Thanks, but I already have the hard copy.

Nice. I once checked it out from the UCLA library (and made some judicious scans :-) but it will certainly be useful to have it in my pocket, forever. :-)

hhp

They may say high resolution, but I doubt it is adequate for full typographic examination, especially of the small size settings.

And “full color” (RGB pixels) is not quite as good as it sounds, because unscreened spot color was used in the document.

Some of the color pages are fairly sharp, but the majority of the pages are just fuzzy, not suitable to study any details.

It's only 400 dpi, so certainly not suitable for close tracing, but the "atmosphere" of the typefaces (to me the main point) should come through.

hhp

@Hrant, have you seen the hi-res yet? If not, download one section.

If you just want to look at old faces, that's one thing; they're not usable as source material.

I'll be taking a closer look later, definitely.

I guess it would depend on what one means by "source material"; since I'm not a "literal revivalist" to me the only point of such specimen books is for tangential inspiration.

hhp

If you don't have the hard copy, you can buy one for $300-350.

You have your choice of a 149MB or a higher-res 999MB PDF version.

If you don't give a rat's ass, or free hi-res digital isn't good enough, do without, fuhgeddaboudit, and QYB.

Sheesh!

There was an earlier version at the Internet Archive, but as that had significant quality problems, this is still very good news.

Steve, thanks for posting this.

Questioning the quality of resolution and color on a typographic forum is not bitching.
400 dpi is not high resolution for type.
I always like to see the impression that metal type makes, and the ink gain, and can never get enough res.
One doesn’t have to be a “literal revivalist” to want that.

Of course, it would be too many megabites now to scan the whole catalog at “super high res”, but if the internet is to present a simulacrum of the real object, some artefacts must be shown, say a few pages, in higher resolution, and perhaps in oblique views revealing depth, or even in 3D.

One day, thanks to Moore’s Law, this sort of zoom in will be possible:

As an aside, scanning a 35 mm color slide on a transparency scanner ($150 scanner, 9600 dpi), you can get down to the film grain. Accordingly, I calculate the 35 mm recorded image as around 15 megapixels.

Hmmm, well it turns out it's not effectively 400dpi (which to me is decent resolution for anything but "forensic type design") because they seem to have used the lowest possible JPEG quality - here's a screengrab for people who don't want to bother downloading the very large files:

It is pretty blurry... Now, this doesn't make it useless*: it still does an acceptable job giving a feeling of what the book -and the era- was about; and people can certainly find plenty of forms and ideas to be inspired by. And really, you can't complain about a no-strings-attached gift.

* BTW Nick, I love looking super close at things like ink gain too, but that's not really what's worthwhile about that ATF specimen book.

hhp

Your screengrab seems to show a highly satisfactory resolution for most purposes.

It is sufficient for type revivals, I would think, to give a specific example.

@Nick, really nice image! I'd love a print of that at 24" x 24"... super awesome texture.

Over the past little while I have irrefutably proven that when it comes to digital photography = cash in the bank ... it's all about quality not quantity. I picked up a cheapo A630 (capable of only 8mp) and a 19teen something manual typewriter all for less than 50 bucks. I then applied all my inbred talents to produce some incredible imagery which I then sold as high end prints!! So far I have turned 8 mp imagery (plus a 5ive dollar typewriter) into just over 2k ...and still counting... :)

When it comes to technology = earned cash ... for the digital artist/craftsman it's all about quality not quantity.

Point being that with the right amount of skill plus your own talent(s) you can realize a handsome return on a minimal investment.

n.

This was posted in case anyone wanted to see what the 1923 ATF Specimen Book and Catalog looked like, out of curiousity or historical interest, without spending hundreds of dollars.

Help me out. A digital copy of a digital original is an exact duplicate. A scan of anything—theoretically—can never be as good as the original. If the hi-res PDF is viewed at the same size as the original book, how much difference is there to someone with normal and unaided vision?

To those who bemoan the lack of resolution, what magnification would you take to the book? 10X, 30X, 100X, 1000X? How big an enlargement is too big, assuming for typographical purposes, and not to sell 24" X 24" prints?

Are we talking perfection and theory here or something practical?

If you view something onscreen at the same size as the original you're at the resolution of the screen, which is way less than the physical version (Retina, shmetina). Unless you're legally blind you will see -or at least feel- a difference. If you're illegally blind go hide in a cabin in Big Bear. ;-)

As I alluded, the usefulness of a scan depends on the user's intentions. This PDF is fine for some things, just not making faithful revivals (which might very well be a feature and not a bug :-) or seeing delicate details of craftsmanship (which is sad).

hhp

Steve, I don’t have a copy of this particular specimen book.
This new online project is of absolutely no use to me.
And yet surely I represent the kind of person that would be interested in an historical type specimen book.
I have been all over the internet looking at scans of old typography, and there is NEVER ENOUGH RES!
I’m not satisfied with admiring page layouts, as a typophile—a connoisseur of type—I want to see those delicate details of craftsmanship (both founding and printing) that Hrant refers to. That’s the sine qua non of letterpress.
I don’t think it’s too much to ask for bigger files—after all, people are downloading gigabytes of movies—and this is surely the trend. So I say to those altruists who scan and post, get with the tour!

