Stellar and Galactic Ages | Berean Archive

Summary

This is not the distant-starlight problem, and confusing the two is the central mistake. Every published young-earth answer to distant starlight — light created in transit, a decaying speed of light, Humphreys' white hole, Hartnett's Carmelian cosmology, Lisle's anisotropic synchrony convention — works by changing how long the light took to get here. Not one of them changes anything about the object that emitted it. A stellar age is a statement about the object: the mass still sitting at a cluster's main-sequence turnoff, the surface temperature of a dead star that has been radiating stored heat, the thorium left in a halo star's atmosphere.
So the argument lands hardest on nearby objects, where the light-travel budget of a 6,000-year universe is not even strained. HD 140283 is 200.5 ± 0.3 light years away by Gaia parallax, and no one has ever published an age for it below about 11.4 billion years.¹² WD J2147−4035 is 91 light years away, has a surface temperature of 3,048 K, and has been cooling for 10.2 billion years.³ The Sun is eight light-minutes away and its ringing gives 4.60 ± 0.04 billion years without any radioactive decay in the calculation.⁴ Its light left this morning.
A young-earth astronomer drew the boundary himself, and it excludes exactly these objects. Ron Samec, writing in the Creation Research Society Quarterly in 2012: "all nearby objects partook in the same dilation effects as the earth during Creation Week and … they all have ages measured in earth time on the order of 10,000 years." He defines nearby as within about 3,000 light years.⁵ HD 140283 and every cool white dwarf discussed here are inside that line by a factor of fifteen or more.
A 6,000-year-old white dwarf would still be white-hot. Read straight off the published Montreal cooling tables, a 0.6 solar-mass white dwarf starts at 98,455 K and is at 97,900 K after 6,000 years.⁶ It has lost six-tenths of one per cent of its surface temperature. It takes 12.08 billion years to reach 3,000 K. The required error is a factor of two million, and the whole published systematic uncertainty budget for white dwarf cooling ages is about one billion years in ten.⁷
Four clocks that share almost no physics agree on the same objects. In NGC 6397 the white dwarf cooling age is 11.47 ± 0.47 Gyr and the main-sequence turnoff age is 11.6 Gyr on the same distance; the authors call it "a successful end-to-end (main sequence to white dwarf) comparison."⁸ In 47 Tucanae two eclipsing binaries weighed by Kepler's laws give 12.0 ± 0.5 Gyr.⁹ In M4 the isochrone mass, the eclipsing-binary mass and the asteroseismic mass of the red giants agree at 0.84–0.85 solar masses.¹⁰ And uranium decaying in the atmosphere of a halo star gives 12.5 ± 3 Gyr.¹¹
The main-sequence lifetime can be derived from measured inputs in five lines, with no stellar model and no cosmology. Take the mass from an eclipsing binary, the luminosity from a parallax and a photometer, and the fusion efficiency from atomic masses measured in a Penning trap. A metal-poor 0.80 solar-mass star lasts about 12 billion years. The arithmetic is in Part 3, and it can be checked on paper.
The young-earth literature has never engaged three of the four clocks. Danny Faulkner's 2019 audit grades 23 young-age astronomy arguments and has no heading for main-sequence turnoff ages, none for white dwarf cooling, and none for stellar radioactive dating.¹² Nine searches across 329 young-earth sites returned nothing on the white dwarf luminosity function, nothing on the Gaia crystallisation result, and nothing on the fact that independent clocks in one cluster agree. That is a statement about the literature, and anyone can re-run the searches.
What creationists do say is honest, and two of them say the hard part out loud. Faulkner in 2007: "Lacking an internal energy source, white dwarfs must tap their considerable store of internal heat and so they must cool as they age. … Did God create 'fossil' stars? Creationary geologists would object adamantly if this were suggested in the realm of fossils of biological origin. … Is the astronomical realm so different?"¹³ Samec in 2015: "I believe that God did not create 'dead' stars."¹⁴ Both then reach for gravitational time dilation — and the two cosmologies that supplied it have been abandoned by their own authors.
Mature creation survives, because it is not the kind of claim astronomy can touch. If God made stars already fusing, already with settled interiors, then they had positions on the Hertzsprung–Russell diagram from the first moment. Jason Lisle puts the sharpest version: "We don't know that God created each and every star with a 'full tank of gas.'"¹⁵ That is unfalsifiable and he says so. The article does not pretend to refute it. It asks the question Faulkner asked first: which side of his own line — created state, yes; created record of events that never happened, no — does a colour-magnitude diagram fall on, when four independent clocks inside it agree on one date?
The heavy elements record a sequence, not a snapshot. The alpha-to-iron ratio in a galaxy's stars falls at a metallicity that differs from galaxy to galaxy by nearly two orders of magnitude, and the differences track each galaxy's independently measured star-formation history.¹⁶¹⁷ The mean metallicity of neutral gas across the universe falls by a factor of ten from the present out to redshift 5.¹⁸¹⁹ A single creation event produces one composition, not a family of compositions indexed by history.
The JWST "impossibly early galaxies" story is the strongest live card the young-earth side holds, and it has to be told through 2026 rather than 2022. Lisle made a dated pre-launch prediction and its qualitative form came true.²⁰ But the most extreme candidate, photometric redshift 16.4, turned out to be at redshift 4.912.²¹ Four of the six famous Labbé candidates got spectra and all four came in lower. And the logic does not reach: a model saying galaxies assembled faster than expected still says they assembled 13 billion years ago.
Verdict: against the young-earth position, and not narrowly. The measurements are not model-dependent in the way the objection assumes. Every uncertainty in this literature is worth a few billion years out of ten to fourteen; the young-earth claim needs the answer wrong by a factor of a million. What survives is a theological position about initial conditions, held honestly by people who say plainly that it makes no predictions.

