Archive through June 20, 2003

William Berkson's picture

Here is a sample of what I have been admiring. The baselines do really jump around a lot. Is this a linotype thing?

Sample of Caslon - Linotype Caslon Old Style (?)

kentlew's picture

I don't recall if any of this was already mentioned in this
thread, but . . .

There were several Caslons produced by all of the
major foundries. This particular cut traces its lineage in
this country back to the Old Style of Philadelphia
typefounder Laurence Johnson. Legend has it that in
1858 he arranged with Wm. Caslon's successors to
duplicate the Caslon fonts in this country. Accounts
vary about whether he obtained original matrices,
unfinished strikes, or freshly cast fonts from which he
made electrotyped matrices.

Johnson later merged into MacKellar, Smiths & Jordan,
which in turn became part of the American Type
Founders Company. ATF re-released the Caslon fonts
under the name Caslon Oldstyle No. 471. This is the
face that Linotype used as a model for their Caslon Old
Face (note: not Old Style), which is what your sample is
set in.

The uneven baseline and x-height -- look at that
v, w, & x! -- are presumably carried over from the
original, rough hand-cutting by Wm. Caslon himself.
These effects are seen in the ATF foundry type, though
perhaps to a slightly differing degree. If Johnson
created his own mats from either unfinished strikes or
by electrotyping, then some of the unevenness could
have been introduced at that point.

Some of the bounce in this sample, however, may
be unique to this setting somehow. That excitable u
is not to be seen quite so jumpy in Lino samples that
I have.

Note that the Linotype cutting has slightly different
vertical proportions, most notably in the shortened
descenders, to accommodate Lino's standard

-- K.

William Berkson's picture

Very interesting, Kent. What books would you recommend that have samples that you refer to?

Also I am curious, if you know, whether Caslon's letter forms were following the Dutch. Hrant argued above that Fleishman was the one who deliberately introduced such variation for readability. Do you think he has a case?

hrant's picture

I think the Caslon variability is/was an issue of low fidelity, not intention.

BTW, the more significant irregularity in your sample might be the color variance.


William Berkson's picture

>I think the Caslon variability is/was an issue of low fidelity, not intention.
Do you mean that the u x w v being higher in x height was not a choice but lack of precision? I find this hard to believe, as it is so marked.

>BTW, the more significant irregularity in your sample might be the color variance.
Can you explain this further? Do you mean the variability in the width of the strokes, which is quite noticeable? I am not sure how much of this is due to inking variability, and how much the actual type.

hrant's picture

Frankly, I really don't know enough about the history of printing technology to know if the lack of precision must have been intentional. But I'd be surprised if it was. If you look at the Fleishman stuff, the variability seems highly controlled.

As for the color variance, I mean that some letters are notably darker than others. This can be due to a number of things: bad design (assuming it's unintended); wear in type; but not bad inking, I don't think.


kentlew's picture

William, I am looking at the 1923 Manual of Linotype
which has about 20 pages of sample
book layouts, as well as paragraph showings, in various
sizes of Caslon Old Face.

I think the common opinion is that Caslon was
influenced in his designs by Dutch models in general.
I do not know what influence Fleischman's types may
have had. I'm pretty sure that most of Caslon's seminal
types (including the Pica under discussion here) were
shown in his 1734 specimen. (The reproduction in
Updike is reduced, so it's difficult to be certain.) The
Fleischman face that Hrant has been analyzing is
tentatively dated 1738 in the Ensched

bieler's picture

At last we have a real heavy hitter on the list. But, no pic?

serafino's picture

Frankly I think old Jim, oops, I mean Bill has a point. It is rather hard to use a computer when you are dead.

Anyway Lanston decided not to bother the gentleman either. That's why we left the face alone. When you have good source you don't have to think about ink spread do you?

Too bad about the lowercase f in the linotype cutting. Not much they could do about that? Should have bought a Monotype. Otherwise not bad. Certainly there are a lot worse.

I must stick to my opinion that ink spread is not an aspiration, it is a carbuncle on the face of Caslon. I found someone who has a copy of the Caslon Challenge. I could ask if he could loan a copy?

Gerald Giampa

hrant's picture

But Bill has historically stated that the old folks didn't know exactly what they were doing, that we often give them too much credit.

> When you have good source you don't have to think about ink spread do you?

If you're printing photopolymer (or in low fidelity, like very small or on absorbent paper) then you most certainly do, because it'll spread some more!


John Hudson's picture

I have a copy of the 1924 Caslon Co. specimen, which is interesting and may be instructively compared to the original 18th Century specimen. The marketing spiel in the 1924 version makes a big deal about the types being 'cast entirely from matrices produced from the original punches engraved in the early part of the 18th Century'. What is interesting about this claim is that it doesn't note that the alignment, spacing and finishing of the matrices was much refined during the intervening 200 years. I don't have a photo of the 18th C. specimen, or the 19th C. specimen produced by Wm Caslon III -- which I do own --, but here is a close up from the 1924 specimen (14pt, but the same observations can be made about other sizes).

