Ksenya Samarskaya

The Proof of the Pudding: A Case Study on Reviewing Type, Featuring Noto’s Cyrillics

Published  12/19/2021

What makes a typeface “good”? In a detailed case study starring Noto’s range of Cyrillics, Ksenya Samarska guides us through Samarskaya & Partners’ type review process. It’s a nuanced evaluative exercise that mobilizes mind, memory, and body: combining what one has learned and experienced with the natural pull of the hand.

At Samarskaya & Partners1 we’re often asked to critique existing work, review in-progress type, or research extended glyph ranges. Some of our takes end up being straightforward. Others are subtler—where there’s both a correct answer as well as a shadow of our personalities. On occasion, we find ourselves arguing against advice someone has received from another expert on a past project, which you could compare to reusing fashion advice from decades ago, or suited to a different occasion. Styles and connotations tend to be ever shifting, and even with type, the speed of that shift escalates in today’s hyperconnected, discursive, digital sphere. (Remember semi-serifs?)

The range of what we’re asked to comment on can be vast. Some glyphs we’ve grown up with or are used to seeing every day, whereas others we’ve only ever encountered in type projects, never naturally in the wild. So, how does one go about giving advice on something that one is both an expert and a novice in? With both a leading hand, and a grain of salt? And how can you work to refine your eye, and confidence, when giving advice on such broad-ranging topics? In this article, I’ll talk you through Samarskaya & Partners’ process, using a recent review of Noto’s Cyrillic range as a case study.

The More You Know, the More You Know

Reviewing type well comes from a combination of marshaling past knowledge, seeing the type in context, spotting patterns, and bringing in the right tools, the right juxtapositions. Things that help me include understanding handwriting gestures and directions, and practicing writing the glyphs myself—we’re not talking about beautiful calligraphy here, just an understanding of where the hand naturally pulls. This implies a habit of continually expanding and adding to a wealth of diverse and authentic sources, and, for me, an ability to twist the hierarchies traditionally created within them.

That practice is regularly supplemented through:

print sources: libraries and vintage and antiquarian book collections;

the internets, which are fast absorbing the former and making them available regardless of geographical location (online lettering archives, digitized libraries from around the world, photo collection hubs);

other type experts: conferences, digital forums and groups in various languages, respected existing typefaces, conversations; and

• travel memories and notations, some of which I’ve either written or talked about.

    Tools and Applications

    Here we’re in the first scene of Amélie, where we’re taking everything out of our handbag or cabinet and looking it over. For proofing or reviewing fonts, I’ll have in front of me the following: a long table if there are print proofs or physical specimens; a standing desk if it’s digital proofing (in order to be able to move and look at the type from different angles); on that desk a computer; and, of course, the set of font files I’m reviewing.

    I’ll need a Unicode spec against which to quickly compare or pull the glyphs to review. My current favorite is Unicode Character Table because it lays the blocks out nicely by use categories. And Wikipedia is still the best I’ve found for spotting the outliers hiding in different Unicode blocks. (For drawing my Latins, I have always leaned on Underware’s research and glyph spec, and Hyperglot is promising for certain reviews.)

    I use Galvanized Jets (manufactured in 2013–2014 with Eric Jacobsen, sped up in 2020 with the help of Simon Cozens) for proofing my own work, or quickly and efficiently spot-checking others’. (The site, like this metaphorical cabinet, is currently mid-sprawl with different tools and bits in different places; if you’d like to fund its cohesion and new feature integration, please get in touch.) Galvanized Jets was designed as a kerning proof and mega-word list, pulling together a long scroll of real-world lexicon examples. In practical terms, it tends to be the fastest way to see all the different contextualization a font can exist in, whether reviewing drawing or kerning, all in one go. (In 2020, Hoefler & Co. published their static word proof; Galvanized Jets is akin to that, paired with an algorithmic backend. Logic and evolution, with an ability to extend far beyond just the English-speaking world.)

    Finally, I use last century’s standby, InDesign, for manually laying out tests that aren’t yet in place as part of Galvanized Jets, such as those static word lists flying around and shared among type designers.

    Screenshot of the Galvanized Jets website showing a menu of options and text typeset in Noto Cyrillic.
    Samarskaya & Partners created Galvanized Jets, a typographic proof site, in 2013.

