Trapping

gerald_giampa's picture

Kent,

You have put it well.

Do you mind if I insert (AND UNITIZING) and see if it meets with your approval. I am using uppercase to show it is not your words but mine. I am "not" rewriting your sentence, I am just asking if you agree with my insertion?

.....

"But Tschichold had to adapt to both constraints -- duplexing (AND UNITIZING) for Lino, unitizing for Mono. So he made the choice to make the two 'a's fit the 9 units (which yields a noticeably wide roman 'a') instead of cramping the italic in the 8-unit slot."

........

Unless systems in Europe were not equal to America? I refer you to Mike Parker

kentlew's picture

Okay, I think I get it now. I was able to replicate your curve by scaling the l.c. alphabet measurements by the ratio of the type size to your base 48 point. That is, I took the 6-point alphabet and multiplied by 8, so that if the 6-point design was 48 points then the equivalent alphabet length would be 692 points, etc.

ATF Garamond Alphabet Scaled

I won't bother listing the values, since they can easily be derived from my previous data. The gray line shows the curve without the unusual 14 point.

I'm not sure what to make of this curve. And I don't have the math skills to derive a formula for it.

>Are you saying that the Linotype had optical scaling limitations?

What I meant is that when I look at ATF designs and try to compare the quality of optical scaling with that of a Linotype design, there are too many other factors and I'm not sure I can sort out my evaluations to focus just on the one aspect.

>as well as things like how the baseline moves.

You realize, of course, that the position of the baseline on the body for any given size was probably dictated by Standard Alignments. I think that's true. I'll have to look into it. These may have been established without the Bentons' input and may or may not skew your study.

-- Kent.

hrant's picture

That's a nice-looking graph. What SW do you use for that, anyway?

I think one can conclude that:
1) There are indeed two segments.
2) The data points are "well-behaved" (unlike the Monotype Centaur numbers for example). This means the ATF boys were using math, not eyeballing. However, this doesn't preclude the possibility (I'd even say probability) that the boundaries of the equations were determined by eye. I feel that's how a good "engineer" (like Benton) would have done it in fact.

And when you couple these two observations, even the 14 point might fit it. BTW, I wonder where your 14 point (the one you told me looks more like a display cut) falls on the graph.

BTW, what do you get when you apply my "point size normalization" to the Falcon numbers?

> I'm not sure what to make of this curve.

If I were a betting man, I'd put the highest odds on a quadratic shape.

> Standard Alignments ... may or may not skew your study.

That's a very good point - one that I hadn't considered.
There is in fact such a chart in one of the ATF specimen books - I'll scan it up for you guys.

hhp

gerald_giampa's picture

I am aware you are trying to reverse engineer ATF Garamond. If instead, or later are dissecting Monotype there were often two versions of 14, 18, and other sizes normally thought of as display. Often two cuttings. One for display, the other for large composition. Alignment was also an issue in both.

Even Hadriano was cut for 24 pt large comp.

I think I still have in Vancouver a optical devise that measures types in one tenth of one ten thousands of an inch. This devise would prove most useful for these purposes.

The optical devise needs to be kept out of the sun, at least when measuring, and the material must be kept in the same room for a period of time to become adjusted to the same room temperature. There are few devises as accurate as this, and almost none, if any devises like it, specifically made for measuring type. Do you think I should see if it is there?

Gerald Giampa
Lanston Type Company

kentlew's picture

Hrant --

A couple of things. Tracy shows a diagram of Standard Line alignment (also called Point Line) on page 49 of Letters of Credit. ATF Garamond was cast on Art Line, though, which allows for longer descenders; but I believe the same principle is in play.

If you still have your hands on that 1923 ATF book, I discovered that a chart of alphabet lengths is given on pp. 584-585. Wish I'd seen this before manually measuring your sample. Turns out my measurements were only off by a point here or there. The 14 point, BTW, is listed as 149, versus my measurement of yours at 148.

Software? Are you kidding? You'll laugh: I make up these graphs in Quark. Good ol' fashioned brute force, man.

-- K.

kentlew's picture

Okay, here's a comparison of the alphabet lengths for the same range of Falcon and ATF Garamond. These are scaled to 12-point equivalence.

