Download font

Designing Eastern Arabic MICR Numerals

Vintage tech is fascinating. Often, its underlying machinery is visible to the naked eye, allowing us to understand opaque abstractions instinctively. Within every man-made object is inspiration, and a narrative diligently instructive of why things can become so complex. Technology is seemingly less arbitrary, less magic-like, when we trace its history and circumstances.

Naturally, technology and design are coupled in some inextricable, amorphous, yet curious ways. Each informs the other, but design is what most visibly reflects the limitations of our tools, materials, processes, tastes, and objectives. Bound by technological constraints, iconic avenues of aesthetic emerge, as though by fate, in dignified defiance to domain standards of beauty.

“PS نوروز‎ 199”. Inside a 1970s Olivetti typewriter: daisy wheel featuring an extended Arabic alphabet for Persian markets. [source: Borna Izadpanah's personal collection.]

In type design, this phenomenon is clearly attested by Magnetic Ink Character Recognition (“MICR”) typefaces which were developed for the banking industry in the 1950s to facilitate automated processing of cheques. The material limitations of the time gave birth to typographic anomalies, the aesthetics of which are now deeply emblematic of computing throughout the 60s and 70s, legends of fraud, retrofuturism, and dystopian world imaginings.

The MICR Duo

The prevailing MICR typefaces are E13B and CMC7. We see these two frequently, some of us on a daily basis, on cheques and banknotes, but they almost never register on our minds as anything special. Some things just hide in plain sight.

Digits that are good for robots! ...and humans!

E13B, the subject of my experiment, was invented in the 1950s by SRI International (these are the same folks that brought you Siri!) as part of a project called ERMA. It is rather routinely named, an instance of revision history as brand; “E” is the fifth typeface proposed for the project if you start counting at “A,” 13 is from 0.013 inches, the elemental grid unit size, and “B” is the second revision of this design. CMC7, which is mainly used in Europe and Latin America, was designed in 1957 by French computer company then named Machines Bull. For a variety of reasons, it didn't have the same cultural impact of E13B.

Sample waveform readout. [source]

Traditionally, MICR fonts are printed in a special ink and thereafter magnetized with an electrical current. Each printed symbol generates a distinct waveform when passed under a magnetic-reading head. The system then analyzes the resulting signals and matches them against digits and routing instructions. The way the ink is distributed inside each symbol determines the waveform, so the design had to surrender to some reality of magnetism: similarly contrasted shapes could generate similar waveforms— something to be strictly avoided. For E13B, the talented designer(s) conceptualized an honest framework of geometric monolines and the occasional slab to achieve this dual-goal of efficient machine and human readability.

A Fount for the Space Age

With how prevasive and ancient MICR fonts are, I was slightly surprised when I couldn't find an Eastern Arabic version, especially since there were numerous attempts at creating alphabet complements in the style of E13B. Earliest of these is Moore Computer (1968) by Visual Graphics Corporation. Its design follows closely the configuration of the original E13B symbols. Notable as well is Bob Newman's Data Seventy (1970) which, while manifests the general mode of E13B, takes the liberty of imagining new digit shapes. Westminster (1971) by Leo Maggs is possibly the most prevalent spinoff, having been bundled as a digital font with Windows 98. There are many other efforts, and the chronology might be slightly off here, but the 3 mentioned are seminal and representative of the genre.

“2001: A Space Odyssey”, 1979. Hungarian movie poster featuring Data 70.
Magnavox Odyssey video game console, 1972. The packaging features the word “Odyssey” set in Moore Computer. [source]

“Style Transfer”

So I set out to design Eastern Arabic numerals in the aesthetic of E13B. One, because Arabic is my mother tongue. Two, because the font is used by financial institutions in Arabia. And three, why not? I think it's about time, and it's incredibly tempting to attempt this sort of thing.

A 5 dirham banknote from the United Arab Emirates. Maybe we can save a ton of ink by using a single representation for the serial number?

Are the red digits functional? A counterfeit prevention measure perhaps? Or do they merely compensate for a cultural pain?