Part 1 — Two different problems

The problem everyone knows about

A galaxy is a billion light years away. Its light took a billion years to arrive. The universe is 6,000 years old. How did the light get here?

That is the distant-starlight problem, and young-earth astronomy has worked on it hard for seventy years. Young-universe cosmologies covers the eight published answers and what happened to each. The short version is that three are dead, one was retracted by its author in five dated statements, and the survivors reproduce standard observational cosmology exactly, by construction, with each author saying so in print.

Look at what all of those answers have in common. Light created already in transit shortens the trip to zero. A decaying speed of light makes the trip faster. Humphreys' white hole and Hartnett's Carmelian cosmology run distant clocks fast while Earth's runs slow. Lisle's anisotropic synchrony convention makes the inbound one-way speed of light infinite by definition, so the trip takes no coordinate time at all.

Every one of them is about the journey. None is about the traveller.

The problem this article is about

Now ask a different question. Forget how long the light took. Look at what the light says about the object that emitted it.

A star's colour and brightness, together with its distance, fix its position on the Hertzsprung–Russell diagram. That position tells you the star's mass and what stage of hydrogen burning it has reached. A white dwarf's colour tells you its surface temperature, and a white dwarf has no energy source, so its temperature is a stopwatch reading. A stellar spectrum shows the strength of the thorium and uranium lines, which is a measure of how much of each isotope is left.

These are not statements about light travel. They are statements about what has already happened inside a particular ball of gas.

Consider what that means for a young-earth reader. Suppose every distant-starlight answer works perfectly. Grant created light, grant c decay, grant infinite inbound light speed. The light from HD 140283 still arrives carrying a spectrum, a colour, a brightness, and a measured parallax. Those quantities describe a star of about three-quarters of the Sun's mass that has finished burning the hydrogen in its core and moved onto the subgiant branch. A star of that mass takes about twelve billion years to do that.

Making the light arrive faster does not make the star burn faster.

The one escape, and who has walked through it

There is exactly one family of answers that does address the object rather than the journey: gravitational time dilation. If clocks in deep space genuinely ran for billions of years while Earth's clock ran for days, then the stars really are old in their own frame and everything measured here is real.

Danny Faulkner named this correctly, and named its only competitor, in a 2007 review in the Creation Research Society Quarterly:

"The accelerated nuclear decay proposed by the RATE project offers the possibility of rapid stellar development early in the universe. The Humphreys white hole cosmology allows for vast periods of time to have elapsed elsewhere in the universe while only a few days passed on earth during the Creation Week. The Humphreys cosmology might allow for much of the stellar development briefly described here."¹³

Look at what is not on his list. No created light in transit. No c decay. No anisotropic synchrony convention. No Carmelian cosmology. He does not offer them, because he is asking about the object's history and those answers do not touch it. That is a working astronomer getting the logic exactly right in 2007, and the rest of this article is essentially an elaboration of his paragraph.

Jason Lisle reached for the same escape in 2008, answering a reader who asked how white dwarfs can exist in a 6,000-year universe:

"In these models, stars can literally age millions of years while only thousands of years elapse on Earth — in which case the secular conjectures regarding stellar aging may be close to the right idea. Such time dilation is the result of Einstein's physics — the well-tested General Theory of Relativity."²²

That was a good answer in 2008. It is not available in 2026, and the reason has nothing to do with mainstream astronomy. Humphreys abandoned the white hole cosmology.²³ Hartnett retracted the Carmelian cosmology in five dated statements between 2015 and 2024; the case file is in Carmelian cosmology. And Lisle's own current model, the anisotropic synchrony convention, contains no time dilation at all — it is a synchrony convention, a relabelling of the time coordinate, which changes when we say a distant event happened without adding proper time to the emitter's worldline. Lisle's own 2024 treatment of stellar ages has dropped the time-dilation move entirely.¹⁵

So the one escape that would work has, at the moment, no model under it. That is not a mainstream complaint; it is the state of the young-earth literature by its own dates.