Caslon Old Face 1924, 14pt

If you compare this image to the Linotype Caslon picture provided by William Berkson, above, you will see that the vertical alignment in the Caslon 1924 version is much better. The Linotype version looks like it has emulated the worst aspects of the 18th Century type, presumably in an attempt to be authentic. Personally, I think the 1924 Caslon specimen shows the Old Face types (and also some of the big types and blackletters) at their very best: the result not of the fetishisation of the original, but of 200 years of proprietorial familiarity with a design that the firm was not afraid to improve. The conceit of the specimen pages is also delightful: re-settings of famous English books, e.g. the first folio edition of Shakespeare's plays, showing how much better they would have looked if set in Caslon Old Face.

serafino's picture


I note the alignment is poor in the

serafino's picture


You point to small types. I am in agreement and I am happy to see you on my side wanting optical scaling to be included in Open Type.

On the matter of absorbant papers however. I am not sure which ones you are speaking of. Most print very crisp.

But if you mean papers that tend to "wick" that is different problem. Sometimes a good ink, or doctoring, could over come that for you. If not, a different stock serves better than a bigger aspirin.

My point was we should not be "imitating results" such as the one you mentioned. Too much time could be spent trying to overcome one specific problem. After all there are many different papers, inks not to mention printers. Which ones do we imitate? Which problems do we attempt to solve.

I realize there are some that think a car looks more beautiful after a crash than when it came from the showroom, hopefully there are very few of them.

Also I realize there are font companies specializing in fonts for differenct kinds of printing processes, and papers. David Burlow's for example. That, however, is highly specialized work, and from the reports I have, the fonts are all for one family of paper, "newsprint".

Imagine were we to follow that example for all known papers, with all fonts in the existing libraries. There would be more fonts than locusts in China.

So if we are to imitate, or indulge in prophecy of circumstances we will certainly be busy doing very little creative work. I'd rather tend shovel bull manure.


The showing of your Caslon specimen is over inked, as such, will hide some of the "fetishisation" of which you speak. You are correct that both the Linotype, and Lanston cuttings were based on an "earlier", Caslon "state". Also, importantly "a smaller point size". A smaller point size is a big consideration when cutting punches. The smaller the size, the more emphasized the imperfections when blown up. The new "state" of the Caslon foundery did not eliminate those imperfections other than perhaps re-aligning, tightening, and making more upright the characters.

The bigger actual hot metal pt. size the smoother it "should look" when cut by a hand punchcutter.

So some of the smoothosity is due to working scale.

Gerald Giampa
P.S. Sorry for any errors, there is pressure to free up this machine.

William Berkson's picture

>1924 Caslon specimen shows the Old Face types their very best ...not ..fetishisation of the original.

I am not 'fetishising' the original, just trying to learn from it.

The jumpy base line of the Linotype in my sample is clearly an undesirable mechanical failing. The overinking on your sample, if Giampa is correct (and it also looks that way to my non-expert eye), is also a failing. I like the 'color' of my sample better. Because of the overinking on your sample I can't compare the variation in color of the letters that Hrant pointed to as a main feature of my sample.

The hypothesis I am pursuing is that digital type has gone too far in uniformity - particularly of color and tightness of fit. Caslon is an apt place to test the hypothesis, because it is obviously less uniform than most faces, yet one of the most used and returned to over the centuries.

Is the looser fit of my sample, which Giampa interestingly says is a feature of limitations of the Linotype machine, a good thing for readibility? The answer isn't obvious to me.

I think it is also an interesting question whether cap height being lower than ascender height - a principle I believe of Morrison - is actually sound. Sure, it does result in less uniformity of color to have full height caps. And I can see that in German this would be distracting, as all nouns are in caps. But in English it may be a nice relief and marker to the eye as a beginning of sentences. (Adobe Caslon does preserve the full size caps.)

Overall, I want to get to the library to see more old samples - but if some people can post more that would be great.

hrant's picture

> absorbant papers

I'm talking about newspapers and phone books.

> imitating

I'm no fan of imitating, but:
1) If you're reviving a metal face into digital that is what you're doing. And if your source is an output (which has suffered some gain) then you have to "retract" the gain if your own eventual output will also have gain.
2) Trapping is good practice of its own (in a text face), imitation or not.

> Which problems do we attempt to solve.

Some kind of weighted average. Imperfect, yes, but helpful nonetheless. Design is after all the balance of compromises.

And if your application is specific (like a custom face for a certain newspaper) then you can be much "tighter" in your decisions. This is especially significant when you consider that custom work is where the real money is anyway, not retail.