ATF Garamond and Falcon alphabet length compared

Here's the actual data --

Falcon:
6 pt : 180 pts
7 pt : 168
8 pt : 160.5
9 pt : 153.33
10 pt : 148.8
11 pt : 144
12 pt : 141

ATF Garamond:
6 pt : 174 pts
7 pt : 162.8
8 pt : 154.5
10 pt : 141.6
12 pt : 134

Note that there are no data points for ATF Garamond 9 pt and 11 pt. The ATF alphabet lengths are revised from the chart in the specimen book, except 7 point, which isn't listed. I revised that measure up from 94.25 to 95, since my other measures were usually on the order of 1 point short.


Okay, I've had enough math for a while. I got to get back to earning a living -- using type, instead of analyzing it to death, thank goodness.

-- K.

rcapeto's picture

Speaking of optical compensation.... I

hrant's picture

> Tracy shows a diagram of Standard Line alignment

And here's the one from that ATF book:
(39Kb) http://www.themicrofoundry.com/other/ATF_line.gif
I think it's from 1904, but I'm not sure.

> ATF Garamond was cast on Art Line, though

Yes, it seems the descenders of the 6, 7 and 8 points don't follow the above graph - but I'll have to check that.

> a chart of alphabet lengths is given on pp. 584-585.

You know, I saw that, but didn't realize it would have saved you the effort... :-/
But it's still good to have a check. Maybe I can see if the fonts I used were the real thing or not.

> The 14 point, BTW, is listed as 149, versus my measurement of yours at 148.

But that's strange, isn't it? I mean because the letterforms are different.
I wonder about the dates the two 14s were each made. Maybe they changed their mind about it being text or display.

> Good ol' fashioned brute force, man.

I like.

--

Your ATF/Linotype comparison is amazing! So it seems Linotype optical scaling was not bad at all! The thing is, now I'm thinking it's too similar. Maybe Linotype did what I'm doing now, if you get my drift...

> 7 point, which isn't listed.

That's interesting. Maybe it helps explain the off color on my 7.

Anyway, I hope you come back to the analysis soon!

hhp

hrant's picture

> It would seem strange that there would be optical adjustment of stroke weight and contrast but not of extenders

rcapeto's picture

But note that the pantograph was capable of a lot of compensation
(even "automatically", through cutting slips)


Do you have (or know of) any diagram showing how this worked?

hrant's picture

Oh, believe me, I'm working on that! :-)

hhp

hrant's picture

Where can I see what ATF's "Art line" alignment was like?

hhp

hrant's picture

I have the '23 book, as well as two others that don't have dates, but none of them seem to show a chart of exactly what the vertical cutoffs were for each size of the Art Line, the way the chart below does for their Standard Line.

(39Kb) http://www.themicrofoundry.com/other/ATF_line.gif

hhp

kentlew's picture

Yes, Art Line refers to an alternative proportion of vertical space. It is used for designs that ATF wanted to allot more descender space for -- usually, of course, this is at the expense of some x-height, and thus the results look more bookish and "elegant", like Garamond, Bodoni, Cloister.

I presume that the point of declaring in the catalog that a face was cast on Art Line was to indicate to the customer that these fonts would not align (baseline) with their other fonts cast on standard Point Line.

Nevertheless, I would guess that there was a standard established for each size of Art Line that determined the placement of the baseline on the body. That is, I would expect all fonts of the same size, cast on Art Line, to align with each other. Is this not so, Jim?

It is this standard aspect of the alignment that I believe Hrant is interested in.

I presume, Hrant, that it could be derived from samples. I would guess that it follows similar principles to Point Line, but with different figures. As you will see if you examine the Point Line arrangement carefully, the purpose of the standard was to make it possible to place different sizes on the same line and align them by cutting appropriate point-size leads to make up the vertical differences. That's why the division of ascent/descent was established in whole points (thus "Point" Line).

But perhaps I am wrong, and Art Line did not accommodate mixed sizes on a line easily, in which case it may not be so simple to derive from examples.

-- Kent.

kentlew's picture

Okay, here's the reference I was trying to remember. McGrew states on page xix:



quote:

"The successful revival of Caslon Oldstyle [which I believe was cast on the original alignment -- i.e., from original matrices with no standard baseline position -- K.] encouraged ATF to design Bodoni with longer descenders than standard alignment permitted. To accommodate them, Art Line was developed, with an allotment for descenders of about 30 to 33 percent of body size. Some foundry faces are made to special alignments. But in all cases, foundry types maintain a point-multiple scale of alignments, which are precisely adhered to."




