would be awesome, and a testament to his power as a designer.

> Could this have been a layout error or what?

Extremely unlikely to be a compositing error.
And almost as unlikely to be a casting error.

Definitely, absolutely, totally keep it. It would be such a statement. Certain people will sputter in outrage that "it's backwards", at which point the historical precedent (which happens to be the sort of thing they [also] worship) would be a dramatic and much-needed slap in the face.


matthew_desmond's picture

I think it's interesting, because I didn't notice the backwards Xs until I looked at other fonts! It actually doesn't look out of place on the pages of the book at all.

Miss Tiffany's picture

An analogy to always keep in your toolbox. It isn't the separate glyphs that create the beauty of the typeface in use, it is the glyphs in conjunction with one another.

hrant's picture

That sort of thing only looks out of place to people who design
fonts with strict "ductal logic". Quite a small number of people
in the scheme of things.


matthew_desmond's picture

For sure, Tiffany.
Type is a machine and if any parts are broken, the whole machine is useless.

William Berkson's picture

Matthew, I am skeptical that the backwards X won't cause problems. Digital type with offset press is an inherently exact medium, with less tolerance than letter press. I would look at things like SEXY and WAXWORKS to see whether having one diagonal letter different is going to look 'off'.

hrant's picture

Testing is certainly in order, but there's no reason to be skeptical, simply because the conventional contrast distribution is a result of circumstance, not some Righteous Order. In type design Familiarity is routinely over-played by both conservatives and iconoclasts.

The interesting question is why would Griffo do it? The only clue, and arguably a tenuous one, I can think of is that "X" is almost never the first letter of a word, plus the tops of glyphs carry more visual weight, plus the "X" (and notably not the "x", which does maintain conventional contrast distribution) has a lateral spacing "issue". Maybe.


Lex Kominek's picture

But Hrant - what about ductal logic? Chirography? Are you trying to tell us that we should not design type based on a strict set of arbitrary rules but based instead on what is pleasing to the eye?

- Lex

hrant's picture

What is this, a two-man show?! ;-)


matthew_desmond's picture

The backwards X is counter intuitive because it doesn't go along with the rest of the letters being derived from pen strokes.

Maybe Griffo was left handed?

William Berkson's picture

The Z and z commonly have reversed stress, and that works, so, to repeat, I'm not dogmatically saying the reversed X won't work, just that it is wise to test.

You might also check other examples of Griffo's and Aldus's work to check whether the reversed X in your sample is a result of a mistake in casting the type.

hrant's picture

As William points out, in order to look nicer some characters have evolved beyond "ductal logic", and clearly that's not violating anybody's "intuition". My crazy idea is to let all the characters enjoy that!

> Maybe Griffo was left handed?

Interesting idea. If so, it's a good thing he never
ran the risk of being a student of Noordzij!


matthew_desmond's picture

Haha, Hrant.

I completely understand what you guys are saying, though.

This will most likely be my most tested face yet. Since the goal here is for a readable book text, I'll be attempting to get the color and forms to all be harmonious. If anything sticks out I will change it, even if it goes against what the source materials are showing.

matthew_desmond's picture

Hi again.
I would really like to get some more feedback on the actual letterforms and other details of this work in progress. Here is a new specimen page to look at, this time including lowercase: (pdf 230k)

I'd especially like to hear from you, Gerald, as you have been immersed in the Aldine materials.

Features of the final fonts will be:

• Alternate brackets and parens that align with the uppercase in addition to the standard glyphs that align with the lowercase.
• Lots of ligatures (ff, fi, fj, fl, ffi, ffl, ct, st, etc.)
• Ending swash glyphs for "e" "m" "n" etc.
• Small caps included in each font.
• Both lining and old style figures included.

Once I get this regular weight completely locked in, I will then work on the bolds and italics.

I have to say that I am happy with how the font is looking when printed out. It's feeling more heavy duty to me than Bembo, which was my original goal.

hrant's picture

The flipped "X" seems just fine (although -interestingly- it looks too wide).


piccic's picture

Hi Matthew, how are you? May I ask you why did you feel the desire to approach a typeface in such a "revival" spirit?

Could Griffo have so consciously been going against so-called “ductal logic”?
Griffo, as far as I recall, was killed in a brawl, so it could be…
Designing type wasn't such a "rigidly intellectual" activity, back then.

For everyone else: I'm sorry I disappeared for almost two years. It's not been a "defined" decision… I simply gradually stopped to work on letters, I had to reflect about what I was doing. I would like to restart (and see if I can change my job), although I still can't see clearly how.

hrant's picture

Griffo was hanged for the murder of his son-in-law.
A brawl he would have won, thanks to his burin. ;-)

And welcome back!


William Berkson's picture

>Griffo was hanged

The version I read is only that he was accused, and it is not known what happened to him--whether he was convicted and hung or eg. let go on grounds of self-defense. Also there is a version with the son in law and a version with the brother in law.

In general Griffo modified the handwritten ductus for more even color. This was his breakthrough, ever since imitated. Still the letters retain the 'underlying force' of the hand written models, as that was their point of departure.

