Dollo's Law

The evolutionary generalization that a complex trait once lost is overwhelmingly unlikely to re-evolve in identical form, because the genetic-developmental scaffolding that built it pseudogenizes faster than selection can reassemble it.

Core Idea

Dollo's law (Louis Dollo, 1893) is the evolutionary generalization that a complex trait once lost in a lineage is overwhelmingly unlikely to re-evolve in identical form, because the genetic and developmental scaffolding that originally built it decays faster than selection can reassemble it.

The mechanism is sequential and grounded in molecular genetics. A complex character — one assembled from many coordinated loci, regulatory elements, and interlocking developmental interactions — is maintained by purifying selection: every mutation that degrades a component is removed because the trait pays a fitness dividend. The moment the trait ceases to be expressed — because an ecological shift, colonization of a new habitat, or loss of the relevant predator or resource removes the selective premium — that purifying selection relaxes. Genes encoding structural proteins for the trait accumulate neutral substitutions; regulatory networks that once orchestrated its developmental cascade are co-opted for other functions or simply erode by genetic drift; the cascade itself frays at multiple nodes simultaneously. On the timescale of millions of years, formerly functional loci are converted by disabling mutations into pseudogenes — molecular fossils that record the existence of a once-working system but cannot reconstitute it. The key asymmetry is combinatorial: building the original trait required the rare historical event of assembling many genetic elements into a coherent working system, an event that natural selection could accomplish slowly over deep time; disassembling that system once selection is removed is rapid and effectively stochastic.

If a later selective pressure then favors a structurally similar function, the lineage cannot simply reverse. The original genetic-developmental architecture no longer exists as a coherent template from which selection can draw. Whatever evolves instead must be built from the surviving genomic material and developmental pathways currently available — which may yield a superficially convergent phenotype built through different genetic routes, different tissue sources, and different developmental logic. This is the distinction Dollo's law enforces: true reversal, in which the original pathway is reinstated from latent machinery, versus convergent re-evolution, in which a functionally similar but structurally distinct solution is built anew.

The law is probabilistic, not absolute, and its exceptions are informative. Atavisms — the transient developmental re-expression of ancestral characters in individual organisms, such as hind-limb buds in cetaceans or extra digits in horses — demonstrate that some developmental machinery can persist in a latent, partially intact state for millions of years. Documented evolutionary reversals, where a lineage genuinely recovers an ancestral complex phenotype, typically involve traits lost only recently (before pseudogenization is complete) or traits whose encoding genes were maintained by selection for a second function, keeping the machinery from decaying. Each genuine exception is diagnostic: it specifies how much of the developmental machinery remained recoverable and on what timescale — effectively calibrating the pseudogenization horizon, the approximate window after which reversal becomes mechanistically impossible regardless of selection strength. The existence of that horizon is what makes Dollo's law not merely a probabilistic claim about contingency but a statement about molecular kinetics: complex-trait irreversibility is grounded in the predictable rate at which unused genes accumulate disabling mutations and cease to be recoverable by selection.

Structural Signature

Sig role-phrases:

• the complex character — a trait assembled from many coordinated loci, regulatory elements, and interlocking developmental interactions, costly and rare to build
• the maintaining selection — purifying selection that holds the genetic-developmental scaffolding intact by removing every component-degrading mutation
• the loss event — an ecological shift that removes the trait's fitness premium and stops its expression, relaxing the purifying selection
• the decay process — once selection lifts, neutral substitution, co-option, and drift fray the scaffolding's nodes, converting functional loci into pseudogenes
• the assembly–disassembly asymmetry — building required rare slow combinatorial assembly under selection; teardown once selection is gone is fast and stochastic
• the pseudogenization horizon — the roughly predictable window after which the machinery has decayed past recovery and reversal is mechanistically foreclosed
• the reappearance verdict — a later similar phenotype is read as true reversal (same loci/cascade/tissue, machinery still latent) or convergent re-derivation (different parts, original architecture gone)