-- K.

gerald_giampa's picture

Kent,

Quote McGrew,

"The successful revival of Caslon Oldstyle [which I believe was cast on the original alignment -- i.e., from original matrices with no standard baseline position -- K.] encouraged ATF to design Bodoni with longer descenders than standard alignment permitted. To accommodate them, Art Line was developed, with an allotment for descenders of about 30 to 33 percent of body size. Some foundry
faces are made to special alignments. But in all cases, foundry types maintain a point-multiple scale of alignments, which are precisely adhered to."
......

I may be a little picky about your note. But I do not think this holds truth.

Quote, note insert, Kent

[which I believe was cast on the original alignment -- i.e., from original matrices with no standard baseline position -- K.]

Caslon always had a standard baseline position, based on its own accommodations. Line up standards were a primary foundry issue. Line up for Caslon had to change for the point and pica system. Its position had to be "first found", then imposed. ATF inherited Caslon matrices with yet a different body system than the original Caslon foundry. I can not remember whether the Caslon Foundry based their transfer point and pica standard to mimic American Type Founders. I believe that is the case. I wish I had the source here or I would check. What I do know is that Lanston had their cutting in 1915 which predates the Caslon Foundry refitting released in their 1924 Catalogue. Theirs, ATF and ours seemed to line up pretty dam good. Made me wonder.

Unfortunately I am rusty on all this now.

In any event it would be impossible to preserve the original line up standard from Caslon's original matrices. There would be no point. (Pardon the pun.)



Gerald Giampa

hrant's picture

> usually, of course, this is at the expense of some x-height

I'm not sure this is a correct view. I do think that designs with long descenders tend to have smaller x-heights, but that's irrespective of the ATF alignments.

> I would expect all fonts of the same size, cast on Art Line, to align with each other.

I'm not Jim, but yes.

> the purpose of the standard was to make it possible to place different sizes on the same line and align them

But also to put different fonts on the same like and align them (with no effort).

I was going to quote McGrew, but...
As a side note, he also says that the standard point line descenders were about 20-25% of the body height, but my own calculations from that chart give a range of 16.67 to 28.57 percent, with an average of 21.61 .

BTW, I see about half a dozen versions of Caslon in the '23 book, cast on different Lines.

hhp

hrant's picture

I think the two baselines in the Standrard verus Art Lines would have to NOT align.

BTW, I have access to ArcheType's collection, and it's very good. But they're moving now, and won't be back up until January. In the meantime, I can always scan the stuff in the '23 book. But thanks for the offer - maybe one day I will indeed need to get in touch with the Unusual Mr Botterill!

hhp

gerald_giampa's picture

Jim,

I think it would be great if you introduced Hrant to Guy Botterill. Guy is a fabulous resource. Could answer many questions. In fact Guy should partipate in this very forum.

Gerald Giampa

gerald_giampa's picture

Jim,

I think it would be great if you introduced Hrant to Guy Botterill. Guy is a fabulous resource. Could answer many questions. In fact Guy should participate in this very forum.

Gerald Giampa

kentlew's picture

>In any event it would be impossible to preserve the original line up standard from Caslon's original matrices.

Gerald, this makes sense and I suppose you're right. I was trying to interpret two things: the ATF claim that Caslon Oldstyle was cast from original matrices (a little bit dubious) and the indication in the 1923 catalog that it was cast on "Original Line" (whatever that means). But you're right, of course, the original Caslon types would not have been cast on point/pica bodies.

>I'm not sure this is a correct view. I do think that designs with long descenders tend to have smaller x-heights, but that's irrespective of the ATF alignments.

Well, Hrant, the addition descender length has to come from somewhere. In general, it looks to me like most of the additional space has come from the x-height, not the ascender; although, I'm sure it varies.

It would be interesting to confirm whether or not different faces cast on Art Line baseline-align with one another. I don't think you can do that definitively from a scan.

My main point in bringing up this alignment stuff in the first place, however, is just to point out that the proportion of the ascent/descent in different sizes of ATF fonts may have as much to do with the established alignment standards as it does with any notions (Benton's or otherewise) of proper optical compensations. So take any curves with a grain of salt.

