The Development of Balto

The journey to releasing Balto was a long one. I can trace the germ of the idea back to my first official graphic design job. At that firm, I did a lot of corporate design and every time I would try to use one of the classic gothic typefaces, the bank/hospital/insurance company client or the account executive for the client would tell me something to the effect of, and it hurts me to say this, “We need to use a font that isn’t ugly.” Around the same time, my art director was looking for a clean, organized gothic that could be used for all text in an annual report. Our search found nothing. This was 1997 or 1998. This wasn’t an “Aha!” moment or anything like that. Rather, it was just a typographic problem that lodged itself firmly in the back of my head. Other thoughts were added to this one over the years until they snowballed into the foundation of Balto.

I usually draw lots of pencil sketches when I am thinking about a new typeface. This helps me refine the ideas without getting too deep into visual details. Balto is weird in that I can’t find many of these in my sketchbooks. There are drawings of random letters that clearly point toward Balto, like this g that I drew over and over and over:

Lots of sketches of the letter g.

I have stuff like this, but not my usual iterative, detailed, rambling note filled pages. I think this is because this typeface was in my mind for so long that I just sort of knew what it was supposed to look like long before I placed the first point in my font editor. In fact, I recently looked back at the very first digital file, dated February 1, 2007, and I was startled to see how close it was to the final version of the corresponding weight in Balto. Here is a comparison of the first (black) and final (red).

The first sketch versus the final Balto Book.

Sure, the x-height changed, the spacing changed, the widths changed, the weight changed and the g changed, but overall? It’s pretty close. After looking at this, I started wondering: Why did it take me over six years to finish this!? To answer that question, I ventured deep into the depths of my hard drive…

Development Process

First, a little overview about my general development process. I create typefaces in a series of overlapping and sequential phases.

My development process,

The idea is the first phase, obviously. When I design things for clients, the ideas always come pretty quickly. With my self-initiated work, it can take a very long time for the ideas to mature. I wish I knew why it was that way, but it is what it is. I don’t start drawing until I have some sense of what I’m drawing and why I am drawing it. Once I feel like I have a good idea, I start drawing in a font editor. This phase always takes the longest. I break the drawing into stages that define the order in which letters, numbers and symbols are drawn. For example, my first stage contains a, b, e, f, g, h, i, m, n, o, r, s, t, u and v in upper and lowercase. The second stage contains A-Z, a-z, 0-9 and basic punctuation. The last stage contains… a lot of stuff. I don’t have a set order for drawing the weights or styles. I usually start with what I think is going to be the most important style in the family and work outward from there. Once the drawing stages are all complete, I kern the fonts. While I’m doing this, I will inevitably notice a few drawing issues that need to be resolved. Once all of those issues are taken care of, the fonts are hinted. Drawing changes after hinting has begun are complex for a variety or reasons so I always have my drawing completely finalized before hinting starts. Once the kerning and hinting are done, I master the fonts. In other words, I make the final OpenType, WOFF, etc. files. I am quite persnickety about the quality of my files so this takes more time than you would think. Anyway, the drawing phase is the most important and where most of my time is spent.

While drawing, I use a simple file name based versioning system. Actually, calling it a versioning system makes it sound more impressive than it actually is. Whenever I am going to make a batch of changes to something, I duplicate the previous file and add a date tag to the file name to indicate the day that the file was started. That file may see a week or more of work before I duplicate it and start the process over. This isn’t as precise as a real version control system like Subversion or Git, but it’s easy and it works for me.

Six Years, Illustrated

So, back to what I was wondering before that digression. Why did it take me over six years to finish Balto? I know that it was the design phase that took the vast majority of the time. But, I like science and detail, so I wanted more definitive answers. I wrote a little tool that would analyze the date tags on all of the files and examine the contents to give me a report of exactly what I was working on, when I was working on it and what was different from the previous version. Data mining is fun!

As I mentioned, the drawing phase began on February 1, 2007. It ended at the close of January 2013. These are all the times that I created a new version of a style in the family.

Balto drawing timeline.

So, I didn’t work on Balto every single day for six solid years. There were gaps between periods of feverish work. The gaps represent (in this order, sort of): Torque, Ohm, the arrival of my first son, UFO 2, the great webfont battle, WOFF, UFO 3, Timonium, a typeface that I am hoping to finish in the next few months, the arrival of my second son and this website. Throughout, there was a lot of random code, fun client work and being a dad.

That’s when I was working on Balto, but what was I actually doing? How did each style develop over this time? There are a few factors to this: what was I working on, how big each font was and how much was I changing. The next chart gives me some clues. The gray indicates how many glyphs each style contained. At the widest, it represents 527 glyphs. The red bar indicates what percentage of these glyphs I changed in the particular version. Wider indicates a higher percentage.

Style by style timeline.

Firstly, whoa. Secondly, whoa, that’s a lot of changes to the Book style. Though, it makes sense. Book is the most important style in the family (you are looking at it right now) and it was the first style I started with. Still, that is a lot of changes considering how similar the first draft was to the final file.

Interpolation Interlude

You may be looking at that chart and thinking “Hey, he didn’t work on Light, Medium, Super or most of the italics very much!” That’s sort of true but there is a good reason. In the case of Light, Medium and Super, those styles are interpolated. Interpolation is a mathematical equation that type designers can use to create one or more instances between two masters. (Technically speaking, I use a fancy version of interpolation called Superpolation developed by Erik van Blokland.) This is a useful technique, but in a family with a wide weight range like Balto, it is only a starting point. For example, here are the official Balto weights (top) compared to a linear interpolation between Thin and Ultra (bottom).

