Family planning, or how type families work

Essays by Peter Biľak
2 746 words14 min read
russian dolls

The size and complexity of recently-developed type families has reached unprecedented levels. Look, for instance, at United, a recent release (2007) from House Industries. The family includes 105 fonts composed of three styles (sans, serif and italic), available in seven weights and five widths. It takes a couple of minutes just to scroll through all the variants listed in the font menu. For a further example of this trend, Hoefler & Frere-Jones have just released their Chronicle type family (2002-2007), the range of which extends through widths (from regular to compressed), weights (from extra light to black), and optical size (from text to headline). In terms of sheer size, Chronicle comprises 106 fonts and beats the rival United by a single stylistic variant.

united font family

United, type family of 105 fonts designed by Tal Leming, published by House Industries in 2007.

Of course these ‘superfamilies’ benefit from the inventions of the past centuries; an ongoing series of typographic innovations that broke new ground for generations of designers to come.

History as a continuous series of discoveries

Ever since the earliest use of movable metal type; certain typefaces have included versions cut for specific point sizes. Claude Garamond's type from the 1530s (also known as the caractères de l'Université), included 15 versions ranging in size from 6 to 36 points. Each size was drawn, cut and cast separately; characters were designed specifically for the optical appearance of the printed text, with optimised letter widths and contrasts between the thick and thin parts of the letterforms. When photographically scaled to the same size, it is easy to see significant differences between the different designed sizes. Earlier typographers would therefore choose various sizes, just as we might choose various weights of a particular typeface today.

garamont 6 36pt

Garamond’s caracterès de l’Université from the 1530s includes 15 optical versions ranging from 6 to 36 points. Above are 7pt and 36pt type at 100%. Below is 7pt sample scaled 425% to match the 36pt version. Note the difference in contrast between the thick and thin strokes, and overall differences in details between the two versions.

In the age of the enlightenment, there was a clear need to organise and rationalise these differing sizes of printing types. In 1737, Pierre Simon Fournier published a table of graded sizes of printing types, introducing the first-ever standardized system for producing and using type. Fournier related type size to the 'pouce' (a French version of the inch), and subdivided the ‘pouce’ to 72 'points'. This new system also became the adopted standard in the English-speaking world, and in 1742 Fournier published his Modèles de caractères de l'imprimerie, in which he further systematised the body sizes of printing types, and suggested names for the most commonly-used sizes. The first mention of types being organized into ‘families’ also originates with Fournier’s work.

Pierre Simon Fournier’s printed scale of his  point system, from Modèles de caractères de l’imprimerie, 1742.

Pierre Simon Fournier’s printed scale of his point system, from Modèles de caractères de l’imprimerie, 1742. Fournier was the first to introduce a standard of producing and using type, sugesting a typographic unit called ‘point’.

Subsequent technological discoveries perhaps allowed typographers to forget the great invention of optically-adjusted type sizes. Type produced by pantographic reproduction (scaling a master drawing to many different sizes), and the later technologies of photocomposition and digital type, allowed working from a single master design regardless of the size of the final application. Typefaces made between 1960s-1990s almost entirely ignored optical sizes because photocomposition allowed unprecedented possibilities of mathematical scaling. Optically-adjusted sizes for type designs made a minor comeback in the early 1990s, most notably in ITC Bodoni, featuring size-specific designs similar to those used by the types originator, Giambattista Bodoni. These included Bodoni Six, designed for small captions, Bodoni Twelve for text setting, and Bodoni Seventy Two for display use.

However, optical size is just one parameter which determines the appearance of a typeface. It seems that typefaces need to be linked by several other shared parameters in order to be seen as part of a coherent group or family. Another such parameter is the weight of the type. For about 400 years printers and publishers did well with a single weight of a typeface, using just the type size as the main means of semantic differentiation. Even complex documents, such as Samuel Johnson's dictionary (A Dictionary of the English Language, 1755), use only a single weight of type set in different sizes to show the hierarchical differences between keywords, definitions and descriptions.

The idea of varying the weight of a single typeface probably happened in the mid-19th century. Heavy typefaces did exist before that time, but they were generally seen on their own and not in relationship to the regular (text) weight. The commercial pressures of the industrial revolution inspired the creation of different weights of typefaces. The idea was simple; to differentiate one text from another, or to highlight a particular part of the text. There were plenty of opportunities to use different weights of type in a western market-driven economy in the 19th century. For example, the Besley and Company foundry’s Clarendon type (1842), is widely acknowledged as one of the first bold typefaces, but soon after its three-year copyright protection expired it was extensively imitated and pirated.

