Too Narrow

In 1944 a physicist proposed that a chromosome carries a code-script. The word went round the world, and it is still going.

Four lines later, in the same paragraph, he wrote that the term was “of course, too narrow” — because a chromosome is not only the plan. It is also the builder. Architect’s plan and builder’s craft, in one.

That second sentence did not travel. This chain is about where it stopped, what filled the space it left, and what that space was eventually used for.

It runs from 1944 to 2025, and every joint in it is marked — documented where the step is established, contested where serious people disagree or we cannot source it, leap where we are the ones reaching. Nine, two and one. A chain is built to feel inevitable. This one is not allowed to.

What this chain is not. It is not an argument that genes do not matter, and it is not an argument against behaviour genetics, heritability, or the genome project. Every correction in it — the gene count, the onion, the 2010 review, the consultation that closed a study — was produced from inside the same enterprise, and the sharpest criticism quoted here comes from a working behaviour geneticist. This is an argument for a part that was left behind, and the person who first pointed it out was the one who coined the metaphor.

A physicist gave three lectures in Dublin in February 1943, and published them the next year as a short book called What Is Life? In it he asked what a chromosome could possibly be, and answered with a word that has never left us.

He called it a code-script. A chromosome, he proposed, carries “in some kind of code-script the entire pattern of the individual’s future development and of its functioning in the mature state.” That is the sentence everybody knows, and it is the sentence the next eighty years were built on.

This chain is about the sentence immediately after it.

He got there first and he got there completely. Not a plan that something else carries out — the plan and the carrying out, the same object, at the same time. Of course, too narrow. He did not have to be argued into it. He wrote it in the same breath as the metaphor, in the same paragraph, on the same page.

Read at full text, and quoted from the book rather than from any of the many places it is paraphrased. The paraphrases keep the first sentence.

A short book by a physicist, about biology, turned out to be the recruiting pamphlet for the discovery that followed.

James Watson read it as an undergraduate at Chicago and switched from birds to genetics. Francis Crick read it and concluded a physicist could do something about the secrets of life; he went to the Cavendish. He wrote to Schrödinger on 12 August 1953, months after the double helix, to say so. Maurice Wilkins read it and stayed in science.

All three later received the same Nobel Prize, and all three credited the same book.

So the transmission is not in doubt. This is the rare case where a metaphor’s route from one head to another is documented by the heads it arrived in. Which makes the loss in the next several links harder to explain away as accident: the people who carried the idea forward had read the qualifier. It was on the page they were reading.

What we are not claiming. That any of the three ignored it deliberately, or even noticed it. A qualifier can be read and not registered, and usually is. The claim here is only about what was available, not about anybody’s state of mind.

This is the mechanism of the loss, and it is not villainy. It is what happens when one half of an idea can be tested and the other half cannot.

Watson and Crick’s 1953 paper ends with the most quoted sentence in molecular biology: “It has not escaped our notice that the specific pairing we have postulated immediately suggests a possible copying mechanism for the genetic material.”

That is the plan half, and it was spectacular. Fixed pairing, a sequence that can be copied, information that survives being taken apart. Within a decade the code itself was cracked, triplet by triplet. Every step produced a result you could put in a table.

The craft half produced no such table. How a copied sequence becomes a body — the development, the timing, the cell that is already there doing the reading — is enormously harder, and in 1953 there was almost nothing to measure. It stayed hard for fifty years.

A metaphor with a testable half and an untestable half will be remembered by its testable half. Not because anybody decided to drop the rest, but because the rest had no findings attached to it.

The architects of the Modern Synthesis answered the question how do genotypes produce phenotypes? in a way that made the question go away.

The answer, as Pigliucci summarises it, is that genes determine phenotypes — expressed in what he calls “the oft-cited metaphors of a ‘genetic blueprint’ or a ‘genetic programme’.” Development, on that account, is a step that happens reliably and therefore does not need explaining.

This is the joint where the craft comes off. Schrödinger’s point was that the chromosome is the builder as well as the plan. A framework that files the building under incidental has kept exactly one half of what he said — and it is the half his own next sentence warned was not enough on its own.