Balto versus interpolation.

Yeah, the interpolation is not so great. In Balto, the Thin, Book, Black and Ultra serve as interpolation masters and the other styles are interpolated between those. Furthermore, I wasn’t completely happy with the interpolated results, so I more or less redrew the interpolations. That’s why you see the 100% change indicators at the end of the interpolated styles.

As for the italics, they were largely based on the romans. I didn’t start a particular weight of italic unless the corresponding roman was nearly done. By that point, I had figured out what the style was going to look like so there wasn’t much back and forth. In a few cases, I spent a week drawing them, looked at a proof, made changes and was on to the next style.

Book Breakdown

So, back to all those changes to the Book style. I made a visual report of each of the versions and it all came back to me in a flood of memories. Here, step by step and unabridged are all of the revisions with some notes about what I was thinking and doing. On the left are some important, frequently changed glyphs as they stood in each version and on the right are all of the glyphs that were changed.

Version 1

“And I’m off! This will be easy! And fast!” — Me, dumbly, February 1, 2007.

Versions 2-4

At some point in my thinking years, I decided that the transition from the arches to the stems should be smooth (left), rather than sharp (right).

Arch to stem comparison.

My theory was that this would allow the b, d, h, m, n, o, p, q and u. to have similar proportions and shapes. That, in turn, would give a more consistent rhythm and cleaner text. The opposite is almost always true, especially in serifed typefaces, but in this particular style it seemed like the right thing to do. I figured that I should double check that theory before I got too far along so I made a quick change and test. Yep, my gut was right. This change was immediately undone.

Versions 5-9

In almost every version up to this point I was experimenting with different ways to handle the terminals on the G, S, a, e and s. Here are all of the versions of the S so far.

S terminal variations.

Should they be open? Should they come all the way around and end flat or very near flat? I kept bringing them around more and more and I finally decided that they should end very near to flat. I declared this matter resolved and moved on. (Foreshadowing!)

Versions 10-17

Hey! New glyphs! The “I think this idea is working. I’m going to add more stuff.” point is always fun.

Versions 18-20

At this point, my friend Christian took a look at the whole family. He gave me some really great feedback and some very red proofs. Among other very insightful comments, he questioned my decision to bring the terminals all the way around. His point was that it seemed out of character with the style. He was right. So, I went back to trying to figure that out.

Versions 21-25

I opened the terminals back up. Then I opened them up as far as I could. That looked awful. Then I brought them back a little. Then some more. Then some more. Eventually I settled into something that I thought worked.

More S terminal variations.

These tiny things may seem like inconsequential details but they are very important. I teach type design and I like to tell my students that while these minuscule changes won’t be noticed by most people, they will be felt. Type design is a great example of the whole being greater than the sum of its parts. A small change like this will echoed by other glyphs and all of these can be multiplied hundreds of times across a single page. The details, as small as they may be, need to be right. This change was one of the most important decisions that I made in Balto. Once I had it right, the whole thing started to feel like a real thing.

Versions 26-36

Finally, everything started to settle into place and I added new stuff. I do love drawing these things that will most likely never be used…

Versions 37-38

…though, I do start to question my sanity when I draw things like the § that will only show up in all uppercase text in the Ultra weight. The odds of that ever being used are near zero, but it’s there! And it looks nice! (Someone please use it so that the hour that I spent on it was worth it…)

This is the point when I start to get really excited about the typeface. By now, the upper and lowercase are working well enough that I can actually type some text in it without thinking. “Ohhhhh nooooo. I need to fix that.” The feeling when this happens is one of the best things about making typefaces. It’s also a sign that it is almost done.

Versions 39-42

Accented glyphs. It has to have accented glyphs. Lots of accented glyphs.

Version 43

And, arrows. I love arrows. I thought they would be useful in such a utilitarian face, so I drew some. Then I tried to use them in an early version of the Balto page on this site

Version 44

…and I realized that the shape was wrong. So, I redrew them. These worked much better. I also added a heart and a star because they were fun to draw. I ended up removing these because if I am going to draw symbols, I want to add a bunch of symbols, not just a dinky heart and star.

Version 45

Then, it was done. Type design can pre pretty anti-climactic. It seems like after all of this, the glyphs should have staged some sort of epic final battle against me in which only they or I emerge victorious or something. But, alas, only ogonek adjustments, subscript and superscript. Out with a whimper. The end.

The other styles followed this same pattern, albeit with fewer versions. Once those were done, it was time to kern, hint, master and prep the website. Just another day (well, a lot of days) at the office.

Six Years, Summarized

Now I know why it took so long to finish Balto. Even though I had been thinking about it for years and had very solid ideas when I started drawing, I should have known that it wasn’t going to be a quick and easy typeface to finish. I didn’t set out to radically reinvent the wheel. Rather, I was trying to take the idea of something that has existed for a long time, the classic American Gothic, and reduce it down to its absolute essence. Each step along the way was an iterative refinement towards my idea of perfection. Perfection, even one that is self defined, is elusive. But, you know what? I loved drawing Balto and I would do it all again.


Because this page just isn’t quite long enough, I present to you all of the versions of some of the most changed glyphs in the family all stacked on top of each other.

Pretty pictures.

September 26, 2013