More Than One Way to Spell It
a read-aloud about reading three letters at a time, and finding out that nearly everything has several right spellings — with something you can try, and every source kept on the page for the grown-up
You know the letters already
Last time we met the four letters that everything alive is written with. A, T, G and C.
And we found out the surprising thing about them: nobody checks the order. Any order at all is allowed.
So here is the next question, and it is a good one.
If any order is allowed — what is the order for?
What does all that writing actually say?
This is the companion to No. 105, which was about the alphabet — which letters exist, and how they pair. That zine deliberately left this subject out and said so on its own face: “The genetic code is a different thing… The code is not in this zine at all.” This is the promise being kept.
The word “code” is about to do a lot of work, so it is worth pinning down now. The genetic code is not a secret message and it is not a set of instructions for building a person. It is a lookup table: three letters go in, one amino acid comes out. That is the entire content of it. Spread ten comes back to this, in the child’s own layer, because the popular meaning of the phrase is the misreading rather than a simplification of the real one.
One boundary kept from No. 105. Nothing in this zine says what any particular stretch of DNA does in a body, or what any gene is “for”. The subject here is the table, not the consequences.
Three at a time
The letters are not read one at a time.
They are read in threes. Always threes.
So a long line of letters gets chopped into little three-letter words, one after another, all the way along.
A three-letter word like that has a name. It is called a codon.
That is a proper grown-up word and you may keep it.
Where the chopping starts matters enormously, and it is the thing the picture cannot show. The same row of letters read starting one place to the right gives entirely different words from there on — that offset is called the reading frame, and a change that shifts it is a frameshift. This zine deliberately does not go there; it is a real and much larger subject, and the claim on the next six spreads does not depend on it.
The letters in the picture are U rather than T, and that is not a slip. The strand actually being read three at a time is a working copy made of RNA, which uses uracil where DNA uses thymine — the swap No. 105 noted in passing. Every codon in this zine is written the way codons are conventionally written, in RNA letters, so that a child who looks one up later finds the same spelling.
The row of letters in the figure is invented. It is there to show the chopping, not to be anybody’s sequence.
Each three names one bead
What is all this building? Mostly it is building proteins, and a protein is a very long string of beads.
There are twenty kinds of bead. Twenty, and that is all.
And here is the whole job of a codon. Each three-letter word names one bead.
GGU means: put a glycine bead on next.
That is it. That is everything a codon says.
Not make a nose. Not be good at running. One bead, and then the next word says the next bead.
The beads are amino acids, twenty of them in the standard set, strung into chains that fold into proteins. Glycine is the smallest. The bead image is doing real work rather than softening anything: a protein genuinely is a linear sequence of these units, assembled one at a time in the order the codons come.
“Not make a nose” is the load-bearing sentence on this spread, and it is literally rather than rhetorically true. The genetic code is a mapping from 64 triplets to 20 amino acids plus a stop signal. There is no entry in it for a trait, a behaviour, an ability or a person. Whatever is true about how bodies develop — and it is complicated, and contested, and involves a great deal besides sequence — none of it is in this table. Spread ten says so to the child.
Three of the sixty-four words name no bead at all. They mean stop: the chain ends here. And one, AUG, does double duty — it names methionine and is also the usual signal for start here.
Sixty-four words, twenty beads
Let us count how many different three-letter words you can even make.
Four choices for the first letter. Four for the second. Four for the third.
4 × 4 × 4 = 64
Sixty-four different words.
But there are only twenty beads to name. And three words are used up saying stop.
So there are far more words than there are things to say.
Something has to give. What do you think it is?
The arithmetic closes exactly, and it is worth doing with a child who wants to. 61 codons name an amino acid and 3 mean stop, which is 64. Those 61 distribute as 5 fourfold sets, 9 twofold sets, 3 sixfold sets, 1 threefold set and 2 single ones — and 5 + 9 + 3 + 1 + 2 = 20, which is every amino acid accounted for, once each.
Every codon assignment in this zine was checked against the NCBI translation tables, table 1, “The Standard Code” — not written from memory. That matters more than usual here, because the demonstration on the next two spreads only works if the particular codons named are the right ones.
The picture draws sixty-four squares and twenty circles at their true counts. Nothing in it is schematic except the arrangement.
More than one way to spell it
Here is what gives, and it is the whole point of this zine.
Nearly every bead has more than one spelling.
Glycine has four. GGU, GGC, GGA, GGG.
All four mean glycine. Not one proper one and three near-misses. All four are right.
And leucine has six.
Two beads do only get one spelling each, and they are the odd ones out: methionine and tryptophan.
The technical word for this is degeneracy — the genetic code is said to be degenerate. It is a genuinely unfortunate term for the property it names, which is redundancy, slack, and more than one correct answer. It is not in the child’s layer for that reason, and it is here so you have it if somebody meets it later.