Incidentally, to me the reversed contrast X looks like a mistake.

hrant's picture

Different authors will confine themselves to different degrees of certainty.
Not many have written about how Griffo died (for reasons of decorum as
well as temporal remoteness) so I agree one shouldn't bet the house on it.
But of those sources who talk about his death I don't recall any that claim
anything besides hanging (for some in-law).

> This was his breakthrough, ever since imitated.

I think you're giving him too much credit.

> the letters retain the ‘underlying force’ of the hand written models

I believe often we see what we want to see. People see ductus in my own
work too. Middendorp even sees it in Legato, irrespective of Bloemsma's
consistent, driven and explicit anti-chirographic stance.

> to me the reversed contrast X looks like a mistake.

Which of course is to be fully expected.

But what would be interesting is how [a person like] you
would come to terms with the fact (or at least probability,
until we hear a better explanation) that Griffo did that...


William Berkson's picture

>giving him too much credit.

Certainly Jenson was an important precurssor, but it was more fully executed with Griffo

>that Griffo did that

I am guessing it was a mistake in casting, so the type was upside down. Checking on the samples from 'Seneca' (with Griffo type) in Peter Burnhill's 'Type Spaces', the Xs don't seem to have reversed contrast. Though it admittedly is hard to tell exactly what is going on because the ink spread is so extreme. Also the fact that the lower case X's in Matt's sample are normal stress would tend to bolster the idea that it was a casting or trimming mistake. It could be part of his general effort to rebalance the color of the alphabet, but as I say I doubt it.

>a person like you

Oh you mean "delightfully open-minded." Thanks, Hrant :)

hrant's picture

> Jenson was an important precurssor

Jenson was Griffo's "father".

> a mistake in casting, so the type was upside down.

For that to be plausible the descender space needs to be equal to the ascender space, otherwise the "X" wouldn't sit on the baseline. Is this the case? Plus the matrix would have to fit in the mould upside down - although I don't know enough to claim that wasn't possible.

And what are the chances that Aldus/Griffo wouldn't have noticed/minded?

> ‘Seneca’ ... the Xs don’t seem to have reversed contrast.

Clearly a mistake in casting. ;-)

BTW, I just meant conservative.


William Berkson's picture


By my lights, I am not a conservative, either in type or politics, but a reformist. Reformism, as formulated by my late teacher Karl Popper, is the idea that we appreciate tradition as a source of strength, but criticise it, and make reforms to change and improve it where we find it lacking.

Thus in type revival I do not believe that literal, 'authentic' revivals are worth the effort. But I do think that making 'contemporary classics' is worth the effort. Here I would include such faces as Slimbach's Garamonds and Carter's Miller.

hrant's picture

OK then, you find less lacking in tradition than average. Same difference.


William Berkson's picture

>less lacking in tradition than average. Same difference.

No, it is not a question of less or more, it is a question of changing tradition critically, rather then knee jerk change or persistence. In some areas I find more wrong with tradition than others do, in some places less.

I find both conservatives and radicals annoying because they know the solution before they hear the problem. Change for a good reason, not because of a general rule.

Like you, I am interested in readability in text type. So I would say, there are good reasons to forget about reviving Bodoni as a text type, but to go for reviving Griffo's types for text.

hrant's picture

To me, you're having a knee-jerk reaction to being called something you
essentially are but don't want to call yourself. Terminological differences
are fine; hiding behind them is not.


William Berkson's picture

>hiding behind them is not.

My views don't fit into your boxes, so how about expanding your boxes?

On the possibility of getting it upside down: My understanding is that punches were struck into a copper strip, then a rectangle was cut around the strike, and this fitted into the adjustable mold. The cutting of the copper matrix was a skilled job, as it would also involve fitting. And it was often done by someone other than the punch maker.

So I'd think it would be pretty easy to make a mistake, especially in these very early days of typefounding.

hrant's picture

Minor point: more often brass than copper, as far as I know.

Of course mistakes happen, but then the ink hits the paper and it shows up, right? So is your position that Aldus and Griffo didn't notice the mistake, or didn't care enough to fix it?

> it was often done by someone other than the punch maker.

Typically the typefounding tasks were split up among different people in the case of large operations. I suspect in the case of the Aldine press Griffo did as much as possible; he most probably was the one striking the punches onto the matrix material, and then why not go ahead and prep the matrix for casting? I can however see the actual casting done by somebody else, for reasons of risk (of burning) and skill-set difference.

Also, they must have had a way of marking orientation; think of the risk of ending up with an upside-down "o" for example. Admittedly, even with such a marking the matrix could possibly have been placed upside down in the mould; but have you done the ascender/descender-space test I mentioned?


hrant's picture

Also, an assumption concerning mistakenly flipping the matrix
is that the struck letter ends up towards the middle of the matrix.
If the Aldine matrices were anything like the one illustrated here _
that doesn't seem to be the case.


hrant's picture

Wait a second: when you rotate an "X" 180 degrees the contrast distribution doesn't change! The only way to get that "X" (outside of a horizontal flip in Photoshop, which I suspect Griffo didn't use...) is to cut the punch that way.


