What a Grid Is Actually For

Josef Müller-Brockmann was sixty-seven when Rastersysteme für die visuelle Gestaltung was published by Niggli in 1981. It runs to 176 pages, German and English on the same spreads, and it is not a manifesto. The publisher describes it, accurately, as a practical handbook for designers in training or teaching, with step-by-step instructions for grids of eight to thirty-two fields and worked applications from a business card to an exhibition stand. Four decades later it is the book most often cited by people arguing that the grid is the house style of Swiss modernism.

That is a misreading, and the dates give it away. By 1981 Müller-Brockmann had been working this way for three decades. His first book, Gestaltungsprobleme des Grafikers, appeared in June 1961, 200 pages, 735 illustrations. Its middle section is not about grids at all. According to Niggli's own description it covers semiotics, communication research and information theory, under a chapter heading translated as "Science and Visual Communication". Two years after that book he was a guest lecturer at the Hochschule für Gestaltung Ulm, in the same period when Horst Rittel was teaching design methodology and operations research there.

The grid did not arrive as an aesthetic. It arrived as the visible residue of a methodological argument, and the book everyone quotes is the late, tidied-up teaching version of a position that had been worked out much earlier and stated more sharply by someone else.

The book that says it better

Karl Gerstner published Programme entwerfen with Arthur Niggli in Teufen in 1964, with German and English editions issued together. The English title is Designing Programmes, and its subtitle is the entire thesis: instead of solutions for problems, programmes for solutions.

Gerstner was thirty-four. He had founded the Basel agency Gerstner+Kutter with Markus Kutter in 1958 and expanded it in 1962 with Paul Gredinger into GGK. The first edition runs to 96 pages, set entirely in Akzidenz-Grotesk on a two-column grid, with an introduction by his partner Paul Gredinger followed by four essays previously published in periodicals: on modernising Akzidenz-Grotesk, on integral typography, on image-making, and on structure and movement. He would go on to design the Gerstner Programm typeface between 1964 and 1967 for Berthold's Diatype filmsetting system, sixteen cuts, eight roman and eight italic. In 1973 the Museum of Modern Art gave him a solo show, "designing programs/programming designs", directed by Emilio Ambasz and based on the book.

Here is what he actually argues, in his own words as reproduced from the 1964 text:

Instead of solutions for problems, programmes for solutions: the subtitle can also be understood in these terms: for no problem (so to speak) is there an absolute solution.

And, on grids specifically:

If the grid is considered as a proportional regulator, a system, it is a programme par excellence.

The operative sentence, the one I would put on a studio wall if I put things on walls, is his statement of the goal: the maximum of constants with the greatest possible variability. Gerstner also borrowed Fritz Zwicky's morphological method and built what he called "the morphological box of the typogram", a table with parameters down one side and the available components against each, so that designing a mark becomes a matter of choosing a cell from each row. His phrase for the creative process, in that framing, is that it is "to be reduced to an act of selection".

Reduced to an act of selection. That is the claim. Not that the grid makes things look a certain way, but that it converts an open-ended act into a finite choice.

Photograph by MarkBuckawicki, CC0, via Wikimedia Commons.

What Müller-Brockmann says he is doing

Müller-Brockmann's own account of the grid is more moralistic than Gerstner's and is usually quoted in fragments. The full passage lists what the use of the grid implies, and it is a list of wills:

the will to systematise, to clarify; the will to penetrate to the essentials, to concentrate; the will to cultivate objectivity instead of subjectivity; the will to rationalise the creative and technical production processes; the will to integrate elements of colour, form and material; the will to achieve architectural dominion over surface and space; the will to adopt a positive, forward-looking attitude. He frames grid-based work as submitting to laws of universal validity, and describes the designer's professional ethos as delivering the objective, functional and aesthetic quality of mathematical thinking as a contribution to general culture.

Two of those items are engineering claims and the rest are ethical ones. "The will to rationalise the creative and technical production processes" is the engineering claim, and it is the one that has survived best. "Submitting to laws of universal validity" is not a claim about layout at all. It is a claim about the designer's relationship to their own preferences, made by a man who had watched what unregulated national feeling had recently done to Europe. Read that way, the 1981 handbook is a much stranger and better book than its reputation.

A grid is a device for deferring decisions

Here is the version I would defend. A grid does not decide what a page looks like. It decides when you have to decide.

Without a grid, every element on every page raises a fresh question with an infinite answer space: where does this go, how wide is it, how far is it from the thing above it. With a grid, you answer those questions once, in the abstract, before you know what the content is, and thereafter each page asks you a multiple-choice question instead of an essay question.

That is a real and measurable change, and it is arithmetic. Take a plain specification and count what it permits.

