I think that leads us unto questions about text faces and legibility - what exactly makes a good text face. Now referring to Celeste, I would like to ask if there were elements in Celeste, that in the process of designing it, you had to change - I mean elements that you became aware of and wanted to work in a certain way, to aid and improve the legibility of the typeface?
For Celeste I had a certain optimum stroke weight in mind - not too thin, but so that type would still be relatively elegant and open - and I remember that in the first digital text proofs the colour was too light, because, like anyone else, I was drawing letters at a larger size, digitizing them and scaling them down, and, as a novice type designer, I didn't know until I did it what would happen when I scaled them down to 10 point.

What do you mean when you refer to the characters as open? Is it a question of stroke width compared to x-height...?
Yes. It's also a matter of the internal spaces being quite large, and therefore the spacing of Celeste is probably looser in general than in some typefaces. I don't generally like typefaces that are very closely spaced; I think letters need a certain room to breathe. So the counters were quite large and open, which is obviously quite a basic thing for legibility, and ensures that parts of letters don't fill in, avoiding the risk that a letter will be read as another letter.

In the process of designing Celeste, how did you make proofs to evaluate the type.
From the beginning - as soon as I could - because it was intended as a text type, I had an A4 sheet with two columns of some 10 point text. For me that's the most exciting part of type design, when you get almost a whole lowercase digitized and basically spaced so that you can make text. From then onwards, it's downhill - it can be very mechanistic and a bit boring.
I made proofs of letters at larger sizes obviously as well, to see the forms, and what was going on in the outlines.

On a laser printer?
On a laserprinter yes, on a 300dpi machine initially which isn't that much use so, as I intended for it to be a professional typeface for use with high resolution imagesetters, I had some high resolution output done as early as I could. From that I saw that the stroke weight was generally a bit too thin, so I had to go over everything in Fontographer and beef-up the strokes, which was a very laborious thing to do at the time; but as I say I was learning as I went along. Now I know more or less how bold one needs to draw letters - a lot bolder than one would initially expect for them to work at a small size. You can easily test this by printing a certain type, the weight of which you like in text, at a very large size and then look at the weights of strokes and things like that.

But returning to the topic of legibility in relation to the actual lettershapes: obviously a lot of text type in appearance seems quite anonymous but there are in fact quite a number of elements in text type, such as certain terminal strokes, certain serifs. You touched on the openness of type, and contrast - whether strokes are too thin or too thick...
Well, those are intuitions that I had In my education as a typographer I became acquainted with certain studies about legibility and the basic knowledge that dates from the late nineteenth century about what endangers the legibility of a lettershape, and that we basically recognize the top half of letters - all those kind of things. I had assimilated all that stuff. But in terms of designing a letter for optimum legibility I didn't go about it in any scientific way, it was all intuitions. I like to think that, when I look at proofs of my own typefaces, I am able to look with eyes that are not just my own, but that I am able to imagine someone else looking at this thing, and notice possible dangers of a lettershape not surviving very well under certain printing or reading conditions, or that it doesn't seem to be itself because of a certain stroke or feature - all that is really intuitive to someone whose attached to the idea of a legible text. When I read a lengthy text I don't like my reading to be molested or interfered with by the letters being in some way strange or peculiar.

Are there any particular shapes in Celeste that at a very late stage you realized, this has to change?
The state that Celeste - although it wasn't called that then - was in when I finished my Bachelors degree in typography, wasn't exactly as it was released in 1994. Originally there were practically no straight lines in it - not even the bottom of the serifs, I don't think, were flat. I had this idea, perhaps as a twist on the modern-face tradition, that all the stems are waisted, or with a certain kind of inverted entasis, which, when proofing the type at small sizes on 600 dpi printer and certainly on a 300 dpi printer, makes jagged edges and just ruins your typeface. Then I thought, well there will be designers who will be proofing this typeface until it gets printed on these kind of low-resolution printers, so do I really want it to look like this when they are looking at it? I asked myself whether the subtle curves were really doing anything for a digital type that should be used at about 10 point. So I got rid of all those subtle curves and I didn't think that it lost anything - in fact I preferred it.

But in terms of specific letters, I think that the lowercase 'a' is a particularly difficult letter to get to work - the balance of the two (if they can both be called) counters in a two-storey 'a'. Both curves, the one curving over at the top and the bowl are very difficult to combine. Some type designers have a basic kind of lowercase 'a' shape; for instance, Frutiger's 'a's are very much all variations on a theme, all versions of the same letter, and he has kind of found a solution that works. But in teaching typeface design I found it to be quite a difficult letter for some people to get right, because the directions of the almost horizontal parts, say the upstrokes of the 'a', are sometimes difficult to get to work with the directionality of your type.