-- Kent.

gerald_giampa's picture

Kent,

Actually, I am not sure what the ATF advertising literatures said but their matrices were made from "strikes" off the "original punches". The original matrices were still at the Caslon Foundry. If not they would have to ship them back and forth as the orders came in. (I hate that, I sound like Nicolas Barker?) Right hand column.

http://www.lanstontype.com/Burgess5.html

I can't imagine why they would have attempted to match existing line. Each body would have had significant differences. They would not be attempting to replenish existing stocks for early customers of the original Caslon. God forbid!

But Kent, I have learned one thing in the foundry business. Never presume anything. Never think for some other company. They will do the most peculiar things.

So I reserve the right to be wrong.


Gerald Giampa
Lanston Type Company


hrant's picture

> "Original Line"

That seems to be the original alignment of the source design, as opposed to the one ATF would impose ("Standard" or "Art"). That explains why the descenders of that one aren't "cramped", but it's not Art Line either. So their various Caslons were cast to three different alignments.

> In general, it looks to me like most of the additional space has come from the x-height

Yes, but I was saying that it doesn't have to, that it's a decision outside the scope of the Line it's in.

> It would be interesting to confirm whether or not different faces cast on Art Line baseline-align with one another.

You mean McGrew might be wrong there? Because he does state that they do.

> may have as much to do with the established alignment standards

Ah, but why doubt that the establishment of those standards was devoid of consideration of optical compensation in the first place?

hhp

kentlew's picture

>Yes, but I was saying that it doesn't have to, that it's a decision outside the scope of the Line it's in.

Oh. Okay. I didn't understand you. Sure, it doesn't have to; but then it comes from the ascenders.

>You mean McGrew might be wrong there? Because he does state that they do

Well, he strongly implies it, but I don't think he says so explicitly. He mentions point-multiples, but I suppose one face could have a slightly different arrangement than another, such that they wouldn't align in all sizes. But, I doubt this; so yeah, probably Art Line fonts all baseline-align with one another.

>Ah, but why doubt that the establishment of those standards was devoid of consideration of optical compensation in the first place?

Well, I was thinking it was developed by committee, but I see now that I was remembering something else. So I suppose that there could be some consideration. I'm just saying that a part of what you're going to find has been restricted by what is essentially vertical unitizing. (Back to that.)

-- K.

grid's picture

Okay, this is not really to the point in this thread, but it does have some interesting similarities. Lucas de Groot posted this in 1987, the plot looks similar to at least some of what Kent has posted. PDF attached.

I

hrant's picture

Bill, in fact I previously alluded to de Groot's "quantization" work in justifying my view that what we're seeing is probably quadratic.

But I have to suspect that Bringhurst is too much of an artiste to favor this type of thing. His perspetive of geometry in type design is probably of the metaphysical flavor... The Elam book I have, and I think it's pretty cool.

hhp

kentlew's picture

>justifying my view that what we're seeing is probably quadratic.

You've said that, but now I've been looking at this again and I don't see that it is.

I think I've shown convincingly that the alphabet length is essentially linear. Scaling the data to equivalencies to some master size doesn't really change the essential relationships of the lengths. There should be a set relationship between the linear graph and the scaled curve, and I don't think that relationship is a quadratic one.

I went back to the alphabet lengths given by the ATF book and revised my formula. The text portion of the scale (6, 8, 10, 12, and even 14) can be expressed by y=7.83333x + 40, where x is the point size and y is the alphabet length. (The accuracy of this is within 0.33333 pts of the stated length for 8 and 10, and 0.66667 pts for 14.)

If you assume a 12-point master drawing and scale everything to that, then the new curve (which is shown in one of my graphs above) can simply be expressed as y = 480/x + 94, where x again equals the point size and y equals the alphabet length, scaled to 12-point equivalency. Again, the accuracy is within about a half a point.

It's not quadratic at all.

-- K.

hrant's picture

> the alphabet length is essentially linear.

I guess it depends how you look at it.
From the perspective of digital type design, the necessary thing is to know/decide what kind of modifications need to be done to a "deliberative ideal". As the curve in the graphs you made show, this is higher order than linear.

hhp

grid's picture

>Bill, in fact I previously alluded to de Groot's "quantization" work<

Missed that Hrant, long thread :-)

I too like Prof. Elam's book, but have found at least one small error. It is incorrect to state that you can draw the spiral for a golden rectangle by drawing arcs from the corner points. Rather, you get a good approximation. The true form is expressed by a mathematical equation where there is only one origin.