What It Is Not

• Not an absolute prohibition. The law is probabilistic, grounded in the rate at which unused machinery decays, not a physical impossibility. Before the pseudogenization horizon — when the loss is recent or the encoding genes are held intact by a second function — reversal is improbable but mechanically possible, and documented cases exist; the claim is overwhelming unlikelihood of identical re-evolution, not a ban.
• Not a bar on re-evolving the function. Dollo's law rules out reinstating the original genetic-developmental pathway, not the reappearance of a similar phenotype. A lineage may well re-acquire a structurally similar capacity by convergent re-derivation — built from different loci, tissues, and developmental logic — and that is fully consistent with the law, which forbids only the identical reconstitution.
• Not a law of physics. The asymmetry is not thermodynamic necessity but combinatorial improbability under molecular kinetics: assembling many coordinated elements into a working system is the rare slow event, dismantling it once purifying selection lifts is fast and stochastic. It is a statistical generalization about genomes, calibrated by a decay timescale, not a conservation principle that holds with the force of natural law.
• Not refuted by atavisms or documented reversals. A hindlimb bud in a whale or a recovered ancestral trait is diagnostic, not a counterexample: each genuine exception measures how much machinery stayed recoverable and for how long, calibrating the pseudogenization horizon rather than overturning the rule. The exceptions specify the window in which reversal remains possible; they confirm the kinetics the law describes.
• Not Dollo parsimony, the tree-scoring rule. "Dollo parsimony" is a phylogenetic reconstruction method that permits a trait to be gained once but lost many times — a notational convention for inferring histories. The law is the underlying biological claim about why complex traits are rebuilt anew; the parsimony rule is one analytic tool that encodes its asymmetry, not the law itself.

Scope of Application

Dollo's law lives across the historical and molecular subfields of evolutionary biology; its reach is bounded by that domain, because its predictive teeth — a decay timescale read off pseudogenization kinetics — exist only where the substrate is a genome under selection. Within that boundary it recurs as a reasoning template wherever a complex character has been lost and might reappear.

• Paleontology and macroevolution — its home setting: the adjudication of whether a complex character reappearing in the fossil record (limbs in snakes, teeth in birds, eyes in cave fish) is a true ancestral homology or a convergent re-derivation built from different parts.
• Developmental and evolutionary-developmental biology (evo-devo) — frames the study of latent developmental machinery: which silenced gene-regulatory networks survive in recoverable form (probed by atavism frequency), and how fast unused developmental cascades fray after expression stops.
• Molecular evolution and pseudogenization — supplies the kinetic mechanism behind the law: loss of expression removes purifying selection on the encoding loci, which then accumulate disabling mutations at a roughly predictable rate, defining the irreversibility horizon as a literal molecular-decay prediction.
• Phylogenetic reconstruction methods — "Dollo parsimony," a tree-scoring rule that permits a complex trait to be gained once but lost many times, encodes the law's gain–loss asymmetry as an analytic convention for inferring evolutionary histories of elaborate characters.
• Conservation biology — applies the horizon to management timescales: a population that has lost a behavior or capacity cannot recover it once the supporting genetic machinery has decayed, so the window of recoverability bounds what restoration can hope to reinstate.
• Regenerative medicine — frames the comparative question of why mammals have lost regenerative capacities (limb regeneration) that newts retain, reading the loss as decayed-past-recovery developmental architecture rather than a merely dormant one.
• De-extinction and breed-back programs — bounds the achievable: such projects can recover only phenotypes whose genetic substrate still survives in extant relatives; a trait whose encoding machinery is fully pseudogenized cannot be resurrected, only convergently re-approximated.

Clarity

Naming Dollo's law makes a recurring paleontological puzzle answerable: when a complex character reappears in a descendant lineage, the practitioner can now ask the sharp question — is this the ancestral trait recovered, or a new structure built on the old niche? Before the law, a reappearing limb, tooth, or eye in the fossil record invited an undisciplined inference of "reversion"; after it, the biologist is forced to separate true reversal, in which the original genetic-developmental pathway is reinstated from latent machinery, from convergent re-evolution, in which a functionally similar phenotype is assembled anew from different loci, tissues, and developmental routes. That distinction is what keeps a homology from being mistaken for a re-derivation, and it is testable: the two leave different molecular and developmental signatures.