Which part exactly?
These two curves in particular (points to the 'a' of Celeste in front of us). The shape of the curve can be difficult to relate to the curves in other parts of the typeface because it's a unique letter, in contrast to the way that 'n', 'm' and 'h' are similar or 'I' and 'l', for instance. 'a' and 'g' are perhaps the most difficult or interesting lowercase letters, it seems to me. 'g' is to me the most interesting; I always have lot of fun with lowercase 'g' - I found 'a' to be more difficult, perhaps because I found it bit less fun with the problems that it gave me. The small 'g' in Celeste didn't give me any problems but it took me a while to learn that the lower bowl of the 'g' doesn't necessarily have to align with the basic alignments of the type like the baseline and descender-line. This is also true of great type classics; you look at them and you realize that the lowercase bowl of the 'g' often extends less below the baseline, it doesn't go down to the full depth of the descenders like on 'p' and 'q' and this is a visual thing that people during five hundred years have obviously managed to work out, a balance of the two shapes of the bowls. I remember at some point in designing the lowercase 'g' of Celeste, looking at it and thinking that it looked too elongated, that it goes down too much - and then I fixed it.

There is a kind of convention now, speaking of the letter 'g', that the first stroke on the lower bowl should sort of visually rest on the baseline. is that the kind of thing you are referring to?
Yes, but it doesn't necessarily have to be this way. I think in fact that still is the case with Celeste, but I know that the bottom of the lower bowl doesn't extend as far as other descenders. In another type I have been designing recently, I don't think that it has any of those alignments, because I wanted the descenders to be very short. So in the end, as in typography in general, it is often useful to have alignments and guidelines for designing things, but in the end it just has to look right in terms of the eye, which doesn't follow the invisible lines behind the typeface, it looks at the letters themselves. It's learning when to break the rules that I'm talking about here and where you are able to break the rules without damaging the integrity of a type - but you'll have to learn the rules before you can break them.

What references would you give as a guideline for a want-to-be type designer?
Well, close observation and measuring of existing types, and then read what has been written about the process of type design, which isn't an enormous amount, but there are some things written by Jan Van Krimpen, William Dwiggins or Walter Tracy that have been good sources for me in particular.

You also spoke at one point about the relationship between curves in Celeste: would you say that the actual curves in a typeface, say, for instance, the bowl of the 'a', indicates or outlines a path that adds fluency to the way the letters relate to each other. For example the upper joint in the bowl of the 'a' (drawing on a napkin) may lead the eye in a diagonal direction as opposed to horizontal, and in that way relate to say the top serif of an 'n' or 't'?
Yes, I am talking about how it works in combination with all the other letters because type design is to some extent about designing the letters as isolated entities, but then, if it is to work in text, you have to look at them all in combination, and obviously the rhythmic or directional flow of a typeface is created by certain directional strokes or by the way the stroke of the 'a' has to agree with - or at least contribute to - a rhythm. If you are talking about a text type and you don't want the 'a' to stand out, then it has to harmonize with the rhythm created by the curves in the rest of the characters.

Looking at Celeste I found that some characters, which are kind of key characters like 'b, p, q and d' compared to other perhaps more relaxed curves, have got a particular pull which actually adds a nice type of dynamic and flow to Celeste, creating a balance between more static qualities of the type.
Yes, that was the idea, to make those curves not symmetrical, as it were. One thing you can do in digital type design is make at least two of those letters from at least two other letters by rotating them 180º; you can quickly make a 'q' from a 'b' for instance. Particularly if your typeface has a very calligraphic influence, that might work to start with, but then obviously you might want to change it a bit. But I don't think I did that at all in making the 'b', 'd', 'p' and q' in Celeste - I was making the curves unique from the beginning, and I wanted to give them a certain kind of energy or asymmetrical aspect which added in a certain way to the dynamic of the typeface in a very subtle way; it wasn't conscious then, but now that we are talking about it, I can see exactly what you mean. But it was part of the general idea not too make those curves exact mirror images or rotated versions of other characters.

So you worked on the letters on a very individual basis?
I did partly through intention and partly through ignorance, not knowing how to make the process faster and make it easier for myself - but in the end that probably added something positive to the typeface. There is a temptation and a possibility in digital type design, because of the cut-and-paste way of working, of making type quite modular or shapes being exactly repeated shapes, that can potentially add to a certain lifelessness. This is a contentious view and I know that there are good and respected type designers who would disagree with me, who think that type design, for text at least, is about a comfortable repetition of shapes and that the exactness of digital technology can add to that and accentuate it as a positive thing; but I think that a certain kind of humane, non-repeated element can sometimes be advantageous. Given that Celeste is in the modern-face tradition to a certain extent, with a very modular and regular appearance in terms of the vertical strokes and how they work in a line, I thought that some kind of asymmetric structure was going to provide the difference, the unpredictability - if that's the right word - in the rhythm and make it more interesting than simply a very strict modern-face type.