Where this article goes

Part 2 sets out the four clocks and what each measures. Part 3 does the turnoff arithmetic from measured inputs, on paper. Part 4 is white dwarf cooling, which is the cleanest clock in stellar astronomy and the one the young-earth literature has never engaged. Part 5 is radioactive dating of stars — the same isotopes used on rocks, read in stellar spectra. Part 6 is the cross-checks: what happens when several clocks are pointed at one object. Part 7 is the nearby-object table, which is where the argument actually bites. Part 8 is the chemical history of the Galaxy. Part 9 is galaxies, including a careful account of the JWST controversy. Part 10 goes through the young-earth answers one at a time, at full strength. Part 11 says what creationists get right.

Part 2 — Four clocks

Four independent methods measure stellar ages. They share almost no physics, which is the point.

The main-sequence turnoff reads how much of a star cluster's upper main sequence is missing. A star's lifetime is its fuel divided by its burn rate, and bright stars burn out first. Find the brightest star still burning core hydrogen and you have found the star whose lifetime equals the cluster's age. The inputs are stellar structure physics, nuclear reaction rates, opacities, and a distance.

White dwarf cooling reads the temperature of dead stars. A white dwarf has no fusion. It is a ball of degenerate carbon and oxygen radiating away stored heat on a calculable schedule, and the faintest one in a population tells you how long the population has been cooling. The inputs are a degenerate equation of state, envelope opacities, and crystallisation physics. Nuclear reaction rates barely enter.

Nucleocosmochronology reads radioactive decay in a stellar atmosphere. Thorium-232 has a half-life of 14.05 billion years and uranium-238 has 4.468 billion. Both are made only by rapid neutron capture, in a ratio nuclear theory predicts. Measure what is left and you have a date. The inputs are nuclear half-lives, spectral line strengths, and a production ratio from r-process models.

Asteroseismology reads the star's own oscillations. Stars ring, and the frequencies depend on the sound speed all the way down, which depends on the composition, which changes as hydrogen burns to helium. The pulsation spectrum is a direct probe of the interior.

There is also a fifth thing, which is not a clock but a calibrator: detached eclipsing binaries. Two stars orbiting and eclipsing each other give their masses from Kepler's third law and their radii from the eclipse timing, with no reference to distance and no stellar model at all. Eker and colleagues compiled 509 such main-sequence stars in 2018 and fitted the mass–luminosity relation directly.²⁴ Their fit for the 0.72–1.05 solar-mass range, made to 53 eclipsing-binary components, reproduces the Sun to 1.6 per cent without having been told about the Sun. This is why stellar models are not free to be wrong: the relation they predict is measured in hundreds of systems.

What "the models could be wrong" actually means, quantified

The standard objection to any of this is that the models are assumptions. It is worth being precise about how large the assumption is, because somebody has done the experiment.

Ying and colleagues in 2023 built 20,000 sets of stellar isochrones by Monte Carlo, varying twenty input parameters over their full plausible ranges — metallicity, primordial helium, alpha enhancement, mixing length from 1.0 to 2.5, helium and heavy-element diffusion coefficients, three different surface boundary conditions, low-temperature opacities to ±30 per cent, conductive opacities to ±20 per cent, convective core and envelope overshoot from zero to 0.2 pressure scale heights, and seven nuclear reaction rates with their measured uncertainties.²⁵ The result for the globular cluster M92 was 13.80 ± 0.75 Gyr, a 5.4 per cent error, of which about 80 per cent is the distance.²⁵

Chaboyer and colleagues had done the same experiment in 1996 with over four million stellar models and reported the sensitivity to each input separately: the RR Lyrae absolute magnitude at ±16 per cent, alpha enhancement ±7 per cent, mixing length ±5 per cent, and diffusion, the ¹⁴N(p,γ)¹⁵O rate and the primordial helium abundance at ±3 per cent each.²⁶ Their Monte Carlo's own summary sentence is the one to hand a skeptic:

"In particular, at the lower age limits the rise is steeper than Gaussian, reflecting the fact that essentially all models give an age in excess of 10 Gyr"²⁶

The full plausible range of every uncertain input in stellar interior physics moves the answer by 3 to 16 per cent. Getting from 12.5 billion years to 6,000 years requires a factor of about two million.