The law also dissolves a confusion between contingency and mechanism. Evolution's irreversibility could be read as a soft statement about historical luck — too many forks to retrace. Dollo's law instead grounds the asymmetry in molecular kinetics: once expression stops and purifying selection relaxes, the encoding loci pseudogenize at a roughly predictable rate, so there is a pseudogenization horizon — a window after which reversal is mechanistically impossible regardless of how strongly selection later favors the trait. This converts a vague "easy to lose, hard to regain" intuition into a question with a timescale attached, and reframes the law's exceptions as diagnostic rather than embarrassing: each documented reversal (a recent loss, or a gene kept intact by a second function) measures how much machinery stayed recoverable and for how long, calibrating the horizon rather than refuting the rule.

Manages Complexity

The macroevolutionary record offers a bewildering catalogue of traits that vanish from lineages and characters that reappear in descendants — limbs in snakes, eyes in cave fish, teeth in birds, mandibular dentition resurfacing in frogs — each case otherwise demanding its own paleontological adjudication of homology versus convergence. Dollo's law compresses that case-by-case puzzling into a single asymmetry with a clock attached: build-up of a complex character is a rare, slow, combinatorial assembly under purifying selection, while its disassembly once expression stops is fast and stochastic, and the genes decay at a roughly predictable rate. The analyst no longer re-derives reversibility trait by trait but tracks a handful of quantities — how complex (how many coordinated loci) the lost character was, how long ago it was lost relative to the pseudogenization horizon of a few million years, and whether any encoding genes were held intact by a second function — and reads off the qualitative verdict: reversal still possible, or foreclosed; a reappearance to be expected as true homology, or as convergent re-derivation from different parts. The deep "is evolution reversible?" question collapses to a question about the persistence kinetics of unused genetic machinery, with the law's exceptions not refuting the rule but calibrating exactly where on that timescale the horizon falls.

Abstract Reasoning

Dollo's law equips the paleontologist and molecular evolutionist with a small set of inferences, all turning on the asymmetry between slow combinatorial assembly and fast stochastic decay.

Diagnostic — homology versus convergence from a reappearance. When a complex character reappears in a descendant lineage, infer its history from the molecular and developmental signature rather than the gross morphology. If the trait is rebuilt through the same loci, the same regulatory cascade, and the same tissue source as the ancestral version, infer a true reversal from latent machinery — which in turn implies the loss was recent (pre-pseudogenization) or the encoding genes were held intact by a second function. If instead the convergent phenotype is assembled from different loci, different developmental logic, or different tissues, infer convergent re-derivation, and conclude that the original architecture had already decayed past recovery. The surface phenotype is mute on this question; the genetic-developmental route is what you reason from, and the homology-versus-convergence verdict is what you reason to.

Diagnostic — depth of loss from atavism frequency. An atavism (a hindlimb bud in a whale, an extra digit in a horse) is a probe of how much ancestral machinery survives in latent, partially-intact form. From the recurrence of an atavistic phenotype across individuals, infer that the developmental cascade for that trait, though normally silenced, remains substantially recoverable; from the complete absence of any atavism for a long-lost trait, infer that the cascade has frayed past the point where even individual-level re-expression is possible. The atavism is the readout; the latent-machinery state is the hidden variable inferred.

Interventionist / predictive — foreclosure from elapsed time. The pseudogenization horizon converts elapsed-since-loss into a prediction. Given a trait's complexity (how many coordinated loci it required) and the time since its expression ceased, predict whether restored selection could reinstate it: before the horizon of a few million years, reversal is improbable but mechanically possible; after it, reversal is foreclosed regardless of how strongly later selection favors the function, and any similar phenotype that arises must be convergent. The lever here is not something the analyst changes but something nature has already set — yet the reasoning is interventionist in form: it predicts the outcome of the counterfactual "if selection now favored the lost trait."