Clarendon and its clones, however, although they were clearly designed to be used next to a Roman (regular weight, or text typeface), had not yet established a systematic relationship between the various weights (or widths) of a family-based type design.

Not one but many
From the early 20th century it became standard practice to include several weights of a typeface to complement the release of new type designs. The best example of this may be the work of Morris Fuller Benton, who complemented the many typefaces he designed for American Type Founders (ATF) with both condensed and heavy versions. Technology and aesthetics worked hand-in-hand for Benton, who used his fathers recently-invented pantographic engraving machine (1886), capable not only of scaling a single typeface design to a variety of sizes, but could also condense, extend, and slant the design. These fundamental geometric operations are the same basic transformations that most digital typographic systems use today.


In the later part of the 20th century, the work of Adrian Frutiger uniquely shifted attention from the design of a single typeface to the design of a complete typeface system, seeing the design of a type family as a continuous space defined by two axes; width and weight. The Deberny & Peignot Foundry released Frutiger’s masterpiece, Univers, in 1957 in an unprecedented 21 variants. Frutiger’s systematic approach and innovative naming scheme eliminated confusion in type specification, and was perhaps even more interesting that the actual typeface design itself. He created a novel system of double digit numerically-referenced styles, where the initial number 5 refers to the basic (roman or text) weight, and the subsequent number refers to the width (5 being standard or normal). Higher numbers signified increasing weight or width, so while Univers Regular was Univers 55, Univers Bold was referred to as Univers 75, and Univers Regular Condensed was Univers 57. The Univers system anticipated 9 weights and 9 widths (also incorporating an oblique, or sans serif italic variant), although some combinations of these proved unworkable in practice, so there is no Univers 79, or Black Condensed. Linotype further expanded the Univers family in 1997 to 63 versions – for this the numbering system was extended to three digits to reflect the large number of variants in the family. Frutiger originally envisioned this system to be used with other typeface families, however his systematic numbering convention never gained wider acceptance with either the foundries or his contemporary type designers.


The Univers type family, designed by Adrian Frutiger in 1957, consisting of 21 typefaces. Rather than focusing on a single typeface, Frutiger developed an interrelated typeface system.

The incorporation of two different styles of typeface into one family was probably first explored in 1932, by Jan van Krimpen in his Romulus project. Van Krimpen's intention was to create a large family of types for book printing; these would comprise a roman, an italic, a script type, bold and condensed types, at least four weights of sans serif, Greek text type, and possibly more. This was deliberately more ambitious than the type family of Lucian Bernhard, who released his types (Bernhard Gothic, Kingsley ATF, 1930) two years earlier.

Romulus, designed by Jan van Krimpen in 1932

Romulus, designed by Jan van Krimpen in 1932, was one of the first type families which included Sans and Serif versions in range of weights. The complete package included, a slanted roman, a chancery italic (Cancelleresca Bastarda), and an infamous Greek.

The sans and serif forms of Romulus share the same construction principles, but the resulting letterforms of the two styles are quite different. Van Krimpen quotes the type historian John Dreyfus in his book On Designing and Devising Type 1957: 'The purpose of the Romulus family was to provide the basic necessities for book printing and by means of a series of related designs to make possible consistent, flexible ... style.' Interestingly, Van Krimpen attempted to separate the style of his roman type and to apply it to Greek script as well. Although Romulus Greek is a fallacy, as it misunderstands the translation of letterforms from Latin to Greek, the method which Van Krimpen suggested is successfully used in localising most of non-Latin type today. When type design is understood as a system, it can be seen to consist of many shared parameters amongst letterforms from even such different origins as Greek and Latin.

Parametric design
A radically new view to understanding type families was paradoxically not proposed by any designer but by a mathematician. In 1977, Donald Knuth conceived a programming language that he called Metafont, which defined the shapes of letterforms with powerful geometric equations. Rather than describing the outlines of glyphs (like the later PostScript and TrueType font formats), Metafont describes an imaginary 'pen' that creates the stroke paths for constructing letterforms. Because of this unique approach, one can change a single input parameter for a typeface, such as optical size, angle of slant, or size of serif, and produce a consistent change throughout the entire font. A single font file can thus be a complex type family with many different versions. Metafont can control over 70 different parameters, which theoretically can define the appearance of any typeface designed with it. Even despite the obvious advantages of the system, and Knuth's close collaboration with the celebrated type designer Herman Zapf, Metafont never became widely used. Later technologies such as Apple's GX and Adobe's MultipleMaster font formats were similarly ill-fated.

