Documented real-world cases
The mechanism is the documented cause of named errors on both clocks, in both directions — none of them clock (Axis-A) failures. Older bias: living Nevada snails at ~27 ka; Bishop Tuff cores ~850–892 ka vs ~767 ka eruption. Younger bias: Vindija bone climbing ~29 → 32 → >48 ka. Each is fixed by an age-independent screen.
The mechanism is the documented cause of named errors on both clocks and in both directions—none of them clock (Axis-A) failures. Older bias: living Nevada snails dated ~27 ka (3.3 pMC), Bishop Tuff cores ~850–892 ka vs ~767 ka eruption, inherited detrital populations. Younger bias: Vindija bone climbing ~29 → 32 → >48 ka up the pretreatment ladder; Pb-loss in metamict zircon. Each is fixed by an age-independent screen, not by re-litigating the clock.
The mechanism of §3 is not hypothetical. It is the documented cause of specific, named errors in the published record, on both clocks and in both directions of bias. Table 3 collects representative cases. In every one the resolution is the same discipline the protocol prescribes: classify or screen on age-independent grounds, then validate out-of-sample or by consilience. None of these is an Axis-A (clock) failure; all are Axis-B (attribution / open-system) failures, exactly as the two-axis decomposition predicts.
| Case (material) | System | Bias | Magnitude | Age-independent screen / fix |
|---|---|---|---|---|
| Living snails (shell) | Nevada springs | older | ~27 ka apparent on living shells (3.3 pMC)[13] | dead-carbon reservoir; date discarded |
| Living river mollusks | freshwater rivers | older | living ~2.3 ka; river-shell errors to several ka[12] | humus/limestone C source; carbonate de-prioritised |
| Marine shell/coral | global ocean | older | ~400–500 yr global; regional ΔR[15] | live-collected known-age; Marine20 + ΔR |
| Old-wood charcoal | East Polynesia | older | chronologies varied by >1000 yr; settlement moved ~4 centuries younger[17,18] | restrict to short-lived material (rat-gnawed seeds) |
| Neanderthal bone | Vindija G1 | younger | legacy ~29 ka, ultrafiltration ~32 ka, hydroxyproline >48 ka[22,21] | single-amino-acid (hydroxyproline) AMS |
| Zircon cores | Lava Creek Tuff | older | cores 668.8 ka vs faces 626.5 ka (eruption); +42 ka if pooled[3] | crystal position (faces vs cores) |
| Zircon (SIMS cores) | Bishop Tuff | older | core spots ~850–892 ka vs eruption ~767 ka[27,28] | growth zone / CL; rims date eruption |
| Detrital zircon | sedimentary strata | older | population inherited; only youngest cluster bounds deposition[30,31] | youngest concordant cluster (k_c ≥ 3) |
| Radiation-damaged zircon | metamict grains | younger | Pb loss ⇒ discordance, spuriously young | anneal + partial dissolution (CA-TIMS)[25] |
Radiocarbon, old carbon in (older bias). Snails living in southern-Nevada artesian springs returned apparent ages of ~27,000 yr—on living animals—purely because the groundwater carried dissolved “dead” carbon from Paleozoic carbonate (3.3% modern)[13]. [F] The half-century-old benchmark is the same: living river mollusks dated to ~2,300 yr, with river-shell errors as large as several thousand years, traced to ¹⁴C-deficient carbon from humus and dissolving limestone[12]. [F] The marine analogue is global—surface-ocean water carries a ~400–500 yr reservoir age with regional offsets ΔR, now propagated through the Marine20 calibration[15,14]. [F] These are precisely the homogeneous, self-concealing column of Table 1: invisible within a single date, correctable only by an externally estimated offset.
Radiocarbon, inbuilt age (older bias). The East Polynesia chronology is the textbook demonstration of screening. A meta-analysis of 1,434 dates found proposed settlement chronologies varying by more than a millennium; restricting the corpus to reliable short-lived samples—short-lived plant tissue and rat-gnawed seeds rather than charcoal of long-lived, decay-resistant wood—moved initial settlement about four centuries younger and resolved a coherent two-pulse colonization[17,18]. [F] This is Steps 1–2 of the protocol applied at the scale of a whole literature: classify material reliability before interpreting age.
Radiocarbon, young carbon in (younger bias—the opposite sign). Bone shows the failure ladder directly. For Vindija Neanderthals, legacy collagen gave ~28–29 ka; ultrafiltration raised it to ~32 ka; and dating the single amino acid hydroxyproline—a bone-specific biomarker—returned ages >48 ka, demonstrating that even the ultrafiltered dates were still contaminated by younger intrusive carbon[22,21]. [F] The same intrusive-carbon mechanism rendered the Zlatý kůň modern-human cranium spuriously young until it was resolved against an independent (genomic) constraint[23]. [I] These confirm the protocol’s insistence that accuracy must guard both directions, not only the older-bias case.
Zircon, old crystals in (older bias—but separable). The canonical Lava Creek separation (§7) is one instance; the Bishop Tuff is the sharpest. Core-targeted ion-microprobe dating of Bishop zircon yields ~850–892 ka, against a high-precision eruption age of ~767 ka from whole-grain CA-ID-TIMS and sanidine ⁴⁰Ar/³⁹Ar[27,28]. [F] This case makes the attribution point unavoidable: the older population is contamination if one wants the eruption age and signal (pre-eruptive residence) if one wants the storage history of the magma—the very same grains, two legitimate interpretations. The discriminator is age-independent (crystal position, CL zoning), never “pick the young grains.” [I] Basement-derived and recycled cores overgrown by eruption-age rims are documented in long-lived Andean systems such as Cerro Galán, where autocryst interiors predate eruption by up to several hundred kyr[29]. [F] Detrital populations are inheritance in the limit: the bulk is older source material, and only the youngest concordant grain cluster constrains deposition—a maximum depositional age, with the standing caution that abundant near-depositional grains or Pb loss can bias even that estimate young[30,31]. [F]
Zircon, daughter loss (younger bias). Episodic or continuous Pb loss from radiation-damaged (metamict) domains drives discordance and spuriously young apparent ages—the granular analogue of modern-carbon contamination in radiocarbon. This is the failure that motivated chemical abrasion (§10): anneal the lattice, then preferentially dissolve the damaged, Pb-loss domains, leaving closed-system zircon[25]. [F]
Across both clocks and both directions, the empirical record is consistent with the single claim of this paper: the accuracy bottleneck is foreign-material incorporation and open-system behaviour, it is recurrent rather than exceptional, and it is handled—never by re-litigating the clock—but by age-independent classification and out-of-sample validation.