Boundary-drawing — which regime a case occupies. Decide whether a given lost trait sits in the reversible regime (recent loss, or encoding genes maintained by a second function, keeping the machinery from decaying) or the foreclosed regime (old loss, no second-function rescue, machinery pseudogenized). The diagnostic is the joint state of three quantities — trait complexity, time since loss relative to the horizon, and second-function maintenance of the genes — and the boundary it draws determines whether a future reappearance should even be entertained as homologous.

A unifying move underlies all four: reason from the kinetics of unused machinery rather than from the phenotype. Because purifying selection is the only thing holding the genetic-developmental architecture together, its removal starts a clock, and every inference Dollo's law licenses is a reading of where on that clock the trait now stands.

Knowledge Transfer

Within evolutionary biology Dollo's law transfers as mechanism, and its currency is the same wherever the substrate is a genome under selection. To paleontology and macroevolution — its home — it supplies the homology-versus-convergence adjudication for any reappearing complex character in the fossil record (limbs, teeth, eyes), with the diagnostic intact: read the molecular and developmental route, not the gross morphology. To molecular evolution it is the pseudogenization clock itself — loss of expression removes purifying selection, and the encoding loci decay at a roughly predictable rate — so the law's horizon is literally a kinetic prediction usable on any silenced gene family. To evo-devo it transfers as the question of latent developmental machinery: which silenced gene-regulatory networks survive recoverably (probed by atavism frequency), and how fast unused cascades fray. To conservation biology, regenerative medicine, and de-extinction the transfer is still mechanistic, because all three operate on real genomes: a population that has lost a behavior or a regenerative capacity (limb regeneration retained by newts, lost by mammals) cannot recover it on management-relevant timescales once the supporting machinery has decayed; breed-back projects can only recover phenotypes whose genetic substrate survives. Across all of these the vocabulary — pseudogenization, purifying selection, latent pathway, the decay horizon — carries without translation, because the cargo is genetic kinetics and the substrate is always a genome.

Beyond the genome the transfer is best characterized as mixed, leaning analogy. The recurrent shape — a complex integrated system is cheap to dismantle once the pressure maintaining it is removed, and the supporting substrate decays faster than it can be reassembled, so what later arises must be built anew from surviving parts — does recur across human substrates: institutional traditions once abandoned cannot be revived in identical form; a language lost cannot be re-spoken with its original phonology; a technology whose tooling and tacit craft are gone must be re-invented rather than resumed. These are illuminating, but they are analogy: nothing in them is a gene, there is no pseudogenization rate, no purifying selection, no atavism, and the predictive teeth of the law — a timescale read off molecular decay kinetics — do not transfer, because the rate of "decay" of a lost institution or craft is set by entirely different and unquantified processes. The honest reading is the (B) one: what genuinely recurs across these domains is the parent — irreversibility of a complex integrated structure once its maintaining conditions lapse, with path_dependence and tacit-knowledge loss riding alongside — and that more-general pattern is what the cross-domain lesson should carry. "Dollo's law," as named, packs domain-bound cargo (the pseudogenization horizon, the homology-vs-convergence test, the eponymous paleontological framing) that does not and should not travel; invoking it for a lost institution renames the components (genome → tradition, pseudogene → forgotten practice) and borrows the asymmetry while dropping the molecular kinetics that make the original a quantitative law rather than a proverb. See Structural Core vs. Domain Accent.

Examples

Canonical

The loss of teeth in birds is the textbook illustration of the law with modern molecular grounding. Louis Dollo articulated the generalization in 1893 from the fossil record, but the tooth case shows its kinetics directly. Birds descend from toothed theropod dinosaurs, yet no living bird has true mineralized teeth. Sequencing of bird genomes has found that the genes underlying enamel and dentin — such as the enamel matrix genes — persist only as pseudogenes: molecular fossils riddled with disabling mutations. A comparative study across birds (Meredith and colleagues, 2014) read the shared pattern of these dental-gene lesions as evidence of a single loss of mineralized teeth in the common avian ancestor over 100 million years ago. Because that machinery has long since decayed, the occasional "egg tooth" or serrated bill of a bird is convergent, never a reinstated ancestral tooth.