The Hague school
Gerrit Noordzij, who taught writing and type design at the Royal Academy of Arts in The Hague for 30 years, outlined and developed his theory of writing in several books. His model presents the serif style as high-contrast type, as distinct from the sans serif style as low-contrast type, and arranges them in a coherent model of typographic possibilities. Therefore, instead of ideological discussions of ‘serif’ vs. ‘sans’, Noordzij focuses on the influence of tools while making marks on a surface. Noordzij describes three ways of producing typefaces: translation, expansion and rotation, each referring to different processes and the resulting stylistic differences between various groups of typefaces.

noordzij cube

Published in 1985 in the book, The stroke Dutch typographer and teacher Gerrit Noordzij proposed a theory of writing regardless of the used tool. His diagrame illustrates the main argument of his theory of writing, presenting the concept of translation, expansion and rotation.


Noordzij's pragmatic theories were highly influential amongst a group of designers that studied at the Academy. One of them was Lucas de Groot, who designed Thesis, (1994-99) a typeface family with three constructional variants of the type (sans serif, serif, and mix), comprised of 8 weights and totalling 144 variants. This type ‘superfamily’ was later further expanded by the addition of monospaced and condensed versions. De Groot developed and applied his own interpolation theory to the design of Thesis, which makes non-linear relationships between the weights of the type design. Thesis, first released in 1994, was the largest type family created at the time.

Part of the contemporary program at the Royal Academy of Arts in The Hague is Type & Media, a postgraduate program focused on type design education. It is only natural that such a place should be at the forefront of of typographic experimentation, redefining what we understand by the terms ‘typeface’ and ‘type family’.

Gustavo Ferreira, a recent graduate of Type & Media produced Elementar (2003-2006), a comprehensive system of pixel fonts generated by a series of Python scripts. Elementar draws its inspiration from Metafont and Univers rather than existing bitmap fonts. It is a parametric system responding to selected input criteria; a basic design for a simplified pixel typeface serves as a model, on which other parameters are applied. Because of the limitations of rendering glyphs on screen in small sizes, glyphs are expressed in terms of exact fractions or multiples of the model design.

Gustavo Ferreira’s system of bitmap fonts Elementar

Gustavo Ferreira’s system of bitmap fonts Elementar includes over 500 fonts, so a special application is necessary to select the fonts based on the user’s input criteria.

Such large typeface systems can become quite impractical to use, as the list of stylistic variations in the font menu gets larger and larger. In the case of Elementar, it is over 500 individual bitmap fonts, so an alternative solution has to be offered to the user to select the correct variant. Rather than presenting the full list of typeface possibilities this way, Elementar comes with its own online interface, whereby the user can choose the parameters, and get the right stylistic variant(s).

Kalliculator was a Type & Media graduation project of Frederik Berlaen (2006). Instead of drawing a typeface, Berlaen made a tool that makes typefaces based on a predefined set of parameters. Similarly to Knuth's tools from the 1970s, Kalliculator simulates pens and their relationships to a drawn stroke. Berlaen’s project uses Noordzij's theories as a base and the Kalliculator electronic pen ranges between pointed and broad nib styles. Users can input a line drawing, and the programme calculates the contrast around the skeleton, mixing the mathematical middle of a stroke and and a path made by an imaginary pen. The idea is that the trajectory of the hand is separate from the style of the pen, so users can experiment by applying various parameters to their sketched strokes. A single drawing of an 'a', can result in hundreds of versions, with each one being directly linked to others via its source drawing. In this way, Berlaen's application challenges traditional views of type families, as a typeface generated from the same skeleton is related to its family variants in a uniform manner.

80 kalliculator 1

Kalliculator is not a typeface but a tool that makes typefaces based on predefined set of parameters. User can input a line drawing, and the programme simulates either broad nib or pointed pen (or anything in between), controls the weight and contrast, applies the same parameters to all the glyphs in the database, and finally generates the font file.

So what exactly defines a type family? An analogy with a real physical family is not often helpful because unlike the biological world, different generations of typefaces are usually not considered to be part of the same family.

Similarly, at the level of individual glyphs, each style of the type family must be recognizably different in order to remain functional. Yet each style must adhere to common principles governing the consistency of the type family. It is clear that individual members of the family need to share one or more attributes, and typographic history offers many examples of this; optical size, weight, width, stylistic differences (sans, serif and semi-serif), construction differences (formal and informal), are the most common parameters linking members of type families. We can also find less common relationships such as varying serif types, changing proportions of x-height, ascenders and descenders, or contextually-appropriate possibilities of different versions.

Work by designers like Berlaen and Ferreira build on centuries of typographic innovation and help to explore new territory for type design. They participate in a cumulative, ongoing and inspirational history of type development, requiring that we continue creating this work in progress.