Part 3 — The turnoff clock, worked in five lines

The turnoff method sounds like it depends on elaborate models. Underneath, it is arithmetic that a high schooler can check, and every input except one is a laboratory or geometric measurement.

Input 1: the efficiency of hydrogen fusion. Four hydrogen atoms weigh 4 × 1.00782503 u. One helium-4 atom weighs 4.00260325 u. These are atomic masses measured in Penning traps to eleven significant figures. The difference is 0.7119 per cent of the starting mass. Neutrinos carry off about two per cent of that, so 0.698 per cent of the rest mass of the hydrogen consumed comes out as light and heat — 6.27 × 10¹⁸ erg per gram.

Input 2: the hydrogen content. Helioseismology measures the Sun's convection-zone helium abundance directly, from the equation-of-state signature in its oscillation frequencies: Y in the range 0.24–0.25, requiring an initial Y₀ of 0.27–0.28.²⁷ So the initial hydrogen fraction X is about 0.71. Old halo and cluster stars are more hydrogen-rich, X ≈ 0.75.

Input 3: the mass. 0.80 solar masses for a star near a globular cluster's turnoff. This is measured, not modelled. Kaluzny and colleagues weighed three eclipsing binaries in M4 and got component masses of 0.8035 ± 0.0086, 0.6050 ± 0.0044, 0.7842 ± 0.0045, 0.7443 ± 0.0042, 0.7665 ± 0.0053 and 0.7278 ± 0.0048 solar masses.²⁸ Better than one per cent, from Kepler's third law and eclipse geometry.

Input 4: the luminosity. From the measured mass–luminosity relation, or from a parallax and a photometer. A metal-poor 0.80 solar-mass star runs at roughly 0.5 solar luminosities — brighter than its solar-metallicity twin, because fewer metals means lower opacity.

Input 5, the only model input: the core fraction. A star burns only the hydrogen in its core, roughly the inner ten per cent by mass for a star near a solar mass. Call it f = 0.10. This is the one number from a stellar structure calculation.

Now the arithmetic:

fuel      = f × X × M
          = 0.10 × 0.75 × 0.80 × 1.989e33 g
          = 1.19e32 g of hydrogen

energy    = 1.19e32 g × 6.27e18 erg/g
          = 7.48e50 erg

burn rate = 0.5 × 3.828e33 erg/s
          = 1.914e33 erg/s

lifetime  = 7.48e50 / 1.914e33
          = 3.9e17 s
          = 12.4 billion years

Run the same calculation for the Sun as a check — M = 1.000, L = 1.000, X = 0.71 — and it gives 7.3 billion years, against about 10–11 billion from full stellar models. It is low by a third, and the reason is known: the calculation charges the whole life at today's luminosity, and the Sun was roughly 30 per cent fainter when it arrived on the main sequence. A one-line calculation with no stellar model in it lands within 30 per cent of the model answer.

That is the whole argument. A star whose mass has been weighed with Kepler's laws, shining at a brightness measured with a parallax and a photometer, burning fuel at an efficiency measured in a mass spectrometer, lasts about ten billion years. No isochrone, no colour-magnitude diagram, no cosmology, no assumption about the age of anything.

The honest weaknesses of this calculation

State them, because a skeptic will find them.

The core fraction f = 0.10 is model-derived and could be wrong by tens of per cent. It cannot be wrong by six orders of magnitude, because f cannot exceed 1 and cannot be small enough to matter without the star having no core at all.

Constant luminosity is wrong; stars brighten as they burn. That error runs toward longer real lifetimes, not shorter.

The mass–luminosity relation is calibrated on solar-neighbourhood binaries at roughly solar metallicity, and extrapolating it to metal-poor halo stars is the weakest step. This is why the metal-poor version above uses a hand-set luminosity rather than the fitted relation. A rigorous version needs metal-poor eclipsing binaries, and M4's are the relevant sample.

Eker's scatter in the 0.72–1.05 solar-mass domain is 0.129 dex, a factor of 1.35 in luminosity and hence in lifetime.²⁴

Every one of those is a tens-of-per-cent problem. None is a factor-of-a-million problem.

One more thing the same scaling predicts, and it checks out

Lifetime scales as roughly M⁻⁴·⁷ using Eker's measured low-mass exponent. Run it downward and the prediction is that no star below about 0.8 solar masses has ever left the main sequence in the history of the universe. That is checkable, and it checks: no globular cluster shows a turnoff below about 0.8 solar masses, and on this arithmetic none should.

Part 4 — White dwarfs: the cleanest clock

What a white dwarf is, and why it is a stopwatch

A white dwarf is the exposed core of a dead star. Fusion has stopped. Nothing generates new energy. What holds it up is electron degeneracy pressure, which does not depend on temperature, so the star cannot shrink to make new heat the way a normal star does. All it can do is radiate away the thermal energy stored in the motions of its carbon and oxygen nuclei.