Mapped back: Teeth are the complex character; the shift to a toothless, beaked feeding mode was the loss event that lifted the maintaining selection; the accumulation of dental pseudogenes is the decay process, now far past the pseudogenization horizon. The reappearance verdict for any toothlike bird structure is therefore forced to convergent re-derivation, not true reversal.

Applied / In Practice

The Quagga Project in South Africa, begun in 1987 by Reinhold Rau, applies the law's boundary to conservation breeding. The quagga — a plains-zebra subspecies with reduced striping and a brownish rear — was hunted to extinction, the last individual dying in an Amsterdam zoo in 1883. Because the quagga was genetically a plains zebra, the genetic variation underlying its distinctive coat pattern is thought to survive within living plains-zebra populations. The project selectively breeds plains zebras that show reduced striping, generation by generation, toward the quagga-like phenotype. It works only because the loss was recent and the substrate persists in extant relatives; no such program could resurrect a deeply extinct lineage whose supporting genes had fully pseudogenized.

Mapped back: The quagga coat pattern is the complex character lost recently enough to sit inside the pseudogenization horizon, its substrate un-decayed. The breeding program tests the reappearance verdict: recovering a phenotype whose genetic material still survives in relatives is achievable, whereas a trait past the decay process horizon could only be convergently re-approximated, never truly reinstated.

Structural Tensions

T1: Probabilistic rule versus law-like force (a "law" that its own exceptions are built to accommodate). Dollo's law is named and treated as a law, and its kinetic grounding — a predictable pseudogenization rate defining a foreclosure horizon — gives it genuine law-like teeth beyond mere contingency. But it is explicitly probabilistic, and it absorbs its exceptions (atavisms, recent-loss reversals, second-function rescues) as calibrations rather than refutations. This is methodologically defensible, yet it creates a standing tension: a generalization that reclassifies every apparent counterexample as data confirming its timescale risks being unfalsifiable in practice, because any reversal is explained as "loss was recent enough" or "genes were maintained." The law's strength (it predicts a horizon) and its slipperiness (it explains away breaches of the horizon by adjusting the horizon) are the same feature. The discipline that keeps it a law rather than a proverb is holding the pseudogenization rate to independent molecular measurement, not fitting it post hoc to each exception. Diagnostic: Is the horizon in this case set by independently measured decay kinetics, or is it being back-fitted to whatever reversal or non-reversal was observed?

T2: Molecular route versus surface phenotype (the diagnostic depends on evidence the fossil record rarely preserves). The homology-versus-convergence adjudication is the law's sharpest service — and it insists the gross morphology is mute, that one must read the loci, cascade, and tissue source. This is exactly right in principle but collides with the law's home domain: paleontology works largely from morphology, and the molecular and developmental signature that would settle reversal-versus-convergence is often unavailable for extinct lineages whose genomes are gone. The tension is that the law disqualifies the evidence most abundant in the fossil record (phenotype) in favor of evidence frequently absent (genetic route), so the very cases the law was built to adjudicate — deep-time reappearances — are often the ones where its decisive test cannot be run. Its diagnostic power is greatest where its diagnostic data are scarcest. Diagnostic: Is there actual molecular or developmental evidence of the rebuilding route here, or is a homology-versus-convergence verdict being asserted from morphology alone — which the law itself declares insufficient?

T3: Decay foreclosure versus latent-machinery persistence (two timescales the law must hold together). The law's engine is fast stochastic decay of unused machinery past a horizon of a few million years. Yet atavisms demonstrate that some developmental cascades persist in recoverable, latent form for far longer than naive pseudogenization would predict — hindlimb machinery re-expressible in whales tens of millions of years after limb loss. The tension is internal: the same law asserts both that unused machinery decays predictably fast and that it can persist latently long enough to resurface, and these pull against each other. The resolution — that different components decay at different rates, and second functions or developmental entanglement preserve some — is real but means the "horizon" is not a single clock but a smear of component-specific rates, undermining the clean timescale that gives the law its predictive appeal. Persistence and decay are both true, and the law's precision depends on knowing which dominates for a given trait. Diagnostic: For this trait, is the supporting machinery on the fast-decay clock (foreclosed) or preserved by entanglement or second function (latently recoverable) — and does the case rest on a single horizon or a mix of component rates?