To design the new digits, I decided to first reproduce the original 14 E13B glyphs—a reconnaissance of sorts—to familiarize myself with the typeface's spirit and evocations.

The good news is that E13B and CMC7 character designs are standardized in ISO 1004-1:2013. The specification largely targets printer and reader manufacturers with things like precise print tolerances, ink and paper considerations, and a bunch of other things. Bad news is that it costs $160 to access. Fortunately, with the power of the internet, you can find relevant bits & pieces of it in printer manuals and patent filings. It's also easy to reverse engineer the design language from printed cheques and manufacturer specimens.

Anatomy of E13B

ISO 1004-1:2013 outlines dimensions of the final printed output, so we need to translate those fractions of inches into convenient virtual units that we can use in our type design app. Here's that in type jargon:


Design Decisions

With the original E13B glyphs out of the way, we're ready for the creative part!

For now, we won't explore anything related to magnetism or the waveforms produced by the newly designed symbols. Ours is mostly an exercise in pastiche. I'm sure that with functional criteria these designs will change drastically. However, if you know of a way to digitally simulate the waveforms produced by these glyphs, please get in touch.

Here were some general guidelines I tried to stick to:

Introducing Nimra


Eastern Arabic zero ٠ is usually represented as a relatively small diamond-like blob in calligraphy.

Placing it on our square grid we find a middle ground where it's not so comically large and desperate to fit in, nor is too small and illegible.


One is traditionally drawn at an angle. My take builds on a raster of Naskh ١ at a tiny point size.

There are only so many ways you can represent such a simple shape.


There are 2 common shapes for two ٢ in Arabic: with and without teeth. While the toothed variant seems more dynamic, getting rid of the teeth signals an evolution, and Nimra is a font for the future! A thinner shape is also welcome since our numerals are overall darker than E13B.


Three ٣ adapts its fork configuration from E13B 3 and flips it sideways. Tooth heights were varied until they looked right.


Four ٤ was a tricky one. You would be tempted to merely flip the original 3 to create a passable resemblance. But Naskh four is horizontally darker, so I try to reflect that.

With this grid, the boundary between isolated 'ain ع and ٤ is increasingly blurred.


There are so many ways to approach five ٥. It's an organic shape with exciting thick-thin transitions. It can be modeled after E13B 0 with an added slab for differentiation, or it can be made into a relatively thick rectangle that's thin on one side. I went for an abrupt look that echoes the brutalist modulation in E13B's 1 and the symmetry of 8.


Six ٦ is simple. I stress the horizontal and flourish the stem with a characteristic spur to channel calligraphic slant, and partially to keep our overall rhythm from getting stale.


Seven ٧ is curious. I thought to shape it like an uppercase Latin Y, but that looked like a chicken bone to me! The diagonals are instead abstracted into what you see.


Eight ٨ is a laterally and vertically inverted seven with a thicker main stem for variety— this also implicitly announces that eight is greater than seven!

Looks like a gate drawn at perspective.


Last but not least, our nine ٩ is unabashedly borrowed from E13B 9, but with a slight change. The lower horizotnal stem is thickened to stay faithful to script, and to make it distinct.

All together now:


Yikes! That won't win any awards for sure, but then again beauty is not a criterion.


In a world entrenched in legacy software and hardware, and with OCR being amazing as it is, the uses for Nimra are limited. At best, it's a display typeface for some retrofuturistic effect, and even that is far-fetched when there's no widespread adoption or a collective nostalgia. But at least we can imagine a world where Eastern Arabic and Arabic numerals co-existed in E13B, and it might have looked something like this:

Download Nimra

Download Nimra 1.02 [.ZIP].
Nimra is released under the SIL Open Font License (OFL).
Source code is available on Github.

In terms of difficulty, the design isn't really challenging. You can check out the source files to get some feel for type design...

So how would you design your choice of numerals in the style of E13B? What would you do differently?

Will you use Nimra in your project? It's always delightful to see my work being used out there. Get in touch and I will list it here.

Derived Works

Further dimensions to explore

Additional Reading and Resources