Tremblay and colleagues put it in the first sentence of their 2019 abstract:

"White dwarfs are stellar embers depleted of nuclear energy sources that predictably cool over billions of years at the expense of their leaking thermal reservoir of ions"²⁹

Heat capacity in, luminosity out. There is no free parameter that speeds it up by orders of magnitude.

Leon Mestel worked out the cooling law in 1952. The degenerate core conducts heat extremely well, so it is nearly isothermal; a thin non-degenerate envelope sits on top and acts as a blanket. Solving for the flux gives a lifetime scaling as roughly the surface temperature to the power −2.5. Two consequences matter here. Cooling is fast at first and then very slow — a white dwarf drops from 100,000 K to 10,000 K quickly and then takes billions of years to get from 10,000 K to 4,000 K. And because the age depends on a low power of the luminosity, a factor-of-a-million age error would need the luminosity scale wrong by a factor of about 25,000. Luminosity is flux times distance squared, and Gaia measures the distance by parallax.

The arithmetic, from the published tables

The Montreal group publishes its cooling sequences as plain text tables.⁶ A 0.6 solar-mass white dwarf with a thick hydrogen envelope starts at 98,455 K at age zero. Reading the table:

Surface temperature	Cooling age
29,500 K	0.0094 Gyr
9,966 K	0.638 Gyr
5,961 K	2.46 Gyr
4,958 K	6.65 Gyr
3,960 K	9.99 Gyr
2,999 K	12.08 Gyr

Notice the shape. Getting from 98,455 K down to 10,000 K takes 0.64 billion years. Getting from 10,000 K down to 3,000 K takes another 11.4 billion. Almost all the clock is in the cold part, which is why the faint end of the population carries the age information.

Now interpolate the table at 6,000 years. The answer is 97,900 K.

Read that again. A 0.6 solar-mass white dwarf that is 6,000 years old has cooled from 98,455 K to 97,900 K. It has lost six-tenths of one per cent of its surface temperature. It is a blue-white, blazing object, catalogued from across the Galaxy. For a 1.0 solar-mass white dwarf the same calculation gives 132,600 K.

A 6,000-year-old universe predicts that every white dwarf in the sky is hotter than 90,000 K. What is observed is a population running from about 100,000 K down to below 3,000 K, with the bulk between 4,000 and 20,000 K, and with the coolest members outnumbering the hot ones because cooling slows down.

This does not require trusting any modelling refinement. It only requires that a hot object with no energy source gets colder.

The version that uses only conservation of energy

There is a form of this argument with no cooling model in it at all.

Integrate the luminosity along the same table from 9,966 K down to 2,999 K. The total radiated energy is 1.92 × 10⁴⁷ erg, over 11.4 billion years. To radiate that in 6,000 years instead, the star must average 1.0 × 10³⁶ erg/s, which is 265 solar luminosities. An Earth-sized object radiating 265 solar luminosities has a surface temperature of about 210,000 K — hotter than all but a handful of the hottest known white dwarfs, sustained for the entire 6,000 years.

No such population is observed. There is no reservoir in a degenerate carbon-oxygen sphere that can supply it, and nowhere for the energy to have gone that does not light up the sky.

The prediction made in 1968 and seen in 2019

The standard reply to any of this is that the cooling models are assumptions. Here is why that reply does not work.

In 1968 H. M. van Horn predicted that a cooling white dwarf's carbon-oxygen core would undergo a first-order phase transition and crystallise, releasing latent heat and delaying cooling by about a billion years. The observable consequence would be a pile-up of stars in the Hertzsprung–Russell diagram, running across the cooling tracks rather than along them. Tremblay and colleagues describe the status before Gaia:

"The observational implication of this, predicted over fifty years ago, is an isolated crystallisation sequence in the colour versus absolute magnitude Hertzsprung-Russell (H-R) diagram, yet no direct observational evidence existed until now to characterise this event."²⁹

Gaia DR2 gave accurate luminosities for a volume-limited sample of white dwarfs within 100 parsecs — 15,109 of them, at a median parallax precision of 1.5 per cent. The transverse sequence was there, exactly where models put 20 to 80 per cent of the star's mass solidified. They ruled out the alternatives explicitly: post-main-sequence mass loss and binary mergers cannot produce a tight mass–temperature correlation, only about 8 per cent of the pile-up stars are magnetic, and cross-matching with five other surveys showed the stars are underluminous at all wavelengths and so genuinely high-mass.