T4: Assembly–disassembly asymmetry versus convergence's own repeatability (irreversibility of the pathway coexisting with predictability of the function). The law's core is a stark asymmetry: building is rare, slow, combinatorial; tearing down is fast and stochastic, so the original pathway is not recoverable. But this sits beside the striking evolutionary fact that functions re-evolve convergently again and again (eyes dozens of times, streamlined body forms repeatedly), which means the phenotypic outcome is in some sense highly repeatable even though the genetic pathway is not. The tension is that Dollo's law emphasizes historical irreversibility and lost uniqueness of the pathway, while convergence emphasizes the deterministic re-attainability of the function — so the same lost trait is both "gone forever" (the architecture) and "readily rebuilt" (the capability). The law is precise only if one keeps pathway-identity and functional-outcome rigorously separate, and the interesting biology often lives in how much developmental route is constrained even in convergence. Diagnostic: Is the claim about the irrecoverability of the original pathway (Dollo's actual domain) or about the function, which convergence may re-derive predictably — and are the two being conflated?

T5: Autonomy versus reduction (an evolutionary law or an instance of irreversibility of a maintained complex structure). "Dollo's law" is a specific evolutionary-biology construct with home-bound cargo — the pseudogenization horizon, the purifying-selection mechanism, the homology-versus-convergence molecular test, the atavism probe, the eponymous paleontological framing — and within evolutionary biology it travels intact as mechanism across paleontology, molecular evolution, evo-devo, conservation, regenerative medicine, and de-extinction, which are co-instances on one genomic substrate. But its portable core is the parent irreversibility of a complex integrated structure once its maintaining conditions lapse (with path_dependence and tacit-knowledge loss riding alongside): a system is cheap to dismantle once the pressure holding it lapses, and its substrate decays faster than it can be reassembled, so what later arises is built anew. That parent genuinely recurs — abandoned institutions, lost languages, forgotten crafts — but as analogy, not mechanism, because none contains a gene, a pseudogenization rate, or purifying selection, and the law's predictive teeth (a timescale from molecular kinetics) do not transfer. Diagnostic: Resolve toward the parent (irreversibility of a maintained complex structure, with path_dependence) when carrying the lesson to institutions, languages, or technologies; toward Dollo's law's pseudogenization-horizon-and-homology-test machinery only where the substrate is an actual genome under selection.

Structural–Framed Character

Dollo's law sits toward the structural end of the spectrum but stops short of the pole — best read as mixed-structural: a genuine natural-kinetic mechanism wearing heavy molecular-genetic vocabulary. On four of the five criteria its structural credentials are strong. Its evaluative weight is nil — a lineage losing a trait and its scaffolding pseudogenizing is neither good nor bad; "irreversible" here is a kinetic fact, not a verdict, and the law praises and blames nothing. It is not human-practice-bound — strip away every paleontologist and birds still carry dead enamel genes, whale hindlimb machinery still lingers latently, unused loci still accumulate disabling mutations at their own rate; the mechanism runs on genomes and clocks, not on a judging agent. Its institutional origin is none: the assembly–disassembly asymmetry is a fact of how purifying selection maintains a complex character and how molecular decay dismantles it once expression stops, not an artifact of any survey or convention — Dollo named a regularity nature already ran, the way one names rather than invents. And within its proper range cross-domain reuse is recognition, not import: moving from bird teeth to snake limbs to lost regeneration to de-extinction, the same pseudogenization-and-decay mechanism is recognized intact on a shared genomic substrate, not borrowed as a frame. These four marks place it firmly on the structural side.