> there are some that think a car looks more beautiful after a crash

Few do. But smart people think it's still worth driving the car to work instead of staying home. Unless a wheel is missing.


> digital type has gone too far in uniformity

Very true.
And the true cause is cultural.

> Is the looser fit of my sample ... a good thing for readibility?

Well, there's certainly such a thing as a too-tight setting, but in general things have to be (or actually, seem) slightly tight for good readability. But it also depends on point size. Your sample for example is a bit loose (overall) for 9-11 point. But it's also a bit light for under 8 point. As for John's sample, to me it's a good overall tightness for text - but it's a 14 there.

Tall caps: good for scanning, bad for immersion.


serafino's picture


You feel it is best to retract from "gain" if your own eventual output will "also have gain".

Is it not best to retract from "gain" even if there were

hrant's picture

> What could possibly makes the first gain better than the second?

One gain is good because that's what the original -that you're imitating- looked like. Two gains are bad because it's not.

> we are imitating type, not ink, and paper.

But the type that's relevant -the one the reader reads- is inescapably linked to the medium.

> But smashed, no thank you.

Don't underestimate the psychological power of a smashed up car speeding down the road. I'm speaking from experience. :-)


In fact I've been waiting for a good time to contact you in private: I have a question about ATF cutting slips. But please do give me some time to put things together.


John Hudson's picture

William, I wasn't suggesting that you are fetishizing the original Caslon types. I have become uncomfortable with both the Lanston Caslon and Justin Howes' Founders Caslon -- although acknowleding the former's particular usefulness for letterpress printing from polymer plates -- since discovering the refitted, realigned Caslon in the 1924 specimen. I understand the historical interest embodied in fonts based on the 18th Century originals, but the Caslon Co. refit was so much of an improvement that what I would really like to see is a digital revival of the 1924 version. I think this would be a more useful thing to have.

serafino's picture


We imitated (your word) patterns, not type or overinked specimen sheets.

But we used a comparator, a specialized typefounders projector for viewing types. There were sliding cross hairs. The viewing area
did not suffer from parallax. We used the comparator when comparing Lanston's hot metal casting with the original types. We discovered both had imperfections. We decided to leave them in. The do not show when printed in their intended point size.

Ikarus was used for the outlining which is a discussion that I am not willing to go into at the moment other than I am of the opinion neither Font Lab or Fontographer would have worked as well. The Beauty of Ikarus is that the output is identical in size to the artwork. In our case lightly inked brass patterns.

The Caslon challenge should put your mind to rest. Caslon was indistinguishable when printed. You could not tell one from other.
Either could "John Hudson."

Keep in mind the "original Caslon types" were printed with ink, pasted into the layout along with Lanston's digital Caslon. So we were including comparison to the godliness of ink spread.

Printed by a "craftsman" mind you. Myself.

I am not sure I dare ask about your car experience?


1). Are you suggesting a new digital Caslon should be based on the Caslon 1924 re-fitting?

2) Are you suggesting it should be based on the overinked specimen sheet, or based the lead types which could be easily found?

2) Are you suggesting it should be based on "14. pt as you have illustrated?

3) Do you remember the Caslon Challenge?

Gerald Giampa

William Berkson's picture

>the Caslon Challenge
Let me understand: The bulk of the text was
printed from digital direct to film, then
offset, and the other was letterpress from
monotype metal type to paper?

>a digital revival of the 1924 version
How would this be different from Adobe Caslon? When I typed Adobe Caslon over your sample, at the same size, the spacing matched and the letter forms were very close. The main difference seems to be the greater weight due to ink spread. The Linotype Caslon Old Face to my eye has significant differences in letter form from either. The spacing is also looser. I don't know about Langston's Caslon 337 by comparison. I think Founders Caslon 12 is of limited use because it imitates the random clumps of ink of the 18th century, before Baskerville and others used better inks. This will always have a forced antique look.

But using some of the irregularities of the original is to me an open different question.Here is my late Uncle Ben Lieberman writing about Caslon in 'Types of Typefaces' (1968):

"The typeface which Caslon introduced in 1734 was an instant success, and except for one brief period in the 19th century has remained popular ever since, yet it is perhaps the most controversial face in history. Some persons consider it the greatest type ever (they have popularized a motto "When in doubt, use Caslon") and other think it overrated, a collection of mistakes, elusively out of keeping with everything. But - it works, is hightly readable, alive, with warmth and open dignity that has no pretense whatsoever. Caslon is the prime example of a face in which the individual letters are nothing, but the total effect is strong and honest."

From this thread, it looks like the controversy is not about to go away. But I don't see Adobe Caslon used that much. And I suspect that isn't just changing fashion, but that somehow it misses that elusive something that has made Caslon so successful in past times.

hrant's picture

> Founders Caslon 12 is of limited use

On the other hand, the provision of a different -historically accurate- cut for each size is wonderful.

