10. Reading the Whole Log

Eight families, eight ledgers, sixty-odd rows of dated functions. Any engineer handed a repository like this one knows what to do next, and it is not to read another file. It is to squash the log — collapse every branch into one timeline and read the shape of the whole. Do that with this book's ledgers and something appears that no single chapter could show: the commits are not scattered. They cluster, in layers, and the layers tell a story none of the molecules tells alone.

The bottom layer is metabolism, and it is almost as old as life. Before anything said glutamate, cells were glutamate: the glutamate/α-ketoglutarate pair was moving nitrogen through metabolism before the last universal common ancestor, around four billion years ago, and histidine's biosynthetic pathway — the assembly line behind histamine — was complete on a similar schedule. GABA enters the same way, as a by-product of glutamate's own chemistry. The future vocabulary of the brain existed first as groceries.

The next layer up is stranger: the sensors came before the signals. The protein architectures that would one day detect these molecules — the cys-loop fold behind the GABA and glycine receptors, the channel scaffold behind the glutamate receptors, the transporter fold that serotonin's reuptake pump still uses — all trace to bacteria, and none of them, at that depth, was doing the job it does in you. A resurrected ancestral receptor from the metabotropic family binds glutamate tightly despite predating any synapse that could have used the fact. The hardware was lying around for hundreds of millions of years before the software shipped. Chapter three met the sharpest version of this lesson: a sea anemone's thirty-seven "GABA receptors," seven of them cloned and tested, and not one answering to GABA. Homology names the scaffold. It does not name the word spoken through it.

Then comes the great consolidation. In the window where animals arise and radiate — crown Metazoa somewhere between 650 and 834 million years ago, the bilaterian ancestor inside 600 to 760 — the scaffolds acquire their modern meanings, and they do it nearly all at once. The glutamate-receptor family radiates into its four subfamilies. The vesicular packaging that makes glutamate a sent signal rather than an ambient one dates to the same floor. The amine dictionary — the receptor families for dopamine, noradrenaline and its twin octopamine, the serotonin radiation — assembles by the bilaterian ancestor on the 2023 comparative genomics, though a 2026 deorphanization result would tug the monoamines a floor lower, and that dispute is live. Nearly every family chapter in this book leans on the same dating anchor there: the protostome–deuterostome split, which molecular clocks put at 642–761 million years and fossil calibration at about 558–575. Half the ledgers in this book wobble by a hundred million years on that one disagreement, and the honest reading is not to average it away but to say it plainly: the whole middle of the log is dated in rocks-versus-clocks stereo.

The fourth layer is the loudest event in the file. Around 550 to 450 million years ago, at the root of the vertebrates, two rounds of whole-genome duplication — 2R, the largest single commit in the log — copied the entire codebase twice, and the signalling systems kept the copies. The opioid system arrives wholesale here: four precursor genes and four receptors from one ancestral pair each, graded A on synteny and phylogeny together. The dopamine receptor inventory quadruples. The vasopressin-like ancestral peptide forks into vasopressin and oxytocin. The trace-amine receptors appear, the histamine H3/H4 pair splits, a sympathetic nervous system is wired down the trunk of something lamprey-shaped. Vertebrates did not invent new words. They duplicated the dictionary and let the copies drift into new meanings — and then, a hundred million years after noradrenaline, came the log's most instructive latecomer: adrenaline, whose synthesizing enzyme the jawless fishes never received. The scream travels down wiring far older than itself.

The shape of certainty

Squash the log a different way — by grade instead of by date — and a second pattern appears, and it is the one this book most wants you to take away. The A's and B's live at the bottom of the stack. The C's and D's live at the top. The pre-LUCA metabolic commits are the best-evidenced rows in the book; the receptor-family consolidations run solid B; and then, as the functions climb toward the ones that made these molecules famous — wanting, mood, trust, wakefulness — the grades thin out to D or to the ledgers' most honest entry of all, undatable as stated. Dopamine's reward-prediction computation has no confirmed date below the teleosts. Serotonin's mood-relevant role is a mechanism question wearing a timing question's clothes. Oxytocin's trust effect may not replicate at all. The pattern is not an accident of laziness; it is structural. Gene trees date parts. Behaviour is a performance of many parts, and performances leave no synteny. The log records what was written, not what it has come to mean — and the meanings, the layer we name the molecules for, are the least datable thing about them.

One family sits outside every layer, on purpose. Histamine's receptor scaffolds are ancient, but the binding of histamine itself was acquired convergently, at least three times, and its synthesizing enzyme exists as two unrelated families still being traded between bacteria by horizontal transfer. That is the control group: what a signalling molecule looks like when its history is not frozen — patchy, re-invented, borrowed. Its existence is what makes the rest of the cast remarkable. Reinvention is clearly possible. The frozen families are frozen anyway.

So the squashed log reads, bottom to top: groceries, then sensors, then packaging, then a dictionary, then a duplication that quadrupled it, then — undated, undatable, and most recent wherever it can be dated at all — everything we actually named the molecules for. Nothing in the stack was invented for a brain. The brain is the newest consumer of an API that was already old when the first neuron fired.

The log now says, as well as evidence can, when each line of the code was written. It cannot say why the file looks like this — why these eight, and not eight others; why nothing was ever rewritten. That question has three competing answers, and they are the last chapter's whole business.

Overturned if: the middle of the stack slides — if the placozoan and ctenophore deorphanization programs now under way land amine-specific receptors below the bilaterian line, the dictionary layer drops a floor, and the tidy sequence of scaffold-then-meaning loses its cleanest examples.

Sources

• 2015 · dos Reis, Thawornwattana, Angelis et al., "Uncertainty in the timing of origin of animals and the limits of precision in molecular timescales" — Current Biology (via Vol. I, ch. 09 — the anchor behind every animal-layer date above)
• 2023 · Goulty, Botton-Amiot, Rosato, Sprecher & Feuda, "The monoaminergic system is a bilaterian innovation" — Nature Communications
• 2024 · Moody, Álvarez-Carretero, Mahendrarajah et al., "The nature of the last universal common ancestor and its impact on the early Earth system" — Nature Ecology & Evolution (via Vol. I)
• 2024 · Edens, Stundl, Urrutia & Bronner, "Neural crest origin of sympathetic neurons at the dawn of vertebrates" — Nature
• Every row of the squashed log above compresses rows from the chapter ledgers, which carry the full sourcing; no date appears here that a family chapter does not source.