Also, I think Bringhurst is aware of a number of different ways of applying math theory to typography and design. However, the only thing I

kentlew's picture

>As the curve in the graphs you made show, this is higher order than linear.

Either you're missing my point, or I'm missing yours.

I just showed that the curves I graphed were just linear relationships looked at a different way -- but it still all boils down to a linear expression. The line that was first graphed is expressed as some factor *times* the point size plus a constant; the curve produced by scaling the data is expressed as some factor *divided* by the point size plus a constant. The factors and the constants are closely related and can be derived one from the other. I don't see a "higher order", certainly not a quadratic one.

I'm not following you.

Show me the Math!

-- Kent.

hrant's picture

Kent, let me try to explain this - as much to myself as to you!

You're physically designing a digital font, and you have the "optimal" design let's say for the 12 point cut. Now you want to alter it to work better for smaller sizes. Assuming you're going to use numbers to do so (or at least get the first iteration), then the relationship of the numbers is not linear. Here's why:

Let's say you're working on the width increase. If you follow the ATF Garamond, the ratio you'd adjust for the 8 point is ~115.3%, and for the 6 it's ~129.9%. The percentages 100, 115.3 and 129.9 do not sit on a line when graphed against point size. The point is for example that you can't interpolate (linearly) the 8 point number from the 12 and the 6.

hhp

kentlew's picture

Okay, yeah, I got that. They don't sit on a line. I can see that.

But are you seeing my point? It's the "higher order" and "quadratic" that I'm having trouble with. When you say "higher order" I imagine that there will be a variable squared or something like deGroot's formula.

What I've been attempting to show is that the formula for these curves is actually very simply derived from the linear equation.

Let me review. The general expression for a linear equation is y = sx + c, where s is the slope and c is a constant, right? The equation for your "scaled" curve is simply y = mc/x + ms, where s and c are still the slope and constant from your linear equation, and m is the point size of your master design. It's an inverse relationship -- not what I would call a higher order.

The curve for your multipliers would simply be the same expression divided by the length of your master alphabet.

Practical example:

The equation for the ATF Garamond alphabet length for the sizes 6-14 point is

y = 7.8333x + 40

Here are the results of the equation:

6 : 87
8 : 102.667
10 : 118.333
12 : 134
14 : 149.667

(The catalog gives 8 : 103, 10 : 118, and 14 : 149. I hope you will accept that the equation indeed fits the data if you allow for some rounding.)

If you want these scaled to be equivalent to the 12-point, then the equation is

y = 480/x + 94 There's your curve.

If you want your multiplying ratio for ATF Garamond (12-point master), then just divide by 134.

y = 3.582/x + 0.7015

The results:

6 : 129.85%
8 : 114.93%
10 : 105.97%
12 : 100%
14 : 95.74%

The differences between these numbers and yours go back to the rounding differences.

All right, enough Math! We're scaring away all the art students!

I've made my point. I don't think it's a higher order. Take it or leave it.

>The point is for example that you can't interpolate (linearly) the 8 point number from the 12 and the 6.

The last thing I'll say on this is No, you can't interpolate the percentage linearly, but you can easily derive it from a linear interpolation. That's my basic point.

Give me the length of a 6-point alphabet and a 12-point alphabet and I can easily derive the intermediates according to this ATF Garamond model, without any quadratic or cubic or any higher order math. Don't let those curves fool you.

I'm not saying all the parameters are that way. But this one surely is.

-- K.

hrant's picture

To be honest, I really don't know now... But one thing I'd point out is that the first derivative of a quadratic equation is of course a line - although at this point I've become about 50/50 on this.

hhp

hrant's picture

I found an interesting graph in that '68 issue of the Penrose Annual:

scales.gif

That "Harmonic Set" business is something the author of that article (Louis Rosenblum) is using as some sort of measure of optical scaling. Gotta read the whole thing.

hhp

kentlew's picture

BTW, Louis Rosenblum is a trustee of the Museum of Printing (where all the Mergenthaler Linotype drawings are housed). I don't recognize that particular graph, but I will be interested to hear about the article.