In the meantime, I will continue to collect original historical documents, or reference material with ultra-high resolution photographic reproductions printed by gravure, such as The Art & History of Books by Norma Levarie:
http://www.alibris.com/booksearch.detail?invid=10024732733&qwork=425630&...
(This is the 1968 edition which I refer to, I don’t know how good the reproductions are in subsequent editions.)
Online scans have to be better than those in this book, which has the advantage of being printed on paper, the reflective medium of the originals, and thus has that quality of facsimile authenticity which online resources don’t have—unless one can print out a page from a digital scan at comparable resolution, but that’s a tall order.

What would be cool is streaming for PDFs. Wait a second, I've actually seen that I think - like when you can start looking at a PDF before it finishes downloading. Right? So you could have a multi-gigabyte file but it loads in the background, and jumps to loading the requested page (and predictively the pages around it).

hhp

Although more resolution would be useful and nice, this is considerably better than the previous version (which basically had to be downloaded as individual page images, if one could figure out how to do that - click on "all files (HTTPS)").

http://archive.org/details/SpecimenBookAndCatalogue1923

There are already some other specimen books on the Internet Archive (chiefly by Circuitous Root) that are at significantly higher resolutions, but which pose potential difficulties in downloading. (They can, however, be overcome, since nothing prevents you from using a "download manager" program with the Internet Archive, it should be noted.)

Thanks for the answers. And so nice of you to elaborate, Nick.

there is NEVER ENOUGH RES!

I seems as though you want more res than the original designers ever had or were capable of. (Or am I wrong about that?)

If you could have pristine glyphs that were literally ten feet high is that good enough? What would you do with them?

When I first became interested in typography, I intensely scrutinized letterforms. I often disliked idiosyncrasies, I loved certain glyphs in particular fonts, etc. It took me a while to learn about the color and density and kerning and readability and legibility of a typeface.

I’m not satisfied with admiring page layouts, as a typophile—a connoisseur of type—I want to see those delicate details of craftsmanship (both founding and printing) that Hrant refers to. That’s the sine qua non of letterpress.

No argument here. But for the most part, letterpress is gone. And "those delicate details of craftsmanship" (I don't think) attest to color or readability. When you make a font, how big are the glyphs—either as sketches on paper or on screen?

The delicate details of craftsmanship are the key to understanding letterpress.
Firstly there is the work of the punch cutter and the letter founder.
Then there is the work of the press man.
Those are connected.
One has to see how the metal bites into the paper, and how the ink is deposited.
The letters were designed with that in mind.
So how can one understand an historical type design without intimately observing this three-dimensional physical process?

But what proportion of people looking at an ATF specimen book want to "understanding letterpress"?

hhp

"I seems as though you want more res than the original designers ever had or were capable of. (Or am I wrong about that?)"

Quite wrong, yes.

The only thing that will suffice is an exact recreation of the original, something which maps the original perfectly down to the finest details, the perfect imperfections. Something like the concept described in Borges' On Exactitude In Science, itself an elaboration on a concept from Carroll's Sylvie and Bruno Concluded, that is a map of the thing itself on a 1 to 1 scale.

I seems as though you want more res than the original designers ever had or were capable of. (Or am I wrong about that?)

The machining tolerances for Linotype design and manufacture were down to one quarter of 1/1,000th of an inch. Presumably they could measure that precisely as well. By my calculation, that would be equivalent to a resolution of 4,000 ppi.

The type production drawings were done at a scale of approximately 10 inches high. For a 10-pt type, that would be equivalent to zooming in to around 7200%.

Note that the sidebearings here differ by 0.00025 inches. That would be the equivalent of 1.8 em units in a digital font with a 1000 UPM.

Kent, great stuff.
And now I feel good that 2 units is the smallest sidebearing adjustment I make. :-)

hhp

…a map of the thing itself on a 1 to 1 scale…

A stereotype.

**

I imagine that if pixels were put on pistons, then the effect of relief printing could be simulated.

> if pixels were put on pistons

It's possible that in the future something like that might actually be possible. They are already working on screens that can change shape to make clickable buttons. It's not the pixels that are moving, but who knows what might be possible someday.

http://thedroidguy.com/2013/02/a-keyboard-that-rises-up-from-a-flat-surf...

duplicate post deleted

I'm glad Nick can "see those delicate details of craftsmanship (both founding and printing)" that are the "sine qua non of letterpress" by magnifying the hard copy a bazillion times. I did appreciate his perspective.