What keeps it off the structural pole is the remaining criterion, vocab-travels, which it fails. The operative vocabulary — pseudogenization, purifying selection, latent developmental cascade, atavism, the decay horizon, homology-versus-convergence — is irreducibly genetic and does not float free of genomic substrates the way "growing quantity" or a decay rate does in a pure structural prime; within evolutionary biology it carries full content, but beyond it "a lost institution can't be revived in identical form" keeps only the bare dismantle-fast/rebuild-anew shape and renames every component (genome → tradition, pseudogene → forgotten practice), so the off-domain transfer is analogy, not mechanism — and the predictive teeth, a timescale read off molecular kinetics, do not survive the crossing. The portable structural skeleton is irreversibility of a complex integrated structure once its maintaining pressure lapses — cheap to dismantle, its substrate decaying faster than it can be reassembled, so what later arises is built anew from surviving parts. That skeleton is genuinely portable, and it is exactly what Dollo's law instantiates from the parents the entry names — irreversibility, with path_dependence riding alongside; the cross-domain reach belongs to those parents, while the pseudogenization horizon, purifying-selection mechanism, and homology test distinctive to "Dollo's law" are the domain accent that stays home. Its character: structural in skeleton — a real, evaluatively neutral, recognized-in-nature irreversibility mechanism grounded in decay kinetics — but stated in molecular-genetic vocabulary that pins it to genomes under selection, leaving it mixed-structural rather than a free-floating prime.

Structural Core vs. Domain Accent

This section settles why Dollo's law is a domain-specific abstraction and not a prime, and it carries the reasoning for that placement in the same stroke — isolating the thin irreversibility skeleton that could lift from the molecular-genetic body that cannot.

What is skeletal (could lift toward a cross-domain prime). Strip the genome away and a spare relational structure survives: a complex integrated structure, once the pressure that maintained it lapses, is cheap and fast to dismantle, and its supporting substrate decays faster than it can be reassembled, so what later arises must be built anew from surviving parts rather than reinstated. The portable pieces are abstract — an assembled system held together only by a maintaining pressure, an asymmetry between slow combinatorial assembly and fast stochastic teardown, a decay of the substrate, and a foreclosure past which reversal is impossible and any similar outcome is a fresh convergent build. That skeleton is genuinely substrate-portable, which is exactly why it recurs in the catalog as the general primes Dollo's law instantiates: irreversibility of a maintained complex structure, with path_dependence riding alongside. This is the core the law shares with lost institutions and forgotten crafts, not what makes it Dollo's law.

What is domain-bound. Almost everything that makes the concept Dollo's law in particular is evolutionary-genetics furniture, and none of it survives extraction intact: purifying selection as the maintaining pressure; pseudogenization — the conversion of unused loci into disabled molecular fossils — as the decay process; the pseudogenization horizon, a foreclosure window read off a molecular decay rate; atavisms as probes of latent developmental machinery; the homology-versus-convergence test that reads a reappearance off loci, regulatory cascade, and tissue source; and the second-function rescue that keeps encoding genes intact. These are the mechanisms and instruments the subfield actually measures, and each is welded to a genome under selection. The decisive test: remove the genome and the pseudogenization clock, and "once dismantled it cannot be revived in identical form" is no longer Dollo's law but a bare irreversibility claim — a proverb about lost institutions — because the predictive teeth, a timescale from molecular kinetics, are precisely what does not survive the crossing.

Why this does not clear the prime bar. A prime's vocabulary travels and its transfer is recognition of the same mechanism, not analogy. Dollo's law's transfer is bimodal. Within evolutionary biology — paleontology, molecular evolution, evo-devo, conservation biology, regenerative medicine, de-extinction — it travels intact as genuine mechanism-recognition: all of these operate on real genomes, so pseudogenization, purifying selection, the latent pathway, and the decay horizon carry without translation, and the law's horizon is literally a kinetic prediction usable on any silenced gene family. Beyond the genome — abandoned traditions that cannot be revived identically, lost languages that cannot be re-spoken with their original phonology, forgotten crafts that must be re-invented — invoking "Dollo's law" is analogy: it renames the components (genome → tradition, pseudogene → forgotten practice) and borrows the assembly-disassembly asymmetry while dropping the molecular kinetics that make the original a quantitative law rather than a proverb. And when the bare structural lesson is wanted cross-domain, it is already carried, in more general form, by the parents the entry instantiates: irreversibility of a maintained complex structure, with path_dependence. The cross-domain reach belongs to those primes; "Dollo's law," as named, packs the pseudogenization horizon, the purifying-selection mechanism, and the homology-versus-convergence test that should stay home.