William Berkson's picture

>Wait a second

Whoops. You're absolutely right, Hrant.

As the Seneca face doesn't seem to have it, it would be interesting to find how where exactly he has the mirrored X and where not.

piccic's picture

Why do you (Matt first) see the Xs "backwards"?
The modulation in all the letters (quite mild), the roughness of them, the small letter size, to my eyes they do not seem to set an entirely systematic approach on how and where "apply weight". I don't think when Griffo worked worried about the result in such a way like most people seem to do now. I think he tried to make the whole work the most functional and aesthetically satisfying, and it was enough good.

George Horton's picture

Double post.

George Horton's picture

Hi Matthew,
This is something crying out to be done; I pray you finish it.

I wanted to do pretty much what you are doing at Reading, but I had to quit through illness; I'm sad not to have anything to show for the months spent researching, thinking and working on a variety of averaging procedures, which took me a long way beyond the stuff Bill has kindly pointed you to. Luckily, it seems I you can do the work for me :-)

The tone of what follows is, just for brevity's sake, terribly hectoring; take it with an implied "but if you want to go another way by all means do".

Gerald is right; but if not a set of fonts you can revive single pages of roman and of italic. Nonetheless your interpretation of what you see needs a thorough knowledge of the mature Griffo's work, in the earliest printings from completed, coherent types. That means using the Scriptores Astronomici Veteres rather than the Hypnerotomachia, for which the type had degraded - degraded further, for the lowercase - and been rather poorly set. It means seeing the De Aetna lowercase, which is a jumble of experimental forms in a very clean state unusually well printed, and comparing it with the revised and narrowed 1499 lowercase; you should also look at the 1497(?) roman, broader, more even, more coherent and less interesting than the 1499, which exists in immaculate form well printed. For your italic, you should see and consider using elements not just, or even mostly, from the well-known 1501 italic in its infinite versions up to 1519, but also the 1503 italic for Soncino and the 1516 made for Griffo's own press, which are generally superior and respectively more beautiful and more functionally intelligent. I've come to see the titchy 1516er as a masterpiece, and it impressed Gerard Unger, who doesn't like the 1501 or 1503. For all these, you need both higher-res scans than you have for the Hypnerotomachia and actual physical books or pages in front of you. I have scans and leaves of the 1501 in its last (?) 1519 state, British Library scans and one leaf of the 1499 S.A.V., BL scans of the 1503 and the 1516, and photographs of the crappy early Jensonian roman - it's interesting to see where Griffo started from - and the 1497 roman. I will send JPEGs if you like, and in extremis could post you my own pages. Just a glance at these will show that your roman isn't very accurate, is more regular than the original, and follows norms established for new work only in the twentieth century. Again, if that's fine by you, that's fine full stop. What you've produced so far is absolutely lovely.

You can get much closer, though others will disagree, with averaging, because it retains information that the eye examining a page cannot possibly synthesize. Here is the output of averaging a page of 1503 italic, using my software (I can send it, but Raph has his own which is much less labour-intensive though perhaps - I can't be at all sure - a shade less accurate). The only processing it has had is increasing contrast to full range and giving it a white background.

Unsharp masking - which is crucial - shows clearly that there are two 'c's here (but not, for instance, two 'n's):

Now the 1499 roman, from my own page, unprocessed:

See the slight whitening on the top of 'o', or the baseline serif of 'i'?
Exaggeratedly unsharped and contrast increased for obviousness:

The different texture of, for instance, 'h' vs 'i' comes from having fewer (too few) images of 'h' being averaged.
If you extract only the interiors, you get

Is that metal? Well, since it's produced from the exaggerated version, and the 'n' doesn't clearly join, it must be a pixel or two skinnier than the metal; but it's surprisingly clean for a type often thought of as very rough.

YvesPeeters's picture

Very intresting work Matthew.

Stefan Seifert's picture

Hi Griffo fans!

very interesting work and discussion in progress. I am impressed as I usally am by the expert knowledge that one finds in this forum. Although the fun is reduced a lot after the pages redesign for it seems to me awfully slow now. That's kind of fun killer.
I am personally not a big Griffo fan for I always found Jensons characters more beautiful and even in the text. What was seen in Aldus printings as a progress i.e. the narrower character ruined the genious idea of having circles as basic idea for beauty. And not the circle of inner forms or else but as a whole as drawn from a pen. Jensons letters to my eyes are the more important ones to take as an example for learning about beauty. I am sorry its provoking from my side, maybe I am only frustrated of not getting response from fantastic guys as Raph, Hrant or William...

Matthew your pdf was impressing. I particularly liked the way you treated the 'h' here your face is soft and pen like as it should be. Maybe you can invest some time in the 'a' if you want. I attach a beautiful redesign from Malin and a rough sketch in Photoshop

best wishes
Stefan (Centaur fan ;-)

Syndicate content Syndicate content