That may be fine for close examination, but that's not how people read.

It's Kent who spoke to the issue that I was raising. I had no idea that the tolerances were so small and the ratios so large. Thank you so much, Kent! I like it when replies are actually answers.

that's not how people read.

But it is how people make.

hhp

And now a word from our sponsor…

Hey: among the stuff that I donated to the Albert and Shirley Smalls Special Collections Library at the University of Virginia is a duplicate copy of the ATF 1923 specimen book, all beat up to hell (because it was SERIOUSLY used in a working print shop, I suspect). What am I bid? Or, for that matter, it turns out i had THREE copies of the Barnhart Brother & Spindler Catalog #25. I’ve promised one to Luc Devroye; should I donate the third copy to the Library of Congress (they have only one copy, so it’s housed in the Rare Books room), or does someone want to flash a lot of cash my way?

BTW, anyone else sampled Sam Adams’ Wee Heavy Ale? 10% ABV: oy vay!

That may not be how people read, but this is a type specimen, not a novel.

Sure, the layout and display type are adequate, and there is some art historical interest in the faux ads, but a certain amount of the book is text type, and that is rendered meaningless at the so-called high res of the files in question.

I wonder what my resolution is, when I take off my glasses and look at type from four inches, which is what I prefer.

@kentlew:
The machining tolerances for Linotype design and manufacture were down to one quarter of 1/1,000th of an inch. Presumably they could measure that precisely as well. By my calculation, that would be equivalent to a resolution of 4,000 ppi.

Thank you for a meaningful and informative contribution to this discussion.

I still think that someone interested in performing a type revival would have the data he needs from the specimen book as it is provided, even if it will mean a little more work.

Incidentally, some data in Typographical Printing-Surfaces by Legros and Grant suggests that foundry types were usually designed so that their widths would be multiples of 1/4 or 1/8 of a point. This suggests that a higher-res scan at some multiple of 576 dpi would potentially have extra usefulness.

To adjust for the fact that a printer's point is not exactly 1/72 inch, make that 578 dpi, if practical.

Except the sensors in a scanner are not that funny density, so it's moot.

hhp

Well, if one is using a camera, one can control its distance from the source material. As well, while sensors have a fixed spacing, the time intervals between when they are read during scanning can sometimes be controlled flexibly, giving one near-unlimited choice of resolution in one direction.

As it happens, 0.0081734 mm, or 1/3107.6 of an inch, is very nearly exactly 1/43 of a point and 1/46 of a Didot point, making it a nearly ideal standard scanning resolution for both Continental and British/American type specimen books both.

Neglecting, of course, effects due to changes in the size of the paper due to humidity and suchlike...

Hmmm, maybe NASA can control camera distances that precisely... :-) Not to mention that different parts of the original would have to be at exactly the same distance to the lens, so you'd need some kind of special shallow bowl to place the original in, and somehow get it to sit perfectly flat in it...

Plus I'm actually not sure how pervasive L&G's claim is (although it would certainly make more sense for ATF than in general).

hhp

I just thought I would put the information on record in case any scanner designers are reading this forum.

Also, unless the film is curved, flat originals map to flat sensors at constant scale through most forms of optics, as long as correction for curvature of field (and distortion) is present.

"Presumably they could measure that precisely as well"

We had different brass rules for each size master (8, 10, or 12 e.g.), but they all met at 10,000th of an inch.

Rumor has it that these were Matthew Carter’s brass rules. (From the collection of the Museum of Printing in North Andover, MA.)

David, I was actually referring to the ability to measure finished type — using calipers, micrometers, and such.

And I have no idea what the magnification power is of that microscope device that CHG is pictured using above.

@John: I still think that someone interested in performing a type revival would have the data he needs from the specimen book as it is provided, even if it will mean a little more work.

I used a 10× loupe on 10 pt. type when I was working on my Scotch Modern revival.
I needed to look that closely, because I was trying to make a font that would create a facsimile, with today’s high-res offset lithography, of mid 19th century high-res letterpress printing. The amount of press gain is quite different now, and works differently on different parts of the letter than it did back then. I needed to see the variations in hairline and serif thicknesses in the original, closely, and how much of that was in the metal and how much the result of ink gain, in order to understand the structure and function of that type design and thus enable an accurate simulacrum.

One of the things I discovered in the Scotch Modern project was that facsimile revivals, with their antiquarian flavour, are an abstruse taste. Bembo has a broader appeal and usefulness than Poliphilus.

So you are correct John, that there is enough information in these ATF PDFs for “a type revival”, if not for facsimile quality.

Nice detail, Nick. Thanks.

Ink gain as Nick calls it or "squish" as I call it - is what gave letterpress a warmth and variety that digital set lithography does not have. It also had an impact on the weight of some digital versions of typefaces depending on the source. The spread of the letter depended on press pressure, nature of ink etc.