The three sixfold beads are leucine, serine and arginine; the two singletons are methionine (AUG) and tryptophan (UGG); isoleucine is the one with exactly three. Verified against NCBI table 1.
Synonymous does not always mean identical in effect. Two spellings of the same amino acid produce the same bead, but they are not always interchangeable in a living cell — codon usage can affect how fast a chain is built and how it folds, and some synonymous changes do have consequences. The claim in the child layer is about what the table says, which is exactly and only that the bead is the same. That is true; the broader “so it makes no difference” is not, and is not claimed.
Change one letter
You need a pencil. The little table is underneath.
One. Write GGU. Look it up. It is glycine.
Two. Now change the last letter. Any letter you like. GGC. Or GGA. Or GGG.
Look each one up.
Still glycine. Every single time.
Three. Now put it back to GGU and change the first letter instead. G becomes A. That gives you AGU.
That one is not glycine. It is serine.
The last letter often does not matter at all. The first one nearly always does.
| Word | Bead |
|---|---|
| GGU | Glycine |
| GGC | Glycine |
| GGA | Glycine |
| GGG | Glycine |
| AGU | Serine |
| AGA | Arginine |
Six rows out of sixty-four, and the extract is deliberate. A full codon table on a read-aloud spread is a wall; these are exactly the rows the game needs. The whole table is at NCBI, and every row above was checked against it.
The third position is often the tolerant one, not always, and the child layer says “often” for that reason. Glycine is a fourfold set, so all four third letters give glycine. But UUU and UUC are phenylalanine while UUA and UUG are leucine — same first two letters, and the third letter decides. If your child tests a different word and finds the third letter mattering, they have found a twofold set and they are right, not wrong.
Why the third position gets away with it. The pairing between a codon and the molecule that delivers the bead is looser at that position than at the other two — Crick named it wobble. The child does not need the mechanism to run the experiment; the pattern is real and they can find it themselves.
A lot of changes change nothing
Copying all those letters is not perfect. Sometimes a letter comes out different from the one before.
And a great many of those times — nothing happens at all.
The word is spelled a different way, and it names exactly the same bead. The string comes out the same.
It is not that somebody caught the mistake and fixed it.
It is that there was more than one right answer, so it was never a mistake in the first place.
These are called synonymous or silent substitutions, and they are common enough to be a standard tool: comparing rates of synonymous and non-synonymous change is one of the ordinary ways of detecting selection on a gene.
There is a famous result here and it needs its hedge carried with it. Freeland & Hurst, “The genetic code is one in a million,” J. Mol. Evol. 47, 238–248 (1998), found the standard code extraordinarily good at limiting the damage of errors compared with randomly generated alternatives. Read at second hand — PubMed refused us, so those figures come from an open-access review that cites it, not from the paper.
And the hedge is the part that usually gets dropped. That same review puts the code “halfway from an average random code to the summit” of what is possible, notes that billions of alternative codes would do better, and concludes that Crick’s rival account — that the code is a frozen accident, arbitrary and stuck because “at the present time any change would be lethal, or at least very strongly selected against” — is essentially unfalsified. So this zine does not say the code was arranged kindly, or arranged at all. It says what is plainly true and checkable: there is more than one right spelling, and a lot of changes therefore change nothing. Why that is so is an open argument, and the page does not borrow its warmth.
And you are using two of these
The list of which word means which bead — that is the code. One list, for everything alive.
You would think there is one of it.
People have written down at least 33.
And you are using two of them. Right now. Inside you.
Most of you uses the usual list. But the tiny power parts inside your cells — they are called mitochondria — use a different one.
They do not agree about UGA. In most of you it means stop. In there it means a bead called tryptophan.
The count and the differences are NCBI’s. Their translation tables run to 33, numbered 1–33 with several retired and the numbers left as gaps. Table 1 is “The Standard Code”; table 2 is the vertebrate mitochondrial code, which is the one your mitochondria use. It differs in three places: UGA is tryptophan rather than stop, AGA and AGG are stop rather than arginine, and AUA is methionine rather than isoleucine.
“At least 33” is doing honest work. That is how many have been catalogued and numbered, not how many exist. It is a count of a list, and the list has grown.
This is the sharpest evidence for something No. 7 argued — that the mitochondrion is a former free-living bacterium that “never quite stopped being itself.” It kept its own DNA and its own ribosomes; it also kept its own reading of the table. The human mitochondrial genome encodes 13 proteins, and those 13 are read by a different code from everything else in the same cell.
So “the universal genetic code” is not universal inside one human being, which is a sentence worth saying slowly. The usual phrase survives because table 1 covers the overwhelming majority of genes in the overwhelming majority of organisms — and because a rule is allowed to be a rule with exceptions, provided somebody says where they are.
What the code actually says
People say the genetic code as if it meant the instructions for making you. As if somewhere in there it said what you would be like.
It does not say that. It has never said that.
It says which bead comes next. That is all it says.