Then the quantitative test. They built the luminosity function of white dwarfs with Gaia masses between 0.9 and 1.1 solar masses and compared it with population simulations under three treatments of crystallisation:

"The case without latent heat release by crystallisation is clearly ruled out by the observations. When latent heat is included in the modelling, there is, comparatively, a substantial increase of the predicted number of white dwarfs in the range of luminosity of the observed peak. The Gaia luminosity function is best reproduced when ¹⁶O sedimentation is allowed to occur along with the release of latent heat."²⁹

And the summary claim:

"We report direct evidence that a first-order phase transition really occurs in high-density Coulomb plasmas, a theory that cannot be tested in laboratories because of the extreme densities involved"²⁹

A model that predicts a novel feature fifty years in advance, in a regime no laboratory can reach, is not an assumption. Their own honest limit belongs here too: "The model bump is not perfectly modelled, as the observed feature is narrower and of higher magnitude" than predicted.

The case where the clocks disagreed, told in full

The field's best advertisement is not its successes but the one time it failed loudly.

NGC 6791 is an old, metal-rich open cluster. In 2005 Bedin and colleagues reached magnitude 28.5 in it with Hubble and found the white dwarf luminosity function peaked at 27.4. Their abstract:

"If this peak corresponds to the end of the white dwarf cooling sequence the comparison with theoretical isochrones provides a cluster age estimate of ~2.4 Gyr, in sharp contrast with the age of 8-9 Gyr inferred from the main-sequence turn-off. If the end is at fainter magnitudes, the peak at mF606W = 27.4 is even more enigmatic. We discuss possible causes, none of them very convincing."³⁰

A factor of 3.5 disagreement, published openly, with no attempt to hide it. In the same paper they checked M67 as a control, where the two clocks agreed — 3.5 Gyr from cooling against about 4 Gyr from the turnoff — and wrote that this "makes the disagreement in NGC 6791 even more puzzling."

In 2008 they went deeper and found two peaks in the luminosity function rather than one:

"Canonical WD models seem to fail — at least at ~25%-level — in reproducing the age of clusters of this metallicity."³¹

They then quantified the candidate physics one at a time: a revised ¹²C(α,γ)¹⁶O rate worth less than 1 Gyr, ²²Ne diffusion in the liquid phase worth about 1 Gyr, ²²Ne separation during crystallisation worth about 0.4 Gyr. In 2010 García-Berro and colleagues combined ²²Ne sedimentation with carbon-oxygen phase separation and raised the white dwarf age from 6.4 ± 0.2 to 8.0 ± 0.2 Gyr, matching the turnoff.

And then that fix was itself revised. Salaris and colleagues in 2024:

"the very recent model calculations by Salaris et al. (2022) and Bauer (2023) have shown that the cooling delay induced by neon diffusion is smaller than previously calculated, leading Bauer (2023) to question whether the latest generation of WD models would still be able to lead to WD ages consistent with the cluster age determined from the main sequence."³²

A new mechanism — ²²Ne distillation, in which neon-poor crystals become buoyant and float — had to be modelled to restore the match.

What this episode actually shows. For twenty years two clocks disagreed on this one cluster by 25 to 60 per cent, and the resolution required genuinely new physics. That is a real limitation on the precision of white dwarf cosmochronology and it should not be swept aside. But look at the size of the problem. The disagreement was 2 to 4 billion years on an 8-billion-year cluster, and it was fixed by mechanisms worth 0.4 to 3 billion years each. Nothing in this literature, including its worst failure, comes within seven orders of magnitude of 6,000 years.

How big the uncertainties are, from the people who measured them

Pathak, Blouin and Herwig ran cooling sequences for 0.5 to 1.0 solar-mass hydrogen-atmosphere white dwarfs, varying core stratification, helium envelope thickness, hydrogen content and conductive opacities across the range models and asteroseismology allow:

"For a 0.6 M☉ white dwarf, we find an uncertainty of 0.03 Gyr at 10,000 K (corresponding to a 5% relative uncertainty) and 0.8 Gyr at 4000 K (9%). … We also separately consider the potential impact of ²²Ne shell distillation, which plausibly leads to an additional uncertainty of ~1 Gyr for crystallized white dwarfs."⁷

Their breakdown at 4,000 K, as maximum variations across the full parameter range: core oxygen content up to 0.8 Gyr, hydrogen layer thickness up to 1.0 Gyr, helium layer thickness up to 0.6 Gyr, conductive opacities up to 1.3 Gyr, and neon distillation about 1 Gyr on top.

This is the number to quote when someone says the models are uncertain. Yes — by about one billion years out of ten. That is ten per cent. The young-earth claim requires the answer to be wrong by 99.99995 per cent.