The consequence Pigliucci draws is institutional rather than rhetorical: developmental biology was largely excluded from mid-twentieth-century evolutionary theory, and only came back with evo-devo. That is a field-shaped hole, not a turn of phrase.

This is the weakest joint in the chain and it is marked contested for a reason we can state precisely: we cannot source it.

Blueprint and programme became the ordinary way to describe a genome, in textbooks, in teaching, in newspapers, and eventually in policy. Pigliucci attributes the framing to the architects of the Modern Synthesis collectively and quotes it as oft-cited, which is a description of ubiquity, not of origin.

We looked for a first use and did not find one we would print. There is no single paper, no coinage, no moment. That absence is itself worth noticing: the most consequential metaphor in twentieth-century biology appears to have arrived by accretion rather than by anybody proposing it.

What we can say, and can defend: by the time it was ubiquitous it carried a claim Schrödinger had explicitly disowned. A blueprint is precisely a plan that something else executes. That is the whole difference between an architect’s drawing and a builder’s craft, and the word chose one.

If you can source it, tell us. A documented first use would upgrade this joint, and we would rather be corrected than keep a contested link at the centre of a chain.

On 26 June 2000 the first survey of the human genome was announced at the White House, and the metaphor was given the largest stage it will ever have.

And from Francis Collins, then director of the National Human Genome Research Institute, at the same event: “Today, we celebrate the revelation of the first draft of the human book of life” — and, of the genome, “our own instruction book, previously known only to God.”

Book. Language. Instruction book. Every one of them is the plan half. Not one of them is the builder. A book is a thing that is read by something else; that is what makes it a book.

This is not a criticism of the project, which was an extraordinary piece of work, or of the people speaking, who were describing a real achievement on a day that deserved celebrating. It is an observation about which half of a 1944 sentence was available to reach for on the biggest day biology had had in fifty years — and which half was not.

Here the chain turns, and it turns on evidence produced by the same project.

Before the sequence was in, estimates of the number of human protein-coding genes ran from 40,000 to 100,000. The higher figure circulated widely. You can see the reasoning without anybody spelling it out: a person is a complicated object, and a specification of a complicated object has a great many entries in it.

The count settled near 20,000.

That is not a small correction. It is a factor of five, and it was a prediction of the metaphor rather than of any particular experiment. If the genome is a specification, its size should track what it specifies. It does not.

And this is the collection’s point about science rather than against it. The number was published, the number was wrong, the number was corrected in the open, and the correction is now the textbook figure. That is the process working. What did not get corrected, at anything like the same speed, was the picture that produced the guess.

The second piece of evidence is older than the genome project, cheaper than it, and considerably ruder.

Allium cepa — the kitchen onion — has a genome of about 16,400 Mb. A human has about 3,200 Mb. The onion carries roughly five times more DNA than you do, and it is not a close call.

T. Ryan Gregory put it as a standing challenge, the onion test: can you explain why an onion needs about five times more non-coding DNA than a human?

And the sharpest form needs no theory of complexity at all. Within the genus Allium — onions, garlic, leeks — genome size runs from about 7 to 31.5 picograms. Onions disagree with other onions by more than onions disagree with us, and they are all still onions.

What this does and does not settle. Most of the onion’s bulk is repetitive sequence, and whether that sequence does anything is a live argument this chain takes no side in. None of that is needed. The claim is only that quantity of DNA does not track the organism — which is the one thing a blueprint would have to do.

In 2010 the Philosophical Transactions of the Royal Society B published a review with the conclusion in its title.

He lists four failures. Proportionality — sequence length should track complexity and does not, which is links 7 and 8. One-to-one mapping — very different genotypes produce the same phenotype. Brittleness — blueprint thinking predicts fragile systems, and real development is strikingly robust. And causation — “genes determine phenotypes” simply omits the process that does the determining.

What he proposes instead is not a better picture but a different kind of thing: developmental encoding, in which what is inherited is a set of generative rules whose results depend on the developing system at every step.

Which is Schrödinger’s sentence, arrived at again from the other end, sixty-six years later. Plan and craft, in one.

A metaphor declared finished in a Royal Society review was, eight years later, the title of a major popular book by one of the most cited psychologists alive.

