In the oldest surviving poetry a single colour term will be applied to the sea, to sheep, and to worked bronze, and for a century that fact has been treated as a puzzle in need of a solution. The solution now generally accepted is the lexical-gap account: the vocabulary was thin because hue had little work to do in a world of few dyes and fewer pigments, and a language leaves unnamed whatever it has no occasion to name. On this account nothing follows about how the speakers saw, and the reading the account displaced -- Ollerenshaw's, that the earliest terms were organised by brightness rather than by hue -- is cited chiefly as an instance of philology overreaching.
The displacement was deserved, and it is worth saying how far. Ollerenshaw's successors did not stop at describing a vocabulary. They inferred from the vocabulary to the eye, and held that speakers who lacked a term for a hue could not reliably pick that hue out. No inference of that kind is available. The passage from what a language names to what its speakers can discriminate has been closed for good, and everything Ollerenshaw's followers built on the far side of it goes with it.
What survives is a smaller claim, and the lexical-gap account has not answered it. A gap is an absence, and an absence explains a silence. The texts are not silent. They show a positive distribution: one term shared by the sea and by worked bronze, another by wool and by the sky at dusk, a third by honey and by a lion's coat -- pairings that hue does not organise and that brightness, together with the play of light across a surface, organises rather well. The lexical-gap account tells us why certain words are missing. The distribution is made of the words that are there.
To this the account's defenders answer that the pairings are artefacts of translation: we gloss a term with one English colour word, find it attached to two things of unlike hue, and manufacture a puzzle out of our own gloss. The answer is a good one for any pairing taken singly, and no single pairing can be pressed. It is much weaker against the set of them, because it would have to be made afresh for each, with a different secondary sense proposed each time, and those secondary senses would then have to agree with one another as the primary ones do -- which is the very regularity the reply was introduced to dissolve.
None of this settles the matter. It is possible that a distribution of that shape could be produced by the accidents of a small surviving corpus. What it does establish is that the question is open and the lexical-gap account is not entitled to treat it as closed. Ollerenshaw was wrong about what a colour vocabulary shows about seeing and right about what one is organised by, and only the second half of him was ever needed.