Three more weaknesses belong here, plainly. The atmospheric physics of the coolest white dwarfs is genuinely hard: below about 4,000 K, hydrogen molecules form and collision-induced absorption crushes the infrared flux, turning the star blue again as it cools. Campos and colleagues call it straight — "the blue turn retains some mystery."³³ Two groups can differ by 1.5 Gyr on the same cluster. And the statistics at the faint end are thin: Kilic's 40-parsec sample has only 84 objects fainter than bolometric magnitude 15.³⁴

Note also which way the corrections have run. Elms and colleagues record that earlier white dwarf estimates of the Galactic disc age "are thought to be underestimated due to ²²Ne dilution cooling delays … being omitted from earlier crystallisation calculations."³ Fixing the physics has pushed the ages up.

Part 5 — Radioactive dating, read in a stellar spectrum

The method

This is the least known of the four clocks and the most striking. The same isotopes used to date rocks on Earth are detectable in the atmospheres of individual ancient stars, and their abundance ratios date those stars directly.

Thorium-232 has a half-life of 14.05 billion years. Uranium-238 has 4.468 billion. Both are made only by the r-process, rapid neutron capture, which produces them in a ratio nuclear theory predicts. Measure the equivalent widths of the Th II line at 4019.13 Å and the U II line at 3859.57 Å, compare with the predicted production ratio, and the decay law gives an age. Cayrel and colleagues print the working formulae:

Δt = 46.7 [ log(Th/r)₀ − log(Th/r)obs ]
Δt = 14.8 [ log(U/r)₀  − log(U/r)obs  ]
Δt = 21.8 [ log(U/Th)₀ − log(U/Th)obs ]

where r is a stable third-peak r-process element and Δt is in billions of years.¹¹

Read those coefficients, because they are the honest part. A 0.1 dex error in a thorium abundance moves the age by 4.67 billion years. Cayrel say it plainly: "an error of 0.1 dex in a Th abundance propagates as a time equal to the age of the Solar System." The third equation, U/Th, has a coefficient less than half as steep, which is the whole reason the U/Th chronometer is the one that works.

The flagship measurement

In 2001 Cayrel and colleagues detected uranium in a star for the first time, in the very metal-poor halo giant CS 31082-001. From the Nature abstract:

"²³⁸U (half-life 4.5 Gyr) is in principle a more precise age indicator, but even its strongest spectral line, from singly ionized uranium at a wavelength of 385.957 nm, has previously not been detected in stars. Here we report a measurement of this line in the very metal-poor star CS31082-001 … The derived uranium abundance, log(U/H) = −13.7 ± 0.14 ± 0.12 yields an age of 12.5 ± 3 Gyr, though this is still model dependent."¹¹

The star's iron content is about one eight-hundredth of the Sun's; its osmium and iridium are about a ninth of solar. Fourteen Th II lines were detected and eleven used, ten of them first detections. Their own statement of the result: "Any age between 11.1 and 13.9 Gyr is compatible with the various determinations associated with their error bars."

Since then the list has grown: BD +17° 3248 at 13.8 ± 4 Gyr from six chronometer pairs;³⁵ HE 1523−0901 at 13.2 Gyr, the first star where several different chronometers could be used at once;³⁶ and 2MASS J09544277+5246414 at 13.0 ± 4.7 Gyr.³⁷

What the field admits about it, stated first

This chronometer has large errors and one outright broken variant, and a reader deserves to hear that from us rather than from a critic.

The production ratio dominates the error budget, and it is theory. Frebel's Table 1 for the single star HE 1523−0901 gives the age from one spectrum under fifteen different chronometer-and-production-ratio combinations. The answers run from 9.5 to 17.8 billion years.³⁶ That is an 8.3-billion-year spread from theory alone, on fixed data. The U-based rows are the tight ones, 12.2 to 13.9 Gyr, and Frebel says why: "the U/Th PR should be less sensitive to theoretical r-process model uncertainties."

Oscillator strengths move the answer by billions of years. Hill and colleagues re-measured the atomic data for the uranium line and eight thorium lines in 2002. The log(U/Th) ratio for CS 31082-001 moved from −0.74 ± 0.15 to −0.94 ± 0.11, and with the same production ratio "this leads to an age of almost 17 Gyr, 4.3 Gyr greater than that originally published."³⁸ So the star's published age went 12.5 ± 3 in 2001, to about 17 in 2002 with new atomic data and the old production ratio, to 14.0 ± 2.4 with an updated production ratio, to 11.1 ± 4.0 in a 2023 re-analysis.³⁹ The star did not change. The atomic data and the nuclear theory did.