    Color and Rhythm

    I’ll write this specific case study of a review as a gradient: we’ll start with the most widely used glyphs, and those I’m most familiar with; then navigate through the ranges where there’s some documentation or arguments available; and end up in the obscure dusty corners of the Unicode, where we’ll discuss how to judge something with little to no direct quality sources.

    Similarly, we’ll zoom in and out, noticing color and texture, and the detail culprits responsible.

    Basic Russian and Localized Letters

    What we first looked for was the texture, or rhythm, of Noto’s Cyrillic as it compared to the Latin. What we found was a too-wide lowercase almost across the board. Character-wise, we spotted a lot of the more common pitfalls with Cyrillic. Some glyphs followed geometric precision and forgot about optical adjustments: the uppercase bowls of the БЪЫЬ’s were taken directly from the B, not allowing their needed extra room to breathe and balance. The “combination” glyphs of ШЩЮ got too wide. Other errors involved a lack of cultural understanding of glyph movements: the Д and Л were unsure about the needed slope and curvature on their left side, the lowercase italic б wobbled in the wave of its arm. Similar confusion was present in the gesture of the У, or the way the curve in the Ч meets the stem. Another common pitfall surfaced in the descenders of glyphs such as ДЦЩ, with Noto making them too long, too close to the body, and with odd joints in the serif styles. The lowercase built on the issues first spotted in the uppercase. Lowercase м, н, and в weren’t sure of their needed width, instead scaling from the uppercase too directly. A handful of the design flaws came from not knowing their place amid the type of typeface Noto wanted to be, with some glyphs getting too humanist (л), others using archaic construction styles (ж), or the breve lacking drops that would otherwise be appropriate here.

    The mistakes can be glossed or summarized as showing insecurity in their forms. Drawn with an amateur hand, they’re too newly built. They seem raw and unused. Whereas out in the real world, the letter shapes have evolved from their initial and somewhat awkward foundations. They’ve shifted over time, adapting their shape to accommodate being drawn, simultaneously losing their initial logic and gaining greater harmony.

    Beyond the Unicode spec, certain languages demand their own localized forms. For now, the ones that are well documented include Bosnian, Bulgarian, Macedonian, Montenegrin, and Serbian, as well as some existing notes regarding Bashkir and Chuvash. Bulgarian has its own forms for ДИЙЛФ вгджзийѝклнптфчцшщъью, whereas Serbian, Macedonian, Bosnian, and Montenegrin draw their own versions of the Б бгѓднтптшщ. Thankfully, the local lobbies for those countries have done well, and some great tutorials online cover this extensively. Other languages haven’t spoken out as much, but I’m hoping they will. The geographical sprawl of Cyrillic is vast, and often without a lot of direct, bidirectional communication with the local regions. I would imagine wherever the glyphs started, most would have picked up a local tic or two by now.

    Stacked letters displaying what they look like in “Soviet” Russian alongside their alternate regional forms.
    Bulgarian, Serbian, and Macedonian are the most commonly advocated-for localizations.

    Extended Cyrillic

    In Noto, extended Cyrillic suffered all the same issues of Basic Russian, as many of the forms in this range are adapted directly from the earlier Unicode ranges. For the additional shapes, the Cyrillic palochka (“stick”) picked up an odd serif in the sans style, and didn’t follow its customary proportions. Crossbars in glyphs such as the ҞҜҎҸҟҝҏћђ and ҹ were shoved in with no consideration of the composite design, resulting in dark splotches and/or making the strokes oddly thin and out of harmony with the base typeface decisions, causing it to feel like an afterthought instead of a considered letter. The Serbian ЂЋђћ’s lacked internal cohesion, forcing the shapes toward awkward tension in their curves. The Ҩҩ’s had a slanted relationship to the baseline while closing up too much in the loop. Some letters were unsure whether they should be sloped or italic, with capital Ҽ taking on the lowercase italic curvature while the rest of the uppercase chose the oblique route. Forms that I would expect to show consistency, such as descenders, took on wildly different shapes depending on which glyph they appeared in, as if they didn’t know they were supposed to be part of the same stroke. The ҊӅӉӍҋӆӊ and ӎ glyphs picked up an odd bell-bottom descender flaring not seen elsewhere. Another set I would expect to be part of the conversation, but was not, includes the ЂҦӃӇӺӼԒԠԢԨꚊђҧӄӈӻӽԓԡԣԩ and ꚋ glyphs. Elsewhere, there was a single version of the Д that took on the historical pointed form within a ligature, a possibly justifiable diversion had it not appeared just that one time in the entire font.

