The best sign is the one nobody asks about

I didn’t design this sign, I found it. It hangs in stairwells, schools and office corridors, and hardly anyone has ever looked at it consciously. This text argues the following thesis: a safety sign succeeds when it is obeyed rather than noticed. To show what that inconspicuousness consists of, I take the sign apart with Charles Sanders Peirce’s distinction between icon, index and symbol, and then change a single variable, its colour.

The object

The sign is E002 of ISO 7010, emergency exit, right hand: a green square, inside it a white door, inside that a green figure running through. The colours and basic shapes of safety signs are set by ISO 3864-1, first published in 1984; there, green stands for safe condition and rescue, red for prohibition and fire equipment (International Organization for Standardization [ISO], 2011). Since 2003, ISO 7010 has registered the individual pictorial signs. In Germany, the technical rule for workplaces ASR A1.3 of 2013 makes these signs binding for workplaces.

Fig. 1 — E002 emergency exit, right, under ISO 7010 and ASR A1.3, in its standard green.

The figure goes back to the running man that the Japanese designer Yukio Ota drew in the late 1970s for a competition run by a Japanese fire safety association. According to Julia Turner’s account, some 3,300 entries were submitted; the candidates were tested for whether they could be understood in a smoke-filled environment. Japan submitted Ota’s design to ISO in 1980, and it was adopted in 1985 (Turner, 2010). That visibility in smoke is a measurable quantity was confirmed by later research: in experiments at the US National Institute of Standards and Technology, a sign’s luminance was one of the primary determinants of its visibility in smoke (Collins et al., 1990, p. iii).

Three observations

First, the figure crosses its frame. The rear leg steps beyond the edge of the door, the leading hand disappears into the green on the other side. Without this crossing the sign would show a person in a door; with it, it shows the act of passing through.

Second, a narrow white outline separates the leg from the green ground where it leaves the door. Without it, green would stand on green, and the distinction between figure and ground that Edgar Rubin described systematically in 1915 would be lost at this point (Rubin, 1915).

Third, the head sits above the torso as a free circle. My hypothesis is that a detached circle stays recognisable as a head under blur for longer than an attached one. This has not been shown for this sign; it could be tested with the blur test from no. 012.

Icon, index, symbol

Peirce distinguishes signs by how they are connected to their object. An icon represents by resemblance, an index by physical connection, like a weathercock to the wind, a symbol by convention or habit (Peirce, 1894/1998). From this follows the sentence this article turns on: every symbol is a sign, but not every sign is a symbol. Symbol is one of three kinds of sign relation, not a synonym for sign.

Two qualifications are needed. First, for Peirce the three kinds are aspects, not separate classes; it would be difficult, if not impossible, to find an absolutely pure index (Peirce, 1955, p. 108). Second, Ferdinand de Saussure uses the word in the opposite direction. For him a symbol is precisely never wholly arbitrary: the scales as a sign of justice could not be replaced by a chariot (Saussure, 1916/1959, p. 68). Anyone writing about symbols therefore has to say in whose sense. Here it is Peirce’s.

In this sense, E002 is a mixed sign. Figure and door are iconic; you recognise them by resemblance. The place is indexical: the sign means something only where it hangs, and its mirrored version E001 points the other way. The colour is symbolic. That green means “rescue” lies in no property of the colour but in a standard.

How a sign becomes a symbol

Peirce answers the question with a model of growth. Symbols come into being out of other signs, particularly out of likenesses and mixed signs; once in being, they spread, and in use and experience their meaning grows (Peirce, 1894/1998, Section 8). At the same time he insists that a new symbol can only grow out of symbols.

The running man confirms both. He began as the likeness of a person running. He became the symbol for “exit” not through his form alone but because he was inserted into an existing system of conventions: colour code, standard, regulation, daily repetition in buildings. A sign becomes a symbol when its meaning no longer depends on resemblance but on a shared habit that carries it.

The convention is not entirely arbitrary, however. The studies reviewed by the NIST team found that red signs were scattered slightly less by smoke, but recommended green because people with red deficiency recognise red signs less well (Clark et al., 1985, as cited in Collins et al., 1990, pp. 2–3). The meaning of the green is conventional; its choice is only partly so. In Saussure’s sense, that is exactly what would make it a symbol.

Inconspicuousness as function

In no. 001 I rejected Beatrice Warde’s demand for invisible typography, first made in a lecture in 1930, as a general rule (Warde, 1955). For the safety sign, by contrast, it holds without qualification. It is read in a situation in which attention is scarce, and it has only one statement to make. Any departure from the learned form would first have to be interpreted. That is the core of the thesis: inconspicuousness is not a style here but the result of a stable convention.

One variable: the colour

On my canvas I therefore change only the colour. Figure, door, running direction and outline stay as the standard draws them; the green becomes red. Icon and index are unchanged; only the symbolic part is affected. Under ISO 3864-1, red denotes prohibitions and fire equipment. In the United States, by contrast, the National Fire Protection Association prescribes no colour, and red EXIT signs are widespread there (Turner, 2010).

Fig. 2 — The same drawing with red instead of green. Only the conventional part of the sign has changed.

This yields a testable hypothesis: the red version will be misread more often than the green one, and how it is read will depend on the convention its viewers grew up with. It has not been tested here. The method exists: comprehension tests under ISO 9186, in which a safety sign under ISO 3864 must be interpreted correctly by 67 per cent of respondents and under the American ANSI Z535.3 by 85 per cent (Gungor, 2023). If the red version fell below this threshold and the green one did not, it would show that the sign’s achievement lies less in its drawing than in the habit with which it is read.