Robert Plomin, Blueprint: How DNA Makes Us Who We Are (2018). Its thesis, in his own compression: “DNA isn’t all that matters, but it matters more than everything else put together.” He puts genetic factors at roughly half the variation in psychological traits, and argues that environmental effects are largely unsystematic — and therefore, he says, not much can be done about them.

This joint is contested and the reason matters. Heritability is real, the twin and adoption literature is real, and this chain is not an argument that genes do not matter. The objection that carries weight here comes from inside behaviour genetics:

The historian Nathaniel Comfort put it harder, calling the book a road map for regressive social policy. Others, including geneticists, have defended it as misread. We are not adjudicating that, and a chain is not the place to. What the chain needs is only the dating: the word came back, on a cover, after the review that retired it — and what it carried back was the half without the craft.

In August 2021 a project opened to collect DNA from ten thousand Autistic people and their families. This is the link where the chain stops being about metaphors.

Spectrum 10K — £3 million, funded by Wellcome, led by Simon Baron-Cohen at Cambridge’s Autism Research Centre, the largest genetic study of autism in the UK. Within weeks it was paused, after sustained criticism from Autistic people that the community had not been properly consulted and that the data could be used by others seeking to prevent or eliminate us.

An independent consultation followed, run by Hopkins Van Mil, across three stages and several years. In 2025 the study was closed, and Wellcome said it supported the decision.

We are writing from inside this. Not about a community — from within one. That is why this link exists and why the chain was worth building.

And this is not a story about a villain. A study was proposed, the people it was about objected, a real consultation happened, and the project ended. That sequence is the thing working, not failing. What it took was us saying so, loudly, for four years. The question the next link asks is why the objection had to be that loud — and whether a picture in which a person is a plan that can contain a mistake had anything to do with it.

Here is the leap, and it is the only one in the chain. We are marking it because the eleven links before it do not add up to it.

The tempting sentence is: the blueprint metaphor caused the gene-hunts. We cannot show that, and we are not going to write it as though we could.

What the chain does establish is that a picture was available, everywhere, for eighty years, in which a person is a specification. And a specification is the kind of object that can contain an error. You can look through it for the part that went wrong. That is what the word is for: it is what makes a blueprint useful to a builder.

What we cannot establish is the arrow. Did the picture make the search thinkable, or did people who already wanted to search reach for a picture that suited them? Both stories fit every fact above. Historians of eugenics have argued versions of each.

What we will say without hedging is narrower and is not a leap at all: the picture is wrong. Not unkind — wrong, on the evidence of links 7, 8 and 9, and wrong by the account of the man who coined it in link 1. Whatever it did or did not cause, nobody is obliged to keep it.

A rhyme, not a proof. This chain is an argument for the part that was left behind, and it is not an argument against the science that left it. Every correction in it — the gene count, the onion, the 2010 review, the consultation that closed a study — came from inside the same enterprise.

The thing this chain keeps arriving at is not that somebody was fooled. It is that the fix was never missing.

It was published in 1944, in the same paragraph as the problem, by the person who caused the problem. It was read by the three people who went on to find the structure. It has been in print continuously for eighty years and it is four lines long.

And the field still had to come back round to it the hard way — through a gene count that was wrong by a factor of five, through an onion, through a review paper in 2010 that proposed developmental encoding and arrived, by a completely different route, at the sentence Schrödinger had already written.

Here is what we think that is about, and it is the nearest thing this chain has to a moral. A qualifier survives only if something can be done with it. The plan half generated experiments, papers, prizes and a project that cost three billion dollars. The craft half generated, for fifty years, almost nothing you could put on a slide. It was not refuted. It was unusable, and unusable is how things get dropped.

Which is a warning about the present rather than the past. The parts of any current picture that are hardest to measure are the parts most likely to fall off it, and they will fall off quietly, and nobody will have decided anything.

What we are left holding. Not that genes do not matter — they matter enormously, and every link above rests on work done by people finding out how. Only this: whatever you are, you are not a drawing that something else is working from. The plan and the building are the same thing, and they are happening now. Schrödinger said so first, and said it was obvious, and the word he used was of course.