    Between the issues brought forward from the base glyphs and the odd way strokes or accents were added, we found something to object to in almost every glyph we reviewed (we are rather devout seekers). A few glyphs, ones directly copied over from the Latin, were spared our critical eye: ӒІЁЈЅӦӪӓіёјѕӧ and ӫ. Even some that looked like they could pass with the copy and paste still got marked up, though, such as the Ukrainian Її’s, which tend to prefer tighter dieresis spacing than their Latin counterparts.

    Wall with large extruded text. On the left the word is in Cyrillic with ascenders for the letter ы, and on the right side the word is in Old Permic.
    Cyrillic and Old Permic as seen on a monument in Syktyvkar, the capital of the Komi Republic. Image courtesy of @Incunabula.

    Rare Languages

    A few glyphs from the Extended Cyrillic set I tagged for further research, including the use of қң/ӄӈ in Koryak, the з̌/ҙ in Nganasan, and the descender shape of the Ҫҫ’s as used in Enets. This moves us through the Unicode right along to the rather rare and obscure languages. (According to Wikipedia, Koryak is currently spoken by 1,700 people, Nganasan by 125, and Enets by around forty, with the written language for it developed only in the 1980s.)

    I dove into research, attempting to find in-use samples of Nivkh (spoken by somewhere between two hundred and a thousand people), and ended up unearthing a trove of printed books written in Komi’s Molodtsov alphabet (an alphabet officially in use for only about twelve years a century ago).

    The final sixteen glyphs that round out the Cyrillic Supplement section are additions for Aleut, Abkhaz, Chukchi, Itelmen, Juhuri, Khanty, Kurdish, Mordovin, Orok, Tati, Yagnobi, Yukaghir, and old versions of Chuvash, Komi, Mokshan, and Ossetic/Ossetian. These are mostly ligatures, so I judged them by marking the needed changes in their various components, plus any generalized optical adjustments. If I were to draw these shapes myself, though, I would want to do supplemental research.

    A selection of alphabets from old books with Roman and Cursive forms in both uppercase and lowercase.
    Various Cyrillic alphabets found online. From left to right: Komi, Abaza, and Abkhaz. Images: Wikipedia.

    Historical Glyphs

    Time-traveling backward past mid-twentieth-century experiments and rarely written languages, we get to Old Cyrillic / Church Slavonic. Trying to decide how to do a sans or serif style of Old Cyrillic is like trying to follow the strokes of a blackletter face while assuming there’s a correct sans/serif variation. It’s all heresy and imagination, but, while you’re there, you can at least find things not to do.

    Given that a good amount of contemporary Cyrillic evolved from Old Cyrillic, there’s something near comical about going back in and recapturing and modernizing these alphabets. But comical doesn’t have to mean bad, or without interest. Perhaps, for all you Latin readers out there, we can use Commercial Type’s rendition of Marian Text Black as a visual parallel. Now, imagine taking those forms yet again, and adding in serifs, styles, a monospaced counterpart. For…time-traveling and very confused programmers to code with? A new set of typewriters? And yet the more I mock, the more I’m fascinated and tempted to dive in and redraw these shapes toward a fitting solution.

    A contemporary monoline typeface whose letterforms take the shape of historic blackletter, showcasing an unusual contrast.
    Commercial Type’s Marian Text Black recreates blackletter as a thin, contemporary monoline.

    In my notes reviewing Noto’s Old Cyrillic, I marked for consistency within the chosen direction. Glyphs that share components should have the same base form. Weight should be consistent. If a decision has been made to go with italic elsewhere (as with most global scripts, italic also didn’t exist in these historical manuscripts), we should continue by italicizing all the glyphs. Some of my comments concern drawing the glyphs with enough open space and balance to make them distinguishable (not going to use the term legible here) at small sizes. The stroke flaring should be consistent and match the initial typeface decisions, following the same concept of what “tool” was used to make these marks. There needs to be an understanding of where glyphs start and end, to (at the very least) place serifs in the proper locations. While I’ve never seen any typeface do this correctly, there’s also knowledge about how numerics were marked, which was by placing a stylized circle around another glyph—so to be used properly, the sizing needs to accommodate such overlaps. The accents should be distinguishable at small sizes, and follow the approximate black area they take up in agreement with other accents in the font. If it’s a serif style, there should be a consistency within the serifs, whereas in Noto I found a combination of some serifs, some half-serifs, some italic flaring, and some blunt sans ends, making for a very inconsistent set in an otherwise restrained typeface.