Also, in answer to a question you once asked: I have been told that Griffith regarded Ruzicka's Primer as the best example of Linotype's grading scheme. I don't have any firsthand evidence, but I have no reason to doubt these sources.

-- K.

hrant's picture

> Ruzicka's Primer

Got it - I'll try to track it down.

hhp

anonymous's picture

Regarding unit systems: there seems to be some doubt in this forum as to its value. I do not suggest that posters are detractors, but there seems to be a bit of conjecture about its flexibility. Many are aware that when FW Goudy was presented with the task of designing his Goudy Gimbel (38E) for the Monotype system, he found it a long and difficult task, having never before been restricted to the width of his letters. You have to remember that Linotype and Montype were all there was for machine setting in those days. Neither were considered crude or archaic, rather they were thought of as amazing marvels of ultimate "hi-tek" engineering. They now seem to funny to most, but they were the result of a logical evolution from hand-setting to machine setting from recirculated cold types (the Page machine) to the very first Lino and Mono machines, which were even at that vastly different from the first successful operating versions of both.

It is understandable that FWG would find this restricting. He did, however deal with it, and the type was a great commercial success, as is evidenced by the fact that many matrix cases of it still survive. I don't find 38E to be the prettiest type ever done, but I don't see anything wrong with its set values or fit. I maintain that it is not terribly restrictive to work within assigned unit widths for a letter. The standard C, C1 and C2 arrangements accommodate a grteat many of Monotype's faces. This is due to none of them requiring unusually wide or narrow boundaries for an a or an e etc. What I am saying is that if a face is based on a kind of "ideal" set of proportions everything works out in the end.

In the case of certain new releases it was not uncommon to allow the type designer to create an new set of unit row arrangement for the die case and it attendant keyboard stopbars and caster wedge. This is evidenced in Goudy's Kennerley Oldstyle, whcih uses not only its own keybar arrangement, but is also the first Monotype face to have an expanded system of 20 units. If you look through any jumbo Monotype speciman books you will come to an antire two page spread of noted Special Arrangements that expand on the standard 18 unit C arrangement.

In setting Monotype the keyboard is constantly aware of units and the calculating of them. This is an absolute requirement in the setting of tabular work. In text work one is also constantly aware of unit counting in a much simpler way.

I was not really aware that Liontype even worked on a unitizes system. What made this a revelation to me is that on the Linotype, the channels (72 or 90 channels) carrying the mats have a goodly space each side of the channels of mats, and I assumed that there was lots of latitude to widen or skinny up a character to some degree. This makes no reference to the problems of an italic and roman being on the same matrrix. So I learned something.

In setting straight matter on the Linotype, a person is not required to count units, merely to keep an eye on the assembler so that when the measure is nearly filled he or she can decide how to fill out the line. As Kent Lew has noted the expanding spacebads do most of the work just like your computer does in setting justified). In some cases, thinspaces are dropped into the line of assemble mats by handl, for situations like small caps. For a few years I was a Linotype operator on dialies and weeklies and did a lot of typesetting. On the Linotype, tabular work was terribly difficult, if the shop did not carry the proper sets of tabular values fixed spaces. If this setting was an a narrow measure, there was little room for more than one spaceband, and it was tough going. The most awkward thing I ever had to set was track results for sulky racing. This had to fit in an 11 pica column, in a special very condensed 7 point condensed gothic whose name I cannot recall. This was so unusual that the racetrack promoter carried his own set of special Linotype mats that he presented to the weekly paper in each town that he travelled to for the season's racing, since he made his money on selling the race tout sheets. Sorry for the walk down Memory Lane. Old guys do that.

What I suggest is that with Linotype and Monotype the unit system did not cause any typeface to be a failure because of the restrictions of width. I can't think of a face where an s or an a is pronouncedy too wide or too narrow, and I don't think it is because I am easly pleased.
I am ignorant of the Sabon a too-wide problem, but my guess is that perhaps it was a conscious and decision by it designer. I don't think that is impossible, since humans do make some odd choices.

It's Saturday morning and I have just spent the last five days cutting and recutting an 18 point letter f punch. The matrix is punched and rought trimmed, so this morning I get to cast it an line and fit it . . . so here goes nothing . . .

JIm Rimmer

anonymous's picture

Hrant

ATF's reference to their "art line" referred to the proportions (parameters) of the letter, according to them. One such example is their Cloister Oldstyle and Cloister Light families, which are drawn to have a smallish x-height, with more dimension left to the ascenders and descenders, with the largest share of the space dedicated to the ascenders.

