I found when I was designing type that, after a certain period, it was actually much more a discouragement to look at these typefaces than it had originally been an inspiration. They were finished products. They had been given the final spit and polish by a manufacturer anxious for them to succeed. So they were hundreds of generations away from the tentative, awkward early sketches that begat them. It’s easy to forget this, and start believing that Frederick Goudy’s pen drew the K in Kennerley perfectly first time – complete with all its mock-tudor pomposity. I have to say that I expect the pomposity got there first. Then the perfection.

But even if your inspiration is not highly finished type, it’s hard to approach the model – either to copy, vary on the theme, elaborate or (watching out for hubris) improve on it. The lines curve off into infinitesimally small details, and there’s always more to uncover. At what point do you go beyond the reasonable and become obsessive?

To make matters worse, when you stand back from the model, there’s the ‘swing’ it has – the way it carries itself. In more straightforward terms, it’s the features that dominate when the type is glanced at, seen as an object, rather than read. These are things like the ratio between the x-height and the extender height, the amount of ‘body’ the type has (colour in the central x-zone rather than in the limbs), the width of the face and any quirks which get frequently repeated (as would happen, for example, if the quirk was part of a lower-case e and the typeface was used for text setting). You can get more of an idea about what I mean if you consider the huge impact that the style of a g’s lower half has on a typeface. (If you don’t know what I’m talking about, you’re reading the wrong article!) That swing can be very hard to put back into your design: if you can get it at all, is it at the same height? In the same proportion? On the same page?

Early in the typeface’s design stages, it picks up quirks like these for itself, and most (luckily) don’t survive natural selection. But some do, and it’s interesting to see the way in which that tail on the g is a harbour for so many freaks of typographic nature. There’s been much ink spilled over the topic already. Typeface identification books (such as Rookledge’s Typefinder) use g and its capital, acknowledging their particularly obvious idiosyncrasy. I think the g’s tail is, along with the shape of the a, one of the things I enjoy fiddling around with the most. It’s not illegal yet. And looked at in a more rational state of mind, it is surprising how much variety can be made within the skeleton of these two letter shapes: try finding the same variety in e or n or k.

So if your model is frustrating, filled with tiny details of huge consequence, maybe it’s best to ignore it. And so that is what I tend to do. Once there is an outline design for a typeface in my mind, I try to avoid looking closely at anything similar. So, in the case of Puritan (influenced by Scala sans, The Sans, Syntax and Pragma) it meant avoiding too close a scrutiny of these faces whenever possible. There were also serif typefaces like Ehrhardt and Fournier which were very important influences; again I tried to keep away from them. I felt that they had left a strong semi-conscious impression on me, and that I’d be able to get what I wanted without direct reference. Like sketching from memory, this would be a third-generation report: in through the eyes, filtered by the brain’s process of analysis, storage and retrieval from memory, and out through my hands (and the mouse).

Looking at the foregoing, I suppose I am describing type design pretty much as it must be done. The act of making a close copy necessarily involves interpreting, and so much more is needed when you want to generate an original result. All the time, the designer is thinking about whether the shapes they have created will be harmonious in groups, about whether the original typeface achieved harmony in similar groups, and about whether that harmony is at the cost of something else. There will be ongoing evaluation of the shape and flow of curves, the angle of stress in the face, and the way the strokes end.

Where my process departs from this is in the deliberate removal of the models that have inspired the face. It takes away a very useful and consistent reference, so that there’s no answer to questions like ‘should the top of the f hang over the next character, or should it be cut off short?’ apart from the most obvious, but frustrating, which is to try it and see.

Without making any measurements, Puritan came out to have a set that is somewhat similar to Ehrhardt’s. But you can’t combine them; there’s not enough similarity in character. Puritan is clearly not incised, and neither has it incisors: Ehrhardt needs a type with bite to partner it. On top of that, Puritan isn’t really a very good design anyway, certainly not to partner a ‘commercial grade’ serif face. It’s wilfully sloppy (see my article on drawing methods) and there’s too much variety in the letters. Without serifs to hold the whole wobbly mess together, Puritan is just a bit too weedy.

But this is not a discussion of Puritan. I explored the limits of type design using a ‘concealed model’ most fully when I drew Acknowledgement. Once again there was a rider: I was in a hurry, because I thought I could give the font away to a friend for a birthday present (the cheek of banking on this exercise being a success can be mitigated by the fact that I knew she was firstly enthusiastic and secondly knowledgeable about type, so there was little to be gained from a rotten job). I started with some wooden type as my model – type that I put back into its drawer after doing some sample prints and haven’t looked at since. I looked at the prints I had made quite carefully, but they were not given any kind of formal analysis. They too were put away in a drawer.

Then the type drawing began. I was working on a ‘fat face’ – a really bold, really wide serif type from the heyday of wood type in the mid nineteenth century. Caps only, nothing clever, probably not even a pound sign in the case. Excepting I, most letters were wider than they were tall. It proved marvellously easy to put in the initial outlines; there’s no subtlety in these faces. But there is skill; and so when it came to adjusting the proportions minutely, there was plenty of call for answers to questions like ‘where does the leg of the K join the arm? Does it join the stem instead? Should I thin it a bit, or taper it, to make it fit in? And how should the bar and stem in the G be constructed – as a serif, or as a unique feature?

I found these questions reached well beyond what I could recall about similar faces. I was forced to experiment and see what happened. And for me the most interesting outcome of this was that, looking around after I’d made the decisions and created the typeface, I discovered again the wide range of solutions to those same design questions which the original generation of letter carvers had come up with. Obviously, there’s no one way to go about designing letters, but within such a narrow range of faces, all clearly grouped in the same style, the variety was still remarkable.

That’s what I think a method like mine can offer to somebody who wants to sharpen their feel for typefaces. You won’t remember everything you see, but looking through old books and posters, and at old type itself, gives you access to all the subtleties and tricks, all the combinations of stems, bowls, tails and serifs that have been attempted. Some worked, some didn’t; but taken together they’re a fascinating body of graphic marks, and abstracted from the world around them they can tell us a lot about the need type designers have felt, faced with awkward glyphs which won’t fall neatly into formal shapes, for an elegant solution, every time.