    Seven images showing how to write 1,000, 10,000, 100,000, 1,000,000, 10,000,000, 100,000,000, and 1,000,000,000. In each the inner letter is an ‘а’; the outside form changes and includes a closed circle, a circle made out of dots, a circle made out of commas, a circle made out of x’s, or different lines around the ‘а’.
    Old Cyrillic Numerals. The interior letter designates the first number, and the markings around it the powers of 10 that follow. Images: Wikipedia.

    Typography courses often have an assignment where the student gets to dream up a new alphabet: a unique set of marks, where the challenge is to give the set an internal logic. The new alphabet needs to be a coherent system, even if it doesn’t translate to anything in the outside world. So here too, at the very least, I would want Noto’s Old Cyrillic to pass that muster. And, if I had more time, I would love a chance to go deeper and notate this to a higher level. For that, original sources can be scoured to better understand normative proportions. There might be a way to find unique solutions for complex glyphs; practicing writing the style enough to internalize and understand its rhythm and desired proportionality would help guide design decisions. Perhaps that’s all a story for another time.

    The Proof of the Pudding

    There you have it, a how-to without the how. (Should we make the next part of this a Twitch channel? A fourteen-year course?)

    For Noto Cyrillic, it’s well advised to redraw a good deal of it, starting from the foundational and most frequently used glyphs and working down the path of this essay. The fixes themselves are relatively small, but the glyph set and family itself is large and sprawling. Checking for consistency should be a priority. Creating tools to help get the work there, and to no longer rely on last century’s static proofs, is also advisable. Our type is only ever going to be as good as our ability to review it; if we can't see our work clearly, it's going to be difficult to spot any problems.

    For those reading this in order to guide their own drawing and proofing work, my advice is to do all that measure-twice work as you go. A good foundation leads to lighter work down the road. Create chapters or checkpoints for yourself. If you have time, build tools to help you. If I have time, I’ll resurrect Galvanized Jets and can extend it to assist. Ask, and pay for, outside opinions as you go. Find native speakers. Find native speakers with contrasting opinions. Understand that the same solutions you’ve made for one typeface won’t always apply to the new one (ditto for prior advice you’ve received).

    With reviews, just like the typefaces themselves, another vital question is: How do you know when you’re done? The end can so easily circle back to a beginning, to more research, to deeper nuance. As with any project, you can’t score without a goal, so writing down the brief for the typeface early can help keep you centered. Noto’s mission of creating “harmonious, aesthetic, and typographically correct global communication for all the scripts in the Unicode standard” is laudable. The Cyrillic makes a solid start by filling the slots, but has a long way to go toward reaching the aesthetic and typographical correctness it strives for. And whether my various notes and appendix have found all there is to notate? I’ll leave that thought with a quote from Nicole Sealey: “A work is never finished, only abandoned.”2 And for right now, I’ll leave it here.


    If you’ve reached a natural stopping point in your own work, consider testing its resolve by entering type competitions with diverse and expert judges. TDC’s Type Design competition this year features a panel of native experts in scripts covering Latin, Cyrillic, Greek, Arabic, Chinese, Japanese, Korean, Tamil, Bengali, and Devanagari; and it’s still open for submissions. The final entry deadline is January 21, 2022.

    • 1—

      Reviews for the Noto styles were conducted via conversations between Ksenya Samarskaya, Micha Strukov, and Azamat Kodzoev.

    • Ksenya Samarskaya is a strategic consultant and creative director, passionate about deep research at the cross-roads of design and our lived realities. Clients and collaborators have included Adobe, Apple, Google, IDEO, Intel, Monotype, Rosetta, Snoop Dogg, WeWork, YouTube, and others. She has previously judged typographic competitions for ADC/The One Club, Communication Arts, TISDC, Type Directors Club, the SoTA Catalyst Award, and served on the board of AIGA/NY. Samarskaya currently serves as the United States Delegate for ATypI, and is this year’s cochair of the type competition for TDC.

    At Typegeist, we invite authors to write from their individual perspectives. Opinions expressed in the articles on this site are those of the respective authors, and do not reflect the views of the Typographic Design Center. Quill icon by Ariel Kotzer from the Noun Project.