If you can get your hands on a 1912 or a 1923 big book, you can scout out what other faces are drawn to their "art line".

As you will see if you can read that part in their big books, it doesn't refer to the actual mechanics of alignment specifically, but rather to the faces' artistic consideration of height dimensions.

Jim Rimmer

anonymous's picture

Hrant

I seem to remember that there is no grid shown for their Art Line. They merely make mention of it here and there in reference to their more classically-based types. Sorry I don't know much more than that.

Jim Rimmer

anonymous's picture

Kent, Hrant

I think I understand that the question is as much about standard alignment of Art Line types as it is their anatomy, and it sort of makes sense that if two faces are both drawn to those paramters, they would match in base-alignment. However, I don't have castings of any two such types that I could proof together or put on an alignment gauge to test. So I wouldn't want to assume that they would mechanically align on the base.

If I were going to pursue this line of research, I would first try to get actual cast letters of two differing faces done on the Art line. For my purposes of checking metal type alignment, I select a letter with the most emphatic baseline. Something flat like the cap or lowercase z works well. If the termination on the baseline of the z descends a little, I pay attention to the horizontal stroke which gives a good indication of the baseline.

There is a man in Baltimore who own a metal type treasure-house called the House of Type. He is not listed in any phone book under that name. His name is Guy Botterill. A most UNUSUAL man, with one of the world's largest collections. I can't say that he will give you a specimen letter, but you could ask.

If you can't find him, let me know, and I will dig through my haystack of papers and see if i can find him.

Jim Rimmer

quadibloc's picture

Adding a comment to such an old thread, particularly one which contained a debate that had become acrimonious, may be unfortunate, but I hope that facts will calm things down rather than inflame them. (I had encountered this thread as the result of a web search for the original unit system for TTS fonts.)

The history of the legibility faces from Linotype was, I thought, well known. Alexander Lawson's "Anatomy of a Typeface" gives a good account of it. First there was Ionic No. 5, with low-contrast strokes and a generous x-height. Then there was Excelsior, which was designed to address a problem of ink-trapping that was observed in one particular case.

In high-speed rotary stereotype presses, the rubber rollers that applied the ink had to be made of harder material so that they could withstand more friction. It was at this point that ink-trapping became a problem that had to be addressed in Excelsior's design.

Not ink-trapping per se, but the need to apply ink more heavily so that halftone photographs would print properly, led to a flooding of counters with ink in tabloid newspapers: and so Paragon was designed.

Then a different problem - mold shrinkage - led to the design of Corona.

So while ink trapping was a real phenomenon, it only played a role briefly, and this was basically because presses were being operated in a way that could be expected to lead to poor results.

hrant's picture

Many fonts made today have trapping.*
+
People tend to strongly avoid unnecessary extra effort.
=
?

* Here's one from last week:
http://typophile.com/node/64634

hhp

riccard0's picture

I think nowadays trapping is just an aesthetic fad.

hrant's picture

Yeah, DTL is into fads.

hhp

riccard0's picture

No, they're into "unnecessary extra effort". ;-)

hrant's picture

:-)
I guess any type designer can be accused of that!

hhp

quadibloc's picture

I came across another thread where it was noted by someone who worked for Linotype that their drawings of type were generally done to an 18-unit system or a 54-unit system, and that this became useful later on, when they came out with a phototypesetter that required this.

That came as a surprise to me, since indeed the Linotype mechanism makes no technical requirement for a unit system - since every space behaves like a quoin.

But this reminds me of something else. Given that the normal assortment of spaces and quads in a job case consists of em, en, 3-to-em, 4-to-em, and 5-to-em, plus the hairline space, and even the hairline space has a finite width, in my naivete, when first exposed to that I would have thought that even foundry type would have to be made to a unit system of 60 units to the em. If slugs of type had widths like pi/4 or sqrt(2)/2, they would drive typesetters to despair.

The favored explanation, though, was that the accuracy to which a line of type needed to be justified was finite, and just touching the sides of a slug would change its width due to one's sweat and the natural oils of one's skin and so on.

quadibloc's picture

Also, on the question of Linotype fonts being designed to a unit system:

This page,

http://www.time.com/time/magazine/article/0,9171,855719,00.html

notes that TIME magazine used TTS machines to operate Linotype casters during World War II because they needed to send identical copy to multiple printing plants. TIME, of course, was a magazine, not a newspaper, and I do not believe it was set in one of the newspaper legibility faces.