It does not say what you will be good at.
It does not say what you will love, or what will be hard, or what you will be like to be around.
It does not say who you are.
It is a list for putting beads on a string, one at a time, in order.
And even that list has more than one right answer nearly everywhere you look.
This spread is in the child’s layer deliberately, and it is the one departure from No. 105. That zine protected itself by never saying what DNA does. This one cannot — the code is what it does — so the protection has to come from accuracy instead: the popular meaning of “genetic code” is not a simplification of the real one, it is a different claim, and correcting it is the honest move rather than a reassuring one.
Everything on this page is literally true of the table. The genetic code is a mapping from 64 triplets to 20 amino acids plus stop. There is no entry for a trait. How bodies actually develop involves regulation, timing, environment, chance and a great deal more besides sequence — and none of that is in this table either.
The blueprint metaphor is the usual carrier of the misreading, and it has been argued against at length in the literature — the objection being that a blueprint specifies a finished object deterministically and leaves no room for the developmental process that actually does the building. Cited as a lead, not read: we have identified but not read Pigliucci’s 2010 paper on the end of the genes-as-blueprint metaphor, so it is named here and nothing in this zine rests on it.
What this one refuses
This page is for you rather than for the child. Its companion, No. 105, could stay safe by never saying what DNA does. This one walks straight into the subject where this genre does its damage, so it owes you a plainer account of where it stops.
Sixty-four ways of writing, twenty things to write. Nearly everything alive is spelled more than one way, and all the spellings are right.
Can you think of another thing that has more than one right spelling?
Why do you think the last letter is the one that gets to be different?
If two lists inside you disagree about a word, how does anything work at all?
What would happen if there were sixty-four beads instead of twenty?
Where all of this comes from
Every codon in this zine, and the two tables on spread nine. The NCBI translation tables — table 1, “The Standard Code”, and table 2, the vertebrate mitochondrial code. Read directly. Source of every codon-to-amino-acid assignment used here, the count of 33 tables, and the three differences drawn in figure four. Nothing in this zine was written from memory, which matters more than usual because the demonstration on spread seven fails if a single row is wrong.
Spread eight, and its hedge. S. J. Freeland & L. D. Hurst, “The genetic code is one in a million,” J. Mol. Evol. 47, 238–248 (1998), PMID 9732450. Not read — PubMed refused the request. Its result, and the criticism printed beside it, both come from an open-access review that cites it and that we did read. F. H. C. Crick, “The origin of the genetic code,” J. Mol. Biol. 38, 367–379 (1968) is likewise quoted at second hand, from that review.
One paper named and not used. M. W. Nirenberg & J. H. Matthaei, “The dependence of cell-free protein synthesis in E. coli upon naturally occurring or synthetic polyribonucleotides,” PNAS 47, 1588–1602 (1961), PMID 14479932 — the experiment that read the first codon. Read, after we had said it could not be. This page shipped saying the paper was held only as scanned images and that a spread on it had been cut for that reason. That was wrong: PubMed Central’s reader serves page images, but the PDF at Europe PMC carries a text layer, and the result is in it — “the addition of 10 µg of polyuridylic acid per ml of reaction mixture resulted in a remarkable stimulation of C¹⁴-L-phenylalanine incorporation,” and “no other polynucleotide tested could replace polyuridylic acid.” One blocked route is not unreachable. The spread stays cut, because twelve is the right length for this zine and the argument does not need it — but it is cut by choice now rather than by a wall that was not there.
How this one is built
Two layers on every spread. Large plain type for the child, first in the document, so a screen reader meets it first — then a bordered box for the adult, holding the citations, the exact wording and every place the evidence runs out. Pictures sit between the two. The refusals on spread eleven are addressed to you, not to the child.
Spread ten is the one departure from this collection’s usual caution, and it was a deliberate call. Its companion protected itself by never saying what DNA does; this zine cannot, so the correction is made in the child’s own layer in plain type rather than kept in a box the child never hears.
What we left out, and what is still open
Two zines running have ended on an exception — a virus in pond water, and now a disagreement between two lists inside one body. That is not a habit, and it is worth saying why. A rule stated as universal earns the question where does it give?, and both times the answer was already published and interesting. When a third piece reaches for the same shape without the exception being the best thing in the material, that will be the time to stop.
Open, and said on the page. Why the code has the shape it has is unsettled — frozen accident, error minimisation, or both at different stages — and spread eight leaves it unsettled rather than choosing the flattering answer. Reading frames and frameshifts are named on spread three and deliberately not pursued. And synonymous changes are not always consequence-free in a living cell, which spread six says in its grown-up box rather than letting the child layer overreach.
For grown-ups who want the rest of it. You Were Never One Thing is where the mitochondria on spread nine come from — a bacterium that moved in and never entirely stopped being itself. It is not written for a child, and this zine borrows nothing from it into the read-aloud layer.