WAD to RR a letter about designing =type= Harvard College Library Department of Printing and Graphic Arts 1940 Cambridge, Massachusetts Copyright 1940 by Harvard College Library Department of Printing and Graphic Arts =printed under the supervision of gehman taylor gordon-taylor inc. cambridge= This text is a slightly expanded version of a let- ter written on July 21 1937 to a friend who wanted to know how one went about designing a typeface. =Dear RR: The= way I work at present is to draw an alphabet 10 times 12 point size, with a pen or brush, the letters carefully finished. I start with the lower-case, and let its characters settle the style of the capitals. Ten times twelve point is a convenient size to work; and I have a diminishing glass that re- duces the letters to something like 12 point size when I put the drawing on the floor and squint at it through the glass held belt high. This gives a rough idea of what the reduction does to curves and things. Having got a start on what I want by this means I turn the drawing over to G.~ and he puts a few of the characters through -- possibly lower-case _h_ and _p_. He makes his large pattern drawings (64 times 12 point) cuts, casts and proves the trial characters; and sends me his large drawings, my 10 times drawings, & proofs on smooth and rough paper. By looking at all these for two or three days I get an idea of how to go forward -- or, if the result is a dud, how to start over again. From the large pattern-sheets I can see just how details behave when they get down to size, and can change the weights of serifs, thin lines, etc., etc., accordingly. _Curves_ do all kinds of queer things when reduced; and the way lines running to- gether make spots is a thing that will surprise you -- but one or two tries on these points give you the in- formation you need. I am beginning to get the drift of it and to foresee from the large drawings what will hap- pen in the type. I can _modify_ in the large outline drawings, but so far I can't _originate_ in that medium. In making the Falcon I tried another scheme for arriving at the characteristics of the first-run ex- perimental letters. I cut stencils in celluloid -- a long and a short stem, the _n_ arch, and a loop -- _twice_ the size of 12 point -- pretty small! -- and construct- ed letters from these elements by stencilling. When I had achieved a line of these little 24 point characters that looked good Griffith ran them up with his `shad- owgraph' projector to the pattern drawing size in pen- cil outline. From these enlargements I again cut sten- cils, or, more properly, templets in cardboard, for stems, the _n_ arch, and the _b_ loop, in the 64 times size -- and made my hard-pencil outline patterns through these, la French curves. You allow for the `set-off' of the pencil-point in cutting the templets. I used the tem- plet method in order to keep as close as I could to the `action' of the 24 point originals. I'd say: make an alphabet, carefully finished, 10 "*" 12 point; getting these lines accurately placed: { | top of ascenders -------------------------o-- | --o----------------------|-- | | `z height' putting this line | where you will | | -------------------------o-- established alignment | bottom of descenders -------------------------o-- } Then have Griff.~ cut and cast two letters -- the ones that will tell you the most. I like _n_, and _p_, _d_, or _b_, a straight one and a looped one. Maybe _hp_ would be best. Then, with the `actual size' proofs from the type, your 10 times drawings, and G's large-size pat- terns in outline, you can see what you are doing; you can thicken or thin your stems or modify curves for another trial if needed, or go ahead with the rest of the letters on the original scheme. I adopted `ten times' because it was easy to work with a 0.01 inch scale -- but of course you could work any size you liked so long as it was some exact multiple of 12 point or what ever size you are shooting for. When G and I have settled dimensions, etc., to our liking, I go ahead with the alphabet on thin bond-paper in pencil outline, in the working draw- ing size -- ruling off the horizontal bounding lines accurately, and then drawing the letters quite freely at first, in the `positive' position -- passing the out- line back and forth from one side of the paper to the other, erasing the previous outline as soon as I have established its child on the other side -- modifying toward `the idea' at each change -- until I get a `posi- tive' that is good enough to mark down on the other side of the paper as a precise `negative' in thin pencil line -- =6h=. The patterns are all negative: back side to. This negative is the guide for the foundry staff's French curves and straight-edges. My drawing is free-hand (except in such cases as the Falcon templets.) I haven't any complaint to make about the staff's French curves -- they do a surprisingly faithful job. Just what hap- pens in the next step -- the reduction to the brass-pattern size (the patterns that guide the engraving-tool in cut- ting the ultimate 12 point punch) I don't know. I haven't compared working-drawing with brass yet; not easy to do. But so far as I can observe from the final proof they keep the original touch here too. Up to this point the affair has been pretty much under your control. You have made your individual letter- shapes good according to your lights, and have got them through to metal type.... Will they behave decently _when they are combined into words_? You can't tell yet. All you can do about this question, in your _drawing_ stage, is to lean hard on the hunches you have picked up as to what letters do to each other when they are fitted together. =Fitting= is the process of working out the ex- actly right amount of space to go between letters. Each type-letter, wherever it goes, carries along with it two _fixed_ blank spaces, one on each side. And of course, each one of the 26 is likely to be placed alongside any one of the other 25 with _their_ fixed blank spaces. So the odds against you in the fitting game would seem to be 2704 to 1. (Would it be that, or 2500 to 1?) But it isnt quite so bad -- the letter-shapes oc- cur in groups of similars: when you have solved for _n_ alongside of _n_ you are close to a workout for _h i j l m_ and for the stem sides of _b d k p q_ -- a prop- er fitting for _o_ gives you a line on the round shapes, etc., etc. _a, c, e,_ on their open sides, and _f g r t_ are hard to fit..... Griffith steps in here, with his experience, and takes a first crack: establishes the `side-bearings' and sets up a trial page. If the result is not satisfac- tory you go on from there by experiment. Usually he makes it in one. There isn't any fitting formula worked out yet. G.~ says there can't be any: that it is a job for the eye alone. I have a hunch that a `course' formula could be worked out; because there is certainly a `right' in- terval for a given weight and height of stem, vary- ing as these dimensions vary. To find out and establish these right intervals of emptiness between occupied re- gions is one of the prime jobs of design -- `voids & masses.' WAD This is the third publication by the Department of Printing and Graphic Arts in the Harvard College Library Cambridge HCL