Prohibition of images

Why do traditions that forbid any image of the divine permit elaborate description of it in words?

The commandment against making any graven image or likeness sits inside a text that spends thousands of lines describing God — with an outstretched arm, a face that may not be seen, a voice, a jealousy, a womb-like compassion. Both cannot be a ban on picturing the divine, because one of them is picturing the divine, at length and in detail, in the same document.

So the rule is not "do not represent". It is "do not represent that way". Which raises the real question: what is a painted or carved likeness doing that a spoken one is not?

Start with the obvious candidate: resemblance. A word does not look like its referent, an image does, and the worry is confusion — someone mistaking the statue for the god. Test it, though. Nobody in these traditions seems to have thought worshippers were literally confused; the polemics accuse image-users of misdirected devotion, not of a factual error.

Try a second difference, which I think does real work. Words carry their own cancellation. A tradition that calls God a rock, a shepherd, a consuming fire, a jealous husband and a mother eagle has produced a set of descriptions that cannot all be true at once, and the incompatibility is the safeguard — piling metaphors up forces the reader to hold them as figures. An image cannot do this. A picture asserts a form and has no grammar for "not literally". Sol Worth's essay on the point is titled "Pictures Can't Say Ain't" (1975, recalled): images have no negation, no "as it were", no subjunctive. So a single carved figure commits to a face, an age, a body, with nothing in it that marks the commitment as provisional.

Now a third difference, and the one the practices themselves seem most anxious about. An image is an object in a place. You can stand before it, bow, dress it, garland it, carry it in procession, house it, defend it when the town is attacked. A verbal metaphor is nowhere and cannot be approached. The hazard is therefore not confusion but termination: the danger that the act of devotion stops at the object instead of passing through it.

That reframing explains something the resemblance story cannot. Relics resemble nothing at all — a bone, a splinter — and are policed as heavily as statues, because they are supremely addressable. Illustrations in a book resemble far more and provoke far less anxiety, because nobody kneels to a book plate. Addressability, not likeness, tracks the prohibition.

Then watch the loop this sets up, because it is why the argument never stays settled. An image that draws devotion draws visitors; visitors bring gifts; the shrine grows wealthy and institutionally central; its centrality attracts still more devotion. Nothing in that spiral is checked from inside, so it eventually provokes a reformist purge — and the purge itself generates new sacred objects, in martyrs and in the images hidden and returned. Byzantine iconoclasm ran through roughly the eighth and ninth centuries in two phases, with icons restored at the Second Council of Nicaea and again after the second phase (dates recalled; the chronology is disputed). The pattern is a cycle, not an event.

The strongest reply to all of this comes from the image-defenders, and it is not a dodge. John of Damascus argued that after the Incarnation the case is transformed: the God who could not be depicted took a depictable body, so refusing to depict him denies that he really did. He drew the line differently — the material is honoured because God honoured matter by entering it. Alongside this the tradition set a distinction between latreia, worship owed to God alone, and the proskynesis or veneration that may be given to an icon, with the formula that the honour paid to the image passes to its prototype (attributed to Basil and quoted at Nicaea; recalled). If that distinction holds, the termination worry is answered.

Does it hold? The honest answer is that we do not know, and both sides read the same evidence oppositely. Image-permitting traditions did not collapse into idolatry, which the iconodule takes as proof the risk was overstated; the iconoclast replies that they avoided it only by erecting the elaborate doctrinal apparatus in question, which is evidence the risk was real. Note too that image-forbidding traditions grow image-substitutes anyway — venerated calligraphy, a covered building, a relic, a book that is kissed. The devotional pressure relocates rather than disappearing.

There is also a competing explanation that owes nothing to psychology. Aniconism may function chiefly as boundary-marking: the surrounding cults had images, so refusing them is what distinguishes us, and the intensity of the prohibition tracks proximity to those cults rather than any risk to the individual worshipper. Two facts sit awkwardly with the purely metaphysical reading. Islamic aniconism rests mainly on hadith about image-makers rather than an explicit Qur'anic ban (recalled), and its most visible consequence was that calligraphy and geometry became the great visual arts — the word took over the visual field rather than emptying it. And the third-century synagogue at Dura-Europos carries an extensive figurative fresco cycle, so the ban was not applied uniformly even where affirmed.

• Why the traditions that most strictly bar images produce the most elaborate script and ornament.
• How the worship-versus-veneration distinction is meant to be recognised in practice, by the person performing the act.

Every term the collection defines is gathered in the glossary.