On this page,

http://www.gochipmunk.com/html/teletypeperforator1.html

a manual is quoted as saying that the use of TTS equipment to control Linotype machines has become a general practice.

I had thought that the multifont perforators were needed because even Corona wasn't designed with the same number of units per character as the TTS version of Ionic No. 5. That they could have been used to control Linotype machines setting magazines in Caledonia, however, appears not to be beyond the pale.

Of course, this still wasn't an intrinsic limitation of Linotype. Presumably, if indeed this did happen, Linotype started unitizing its fonts sometime in the 1930s, when the advantages of using TTS for other purposes than newswires became apparent.

Don McCahill's picture

TTS was used extensively in newspapers when I was a junior reporter in the 70s. The reason ... Linotypes were expensive, and had limited capacity. TTS punchers were cheaper, and you could have two (or more?) setters working on the TTS to feed one linesetter, optimizing the resource.

quadibloc's picture

Exactly!

One of the original advantages of the Monotype system over Linotype was that the Monotype caster could be kept continuously busy, at an optimum speed, playing tapes generated on multiple Monotype keyboards.

The 18-unit system was a typographical disadvantage intrinsic to Monotype which Linotype did not share. On the other hand, Linotype had the disadvantage that Roman and Italic were normally identical in width (but bold was on separate matrices, and small caps were put in their own locations), and this was more noticeable.

When the TTS system was introduced for newspapers, it was noted that it allowed a Linotype machine to run more quickly and smoothly than one operated manually.

So, given that someone who used to work at Monotype noted that they were designing their fonts to an 18-unit system, what I think may have happened is the following:

Originally, when Linotype was first designed, what you see in the advertisements and the standard textbooks was right. Monotype had the unit system, Linotype had Roman and Italic equal in width.

After TTS was introduced, and people saw the *additional* benefit of how efficiently it made Linotype machines run, popular demand led to:

  • the creation of the "multifont keyboard" which could be set up to punch tapes based on different unit values for the letters than those standardized for the newspaper fonts, and
  • Linotype making their matrices for fonts other than the newspaper legibility faces to an 18-unit system, so that those fonts would be usable with TTS.

So even where paperbacks or magazines, instead of newspapers, were being set on Linotype, the machines could still be used as efficiently as Monotype casters, thanks to TTS being universally applicable.

kentlew's picture

> Linotype making their matrices for fonts other than the newspaper legibility faces to an 18-unit system, so that those fonts would be usable with TTS.

I think that you may be drawing a conclusion that is not completely warranted.

AFAIK, the Teletypesetter system of tape-driven composition did not intrinsically require unitized fonts.

The TTS system was, foremost, a standardized encoding system for tape perforation that facilitated long-distance [telegraph] transmission of keyed text.

Indeed, a side benefit is that this allowed tape-driven composition on an suitably equipped Linotype machine, similar to the benefit seen on a Monotype.

However, I believe that unitization is a separate issue.

The 18-unit system was an integral part of the Monotype matrix setup and its justification procedure. A unitized system allowed progressive calculation of the composed line and simplified the mechanical computation of space adjustment for justification.

The Linotype, on the other hand, relied upon a physical mechanism for line justification -- the tapered spaceband. So, unitization would not have benefited this process.

I have always understood that the purpose of a unitization spec as part of the TTS system was for standardization in newspaper syndication. Part of the original impetus for the development of TTS was a more efficient means of transmitting news stories. The addition of unitization standards would have allowed a story to set to a consistent measure and length in different newspapers with different body types.

This standardization would not have benefited book compositors.

As I understand it, the whole point of development of the Multiface Perforator was to work with non-unitized fonts and thus a different "counting" mechanism for engaging other operator functions.

What I do know for a fact, having spent some time analyzing the production drawings of Caledonia (drawn in 1937, well after the introduction of the TTS system), is that this book face was *not* drawn on an 18-unit system. Although I have not analyzed other designs as closely, I am highly skeptical that any other book faces were unitized either, prior to the introduction of Linotype's own unitized phototypesetting systems.

Syndicate content Syndicate content