Relationships to Other Abstractions

Current abstraction Dollo's Law Domain-specific

Parents (2) — more general patterns this builds on

• Dollo's Law is a kind of Irreversibility Prime

  Dollo's law is irreversibility specialized to the post-loss decay of complex genetic-developmental scaffolding.

• Dollo's Law is a kind of Path Dependence Prime

  Dollo's law is path dependence specialized to evolutionary accessibility changing with the lineage's trait-loss and scaffold-decay history.

Hierarchy paths (4) — routes to 4 parentless roots

• Dollo's Law → Irreversibility → Reversibility and Irreversibility

Show alternative paths (3)

Not to Be Confused With

• Convergent evolution. The independent re-derivation of a functionally similar phenotype in a lineage that had lost (or never had) the ancestral one — built from different loci, tissues, and developmental logic (an "egg tooth" that is not a reinstated theropod tooth; a cave-fish structure that is not the recovered ancestral eye). Far from contradicting Dollo's law, convergence is exactly the verdict the law forces once the original architecture has pseudogenized: not true reversal, but a fresh build. Tell: does the reappearing trait run through the same ancestral loci/cascade/tissue (true reversal, machinery still latent) or through demonstrably different genetic-developmental routes (convergence)? The molecular route decides, not the morphology.
• Atavism. The transient developmental re-expression of an ancestral character in an individual organism — a hindlimb bud in a whale, an extra digit in a horse — betraying that some silenced machinery persists in latent, partly-intact form. This is an individual-level, one-off developmental event and a probe of how much machinery survives, not a lineage-level evolutionary recovery of the trait. Dollo's law is the macroevolutionary claim about whole lineages; atavisms are its diagnostic readout that calibrates the horizon. Tell: is a stray ancestral structure appearing sporadically in single individuals (atavism, a probe), or has a whole lineage stably re-acquired the trait under selection (the reversal the law addresses)?
• Vestigial structure. An ancestral trait still present but reduced, degenerate, or repurposed (the human appendix, whale pelvic bones) — retained, not lost. Dollo's law concerns traits that have been fully lost and expression halted, and whether they can be rebuilt. A vestige is a trait mid-decay that never disappeared; the law's subject is the trait already gone and the question of its return. Tell: is the structure still there in reduced form (vestigial), or absent and possibly reappearing later (Dollo's domain)?
• Pseudogenization (the molecular process). The decay of an unused gene into a disabled "molecular fossil" once purifying selection lifts — the kinetic mechanism underneath Dollo's law. It is a process operating on individual loci; Dollo's law is the macroevolutionary generalization about complex-trait irreversibility that this process grounds and clocks. Tell: pseudogenization names why a single silenced gene rots at a predictable rate; Dollo's law is the lineage-level irreversibility claim built on the aggregate of that rot across a whole trait's scaffolding.
• Irreversibility and path dependence (the parent primes it instantiates). The substrate-neutral skeleton — a complex integrated structure, once its maintaining pressure lapses, is cheap to dismantle and its substrate decays faster than it can be reassembled, so what later arises is built anew — belongs to irreversibility (with path_dependence riding alongside), not to "Dollo's law." Dollo's law is the evolutionary-genetics specialization, adding purifying selection, the pseudogenization horizon, and the homology-versus-convergence test. Tell: carrying the lesson to a lost institution, a dead language, or a forgotten craft invokes irreversibility + path_dependence; "Dollo's law" applies only where the substrate is an actual genome under selection with a measurable decay clock. (Treated fully in earlier sections.)

Neighborhood in Abstraction Space

Dollo's Law sits in a moderately populated region (51^st percentile for distinctiveness): it has near-neighbors but no dense thicket of look-alikes.

Family — Unclustered & Miscellaneous (309 abstractions)

Nearest neighbors

• Wallace Effect — 0.86
• Haldane's Sieve — 0.85
• Baldwin Effect — 0.85
• Muller's ratchet — 0.84
• Fisher's Fundamental Theorem of Natural Selection — 0.84

Computed from structural-signature embeddings · 2026-07-12