Do you relate this phenomenon - in what you have just said - to the sometimes too precise reproduction of letters in high resolution output for modern offset printing?
I think so. In the typeface I have recently been designing I tried to make some kind of softer shapes, which approach more the slightly more blurred and fuzzier outlines that you get with lead type, because I don't always like the perfect, really crisp and clean reproduction that you get in digital and offset-lithography techniques.
I think it's great that now we have a really precise technology that can reproduce the shapes that type designers make. It is probably the most favourable time to be working for a type designer, because it is such precise technology, but sometimes, because it can reproduce any kind of shape, I get a little bit bored with the extremely clean and crisp.
I did take advantage of the possibility of this sharpness with the serifs in Celeste, for example, which are purely rectilinear and orthogonal: they don't have any bracketing, they are made of purely straight and sharply cut ends. It's a form of serif which is not very old - the first type I know or think it was included in was Trump Medieval, and also it was used in Gerard Unger's typeface Swift some years later - so it struck me as a type of serif which was still ripe for picking and could be used in new ways. But, as I said earlier on, originally there were lots of subtle curves going on in that serif, although I always had a sharp end to it because I wanted a certain type of crispness, as that was part of the modern-face tradition; but in the end I decided to make it just purely straight. It's a kind natural form to give to a digital type because if you are designing a new typeface in a digital medium you'd have to have some kind of other intention to do otherwise. I mean to make serifs that are scooped at the bottom and are rounded at the end, like in some digital adaptations of Garamond types for instance; those are very complex forms in digital type because they are informed by all kinds of historical judgements about how those shapes were - interpretations of lead type models and that sort of thing. Whereas if you are designing a new serif in a digital typeform then the easiest thing to do basically is to make some corner points - it's less trouble and it also gives you a kind of clean and crisp quality in the typeface which you can say is perhaps native to digital typefaces.

As you know, I've dabbled a bit in Fontographer designing type and I've noticed when looking at serifs in small point sizes, that they actually turn out a lot more like Garamond type of serif you mentioned once put through, for instance, a laserprinter.
Sure, that starts to blur the outlines, depending on the level of the ink in the toner and all kinds of other things, like conditions of humidity, it might give you a slightly more blobby image. If you are designing type for output and printing on paper it seems to me that you are never going to get an entirely trustworthy print. If you look at a print that you get from a laser printer now compared to a different one few years ago you might get a completely different image. It depends on the colour of the paper, the quality of the paper and all kinds of other things.
Who's to say that when Garamond cut type in 10 point, he didn't want to make his serifs have purely sharp ends? It was just physically impossible for him to do so with a cutter working on a 10 point piece of steel. Even if he did so, as soon as it was printed it would become slightly rounded because it was a piece of metal being punched into damp paper. So in an interpretation such as Adobe Garamond the very subtle and rounded ends to serifs constitute a very complex interpretation of what people think that Garamond's ideal serif and intended form was. This can only be a form of divination, we can't strictly know unless we have a time machine and we could go and ask him - in antiquated French.

Does it make any sense this intended recreation of typefaces? I don't mean the free interpretation of earlier type but more the revivals pretending to be true to originals.
Yes, I don't see why not. I think the Adobe Garamond is a very well done and beautiful revival or interpretation of a Garamond, but if you look at original Garamond types they are not the same and they can't possibly be the same because they are done in a completely different time and technology. I think making sensitive reproductions of classic typefaces is useful; because they are classics, they still look contemporary. I mean basic forms of roman types haven't changed an enormous amount for five hundred years and the ones from that era still work - perhaps the best, so why not have good versions of them around to be able to use.

I find the reverence for old typefaces curious in the sense that not many people do really have the experience of judging first hand nor reading these types, although they are the ones ending up with revivals of those types - that is to say the common user, the people who buy books from the bookstores.
But there is no reverence for those typefaces among people who buy books. Those people wouldn't even notice the type and that's why those typefaces are good - they have this comfortable quality of nameless and timeless tradition and that's why you don't even have to think about where the typeface comes from. I talked recently at a literary conference in Wales about typography. As I was talking to an audience of literary people I wanted to make an observation about how the publishing industry is notoriously conservative and that novels are perhaps the most conservative form of typography or typesetting there is; and I knew that most books were probably still set in typefaces that were five hundred years old or more - in Bembo, or a version of Garamond. I knew just from reading novels myself that generally they were, and I did a straw poll taking about ten books of the shelf on a book shop, and they were indeed set in five-hundred year-old typefaces, well 20th century digital versions of those types. In trying to explain this to these literary people, I had pictures and everything, but if I had tried to explain this to someone in the bookstore they would have thought I was mad - or wondered what the hell I was talking about. They just want to read the books without any problems. You think that there's a reverence for those typefaces among book designers?

<< previous page ||||||||||| ^ back to top ||||||||||| next page >>