GIVEN

  measure      960
  columns      12
  gutter       24
  column       = (960 - 11 * 24) / 12 = (960 - 264) / 12 = 58
  span width   = 58s + 24(s - 1) = 82s - 24
                 for a span of s columns
  baseline     8
  body         16 / 24    24 = 3 baselines
  h2           32 / 40    40 = 5 baselines

HORIZONTAL CONSEQUENCES

  distinct block widths           = 12
  distinct (start, span) places   = sum of (13 - s), s = 1 to 12
                                  = 12 + 11 + 10 + ... + 1
                                  = 78
  two-block splits of the measure = 11
  three-block splits              = C(11, 2) = 55

VERTICAL CONSEQUENCES

  every block height is a multiple of 8
  n lines of body text = 24n = 3n baselines
  a rule of "gaps are 3 or 5 baselines" leaves exactly two
  vertical gaps = {24, 40}

RESULT

  the question "where does this go" has become
  "which of 78 placements, and which of 2 gaps"
  78 x 2 = 156 permitted positions for one block
  a second designer, given the same brief and the same 156,
  will produce a page that sits beside the first one without
  a visible seam

GIVEN

  measure      960
  columns      12
  gutter       24
  column       = (960 - 11 * 24) / 12 = (960 - 264) / 12 = 58
  span width   = 58s + 24(s - 1) = 82s - 24
                 for a span of s columns
  baseline     8
  body         16 / 24    24 = 3 baselines
  h2           32 / 40    40 = 5 baselines

HORIZONTAL CONSEQUENCES

  distinct block widths           = 12
  distinct (start, span) places   = sum of (13 - s), s = 1 to 12
                                  = 12 + 11 + 10 + ... + 1
                                  = 78
  two-block splits of the measure = 11
  three-block splits              = C(11, 2) = 55

VERTICAL CONSEQUENCES

  every block height is a multiple of 8
  n lines of body text = 24n = 3n baselines
  a rule of "gaps are 3 or 5 baselines" leaves exactly two
  vertical gaps = {24, 40}

RESULT

  the question "where does this go" has become
  "which of 78 placements, and which of 2 gaps"
  78 x 2 = 156 permitted positions for one block
  a second designer, given the same brief and the same 156,
  will produce a page that sits beside the first one without
  a visible seam

GIVEN

  measure      960
  columns      12
  gutter       24
  column       = (960 - 11 * 24) / 12 = (960 - 264) / 12 = 58
  span width   = 58s + 24(s - 1) = 82s - 24
                 for a span of s columns
  baseline     8
  body         16 / 24    24 = 3 baselines
  h2           32 / 40    40 = 5 baselines

HORIZONTAL CONSEQUENCES

  distinct block widths           = 12
  distinct (start, span) places   = sum of (13 - s), s = 1 to 12
                                  = 12 + 11 + 10 + ... + 1
                                  = 78
  two-block splits of the measure = 11
  three-block splits              = C(11, 2) = 55

VERTICAL CONSEQUENCES

  every block height is a multiple of 8
  n lines of body text = 24n = 3n baselines
  a rule of "gaps are 3 or 5 baselines" leaves exactly two
  vertical gaps = {24, 40}

RESULT

  the question "where does this go" has become
  "which of 78 placements, and which of 2 gaps"
  78 x 2 = 156 permitted positions for one block
  a second designer, given the same brief and the same 156,
  will produce a page that sits beside the first one without
  a visible seam

The arithmetic is unglamorous and that is the point. One hundred and fifty-six is a large number for a single block and a tiny number compared with a blank page. The grid has not made the layout. It has made the layout enumerable, which is exactly Gerstner's act of selection.

The span table is the part a team actually uses day to day.

Every figure in that table comes from the single expression 82s minus 24. Nobody has to remember it. They look it up, and because they look it up rather than estimating, the fourth article of an issue lines up with the first.

The multi-person problem is the real problem

A designer working alone on one artefact does not need a grid. They need a good eye and enough time, and the grid may well slow them down and flatten the result. This is why the argument about grids keeps going in circles: both sides are right about different production conditions.

The grid earns its keep at the point where two things become true at once. The first is that more than one person is setting pages. The second is that the thing has more than one issue, edition, release or screen size, and the earlier ones will still be visible when the later ones ship.

Under those conditions the grid stops being a compositional tool and becomes a coordination protocol. Its job is to make the decisions of person A on Tuesday compatible with the decisions of person B on Thursday without the two of them speaking. Gerstner's formulation covers this exactly: the maximum of constants with the greatest possible variability. The constants are the coordination; the variability is what is left for judgement.

I got this wrong for a long time, and it is worth saying how. For years I set the grid first, treated it as the design, and treated any page that broke it as a failure of discipline. What I was actually doing was using the grid to avoid making the hard decisions rather than to defer them until I had the information to make them well. Deferral and avoidance look identical from outside. The difference is whether you ever come back.

The cost of a grid nobody breaks

Here is the part the handbooks do not cover.