Check out the non-linear (not to mention non-topological) variance in the italic UC "Q" in the 18, 12 and 8:



John Nolan's picture

> Founders Caslon 12 is of limited use

On yet another hand...
Although I agree that the theory behind Founder's Caslon seems shakey, in practice it works quite well.

I've been using the full range for quite a while, and if you keep within the size ranges, the effect is very nice. The style variations as one moves from the 18 to 14 to 12 pt. cuts produce a satisfying page.

I have also sucessfully (to my eye) introduced Lanston's 337 into pages along with Founder's.

My advice: Founder's Caslon: try it, you'll like it,

William Berkson's picture

>introduced Lanston's 337 into pages along with Founder's

Interesting. How do you compare the two?

John Nolan's picture

Justin Howes seem to have "cheated" and introduced quite a few kerns in Founder's, and this, along with the "ink effect" makes it darker than 337.

I've used 337 as a sort of "Caslon Light" at intermediate sizes. Sometimes Founder's can look heavy in the 22 to 30 pt. range.

John Hudson's picture

1) Are you suggesting a new digital Caslon should be based on the Caslon 1924 re-fitting?

I'm saying that I would find such a typeface more immediately and more broadly useful than any of the current digital Caslons. Sadly, Adobe Caslon is the only version that is suited to a wide range of typographic application, but it doesn't look like Caslon, so what's the point?

2) Are you suggesting it should be based on the overinked specimen sheet, or based the lead types which could be easily found?

I would like to have two versions: one for letterpress printing from polymer, and one for offset printing, but NOT because the latter should reproduce overinking or ink gain/smear. Letterpress printing puts a bite into the surface of the page, which somehow makes the type seem more there. Offset printing is very often too light, underinked, and ends up looking grey and unconvincing. In letterpress, the ink is pressed into the fibres of the paper; in offset, very often, the paper actually affects the colour of the type, which is not dense enough and does not penetrate enough. Robert Reid told me he often has the impression that offset printed text could be blown off the page by a gust of wind.

I think types printed offset often benefit from being slightly heavier than that used for letterpress printing. I don't think this is really about ink gain/smear, as some people have suggested, but simply about the way in which the ink and the letterform interract with the paper in the two different technologies.

Note also that I adjusted the contrast on the photo because it was rather dark, so this might affect the impression that the specimen is overinked. Unfortunately, looking at photographs of type is no substitute for looking at the real thing.

3) Are you suggesting it should be based on "14. pt as you have illustrated?

Well, ideally we would have a different font for each size. This is especially true for Caslon, who varied his forms considerably between sizes; much more than, say, Bodoni, who seems to have followed a fairly linear pattern

4) Do you remember the Caslon Challenge?

I do, although not well enough to make a guess :-)

It should be noted that none of my thoughts are intended to suggest any kind of absolute rule or preference. I have no doubt at all that a skilled printer can make even the lightest digital Caslon (probably 337) look good even in offset. I am very conscious, however, that quality book typography and printing is a small proportion of the typesetting and printing that happens in the world, and I would like to have a Caslon that could take some abuse and still come out readable. Apart from offset printing, there are webpress, laser and inkjet systems, etc., and I'm not refering to office and home printers but to large scale commercial printing devices.

hrant's picture

> I don't think this is really about ink gain/smear

If you measure things closely, you'll see that it mostly is. Especially when you consider -or at least believe- that good letterpress doesn't leave a very visible impression.


John Hudson's picture

I'm saying that extra weight that might benefit a type in offset printing does not need to be based on ink gain in letterpress printing. Why should it? Modern types designed for the principal modern printing technologies, not based on historical models and not in any sense revivals, also benefit from being heavier than letterpress types. It would be perverse to approach such design from the perspective of trying to imagine what a letterpress version of such a typeface might look like if printed with a bit too much ink. Since you're so fond of saying, Hrant, that type design should be based on how people read not on how they used to write, you might agree that types should be designed for how people print not how they used to print.

hrant's picture

> extra weight that might benefit a type in offset printing does not need to be based on ink gain in letterpress

Huh? I totally agree with that - in fact it actively shouldn't be based on that (unless it's simulating the old stuff).

What I thought you said was that the reason offset fonts need to be darker than letterpress fonts is not related to gain - and I think it is.


anonymous's picture

BTW, the more significant irregularity in your sample might be the color variance.

anonymous's picture

Can't a fellow rest in peace? Have you young'ns ever even filed a punch? I knew exactly what I was doing.

Look at my vertically stressed 'o'. I had to choose whether to make it as thick as the vertical stems, or the thick of angle stressed letters, or the 9 and 3 O'Clock of the other round letters. I chose the last, because it is most lively. Vagaries, rubbish.

Sorry about the last post. It is hard to use a computer when you are dead.

Syndicate content Syndicate content