A grid distributes emphasis evenly by construction. That is its virtue when the content is evenly important, which is the case for a timetable, a catalogue, a price list, a specification, a set of technical drawings. It is a defect when the content is not evenly important, which is the case for almost everything anyone reads voluntarily.

If every element in a publication sits on one of 156 permitted positions, then no element can signal, by its position alone, that it matters more than the others. Hierarchy has to be carried entirely by size, weight, colour and white space, and those channels saturate quickly. After a while the reader stops reading position as information at all, because position has stopped carrying any. You have bought consistency by spending the cheapest and fastest signal you had.

The second cost is slower and worse. A grid that is never broken stops being tested. Nobody discovers that the four-column measure is too narrow for the kind of writing the publication has drifted towards, because nobody ever tries to set a paragraph outside it. The grid becomes a description of what the organisation used to publish rather than a tool for what it publishes now, and because it is enforced rather than argued about, the mismatch is invisible until somebody redesigns from scratch and discovers the last three years look wrong.

The third cost is honest and small: grids make bad work look competent. A page that is aligned and rhythmically regular reads as considered even when nothing about the content has been considered. That is genuinely useful when you are shipping under time pressure, and it is genuinely corrosive when it becomes the standard the work is judged against.

Müller-Brockmann's list of wills contains a clue that he knew this. Every item begins with "the will to". None of them is "the rule that". He is describing an intention held by a person, not a constraint enforced by a file. A grid held as an intention can be set aside for a reason. A grid held as a rule cannot, and a rule that cannot be set aside for a reason is no longer a design decision, it is furniture.

What I would keep

Keep the arithmetic. Write the specification down, in numbers, with the derivation visible, so that anyone can recompute a span width rather than guessing it. Keep the baseline, because vertical rhythm is the thing that most reliably falls apart across multiple hands and is most invisible when it does.

Keep Gerstner's framing over Müller-Brockmann's. The grid is a programme, which is to say a thing that generates solutions, not a solution. A programme with no output is a failed programme, and so is one whose output is always the same.

And keep a register of the breaks. Every time somebody sets something outside the grid, write down what they were trying to do and whether it worked. After a year that register tells you more about what the publication actually needs than the grid does. If the register is empty, the grid is not being used, it is being obeyed, and the two are not the same thing at all.

Sources

• Müller-Brockmann, Josef. Rastersysteme für die visuelle Gestaltung / Grid Systems in Graphic Design: A Visual Communication Manual for Graphic Designers, Typographers and Three Dimensional Designers. Niggli, 1981. 176 pages. ISBN 978-3-7212-0145-1. Publisher's record: niggli.ch/en/products/rastersysteme-fur-die-visuelle-gestaltung

• Müller-Brockmann, Josef. Gestaltungsprobleme des Grafikers / The Graphic Artist and his Design Problems. Niggli, first published June 1961. 200 pages, 735 illustrations. ISBN 978-3-7212-0466-7. Publisher's record: niggli.ch/en/products/gestaltungsprobleme-des-grafikers

• Gerstner, Karl. Programme entwerfen / Designing Programmes: Instead of Solutions for Problems Programmes for Solutions. Arthur Niggli, Teufen AR, 1964, first edition, 96 pages, introduction by Paul Gredinger. Bibliographic record and dealer description:

• "Programme entwerfen." French Wikipedia (publication year, publisher, the four constituent essays, the simultaneous German and English editions). fr.wikipedia.org/wiki/Programme_entwerfen

• Gerstner, Karl. Designing Programmes, excerpt reproduced online, including the passages on the morphological box of the typogram, on design as an act of selection, and on the grid as a programme par excellence. medium.com/@digitalonetwo/designing-programmes-excerpt-62bc64e67b8b

• "Grid and Design Philosophy: Josef Müller-Brockmann" (transcription of the design philosophy passage, including the list of wills and the statement on professional ethos). neugraphic.com/muller-brockmann/muller-brockmann-text2.html

• Wikipedia. "Josef Müller-Brockmann" (born 9 May 1914, died 30 August 1996; professor at the Zurich school 1957–1960; Neue Grafik founded 1958 with Neuburg, Lohse and Vivarelli; guest lecturer at the Ulm School of Design in 1963; IBM Europe consultant from 1967). en.wikipedia.org/wiki/Josef_Müller-Brockmann

• Wikipedia. "Karl Gerstner" (born 2 July 1930, died 1 January 2017; Gerstner+Kutter 1958; GGK 1962; Gerstner Programm 1964–67 in sixteen cuts for Berthold Diatype; MoMA exhibition 5 February to 30 March 1973 directed by Emilio Ambasz; archive to the Swiss National Library in 2006). en.wikipedia.org/wiki/Karl_Gerstner

Related reading

• Munich 1972 Was Public Infrastructure

• Ulm, and the Idea That Design Could Have a Method

• Tschichold Changed His Mind

• The Job Title Nobody Has Any More

More from the Moc Mallen Studio journal