And one chronometer is broken outright. About 30 per cent of r-process-enhanced stars have thorium-to-europium ratios two to three times higher than the rest — the "actinide boost." Because the Th/Eu clock reads extra thorium as less decay, an actinide-boost star reads as younger than zero. Holmbeck and colleagues print all three answers for one star side by side: the Th/U age is 13.0 ± 4.7 Gyr, and "The Th/Eu and U/Eu ages are −9.5 Gyr and 5.8 Gyr, respectively."³⁷ Goriely and Arnould put the same absurdity the other way round: under a universal production ratio, one star "would thus be about 36 Gy old."⁴⁰ Hill and colleagues, who discovered the boost, say what it means: any chronometer based on an actinide against a second-peak element is "doomed to failure."³⁸

Shah and colleagues summarise the wreckage in one sentence: "the application of the Th/Eu chronometer to RPE stars has led to a large range in ages from 21 Gyr to -9 Gyr."³⁹

How the field answers it, and how good the answer is. The U/Th ratio is insensitive to the boost, because a boost lifts thorium and uranium together — both sit in the fission-cycling region. Holmbeck and colleagues showed in 2019 that the boost can be produced by small variations in neutron richness within a single class of r-process event, which removes the need for a separate progenitor class.⁴¹ That is a real physical argument, not a rescue. But the residual cost should be stated: the boost factor needed was 3.1 on the initial thorium and uranium production ratios for one star and 2.5 for another. If the actinide production ratio varies by a factor of three between stars, the U/Th production ratio is being assumed constant while both its numerator and denominator are known to vary by that much. The proportionality argument is good. It is not measured.

What the weaknesses do not do

Every problem above is a problem of several billion years, and one of them is a problem of sign within a single broken chronometer. None is a problem of six orders of magnitude. The published disagreements run from −9 to +22 billion years, and the negative end is a diagnosed artefact of one method that the field identified, named and explained.

There is no combination of oscillator-strength error, non-LTE correction, line blending, model-atmosphere choice and production-ratio uncertainty in any of these papers that moves a star from 13 billion years to 10,000.

Why GW170817 matters here

For sixty years nobody knew where the r-process happened, which meant the production ratio was pure theory. Watson and colleagues state the gap as it stood in 2019:

"Half of all the elements in the universe heavier than iron were created by rapid neutron capture. The theory for this astrophysical 'r-process' was worked out six decades ago and requires an enormous neutron flux to make the bulk of these elements. Where this happens is still debated. A key piece of missing evidence is the identification of freshly-synthesised r-process elements in an astrophysical site."⁴²

On 17 August 2017 two neutron stars merged 40 megaparsecs away and the whole thing was watched. Pian and colleagues got spectra of the kilonova;⁴³ Kasen and colleagues modelled two ejecta components, light and heavy;⁴⁴ Watson and colleagues identified strontium.⁴² Kasen's closing paragraph names the connection to old stars directly:

"In the past, the uncertain origin of the heaviest elements was studied indirectly, by analysing fossil traces of these species in the surfaces of old stars. With AT 2017gfo we have now directly glimpsed and spectroscopically dissected a sample of pure r-process matter, big enough to enrich a million such stars."⁴⁴

The boundary has to be stated and the article will not overstep it. GW170817 identified strontium, probably yttrium, probably tellurium. It did not identify thorium or uranium. The heavy component is inferred from opacity — lanthanide-rich ejecta make a red, slow-fading infrared light curve — not from an actinide line. Nobody has measured an actinide abundance in a kilonova.

So the correct claim is two-step. The r-process happens, it happens in neutron-star mergers, and it has been watched happening once. And the specific number the chronometer needs, the initial U/Th ratio, is still model-dependent. What changed in 2017 is that the remaining freedom is the freedom of nuclear-mass and fission-yield models applied to a known environment, rather than the freedom of an unknown environment.

The same isotopes, counted a different way

There is a check on all of this that uses no starlight at all.

Plutonium-244 has an 81-million-year half-life and is made only by the r-process. Any that existed when the Earth formed is long gone, so any found today arrived recently from space. Wallner and colleagues counted it in deep-sea sediment by accelerator mass spectrometry:

"Here we report the detection of live interstellar ²⁴⁴Pu, archived in Earth's deep-sea floor during the last 25 My, at abundances lower than expected from continuous production in the Galaxy by about 2 orders of magnitude. This large discrepancy may signal a rarity of actinide r-process nucleosynthesis sites, compatible with neutron-star mergers or with a small subset of actinide-producing supernovae."⁴⁵

Three things follow. The same isotopes are being measured two completely different ways — through a spectrograph four kiloparsecs away, and as atom counts in Pacific sediment — and they agree on the same picture. The rarity result is what makes "time since the last event" a meaningful question at all. And it is a constraint on accelerated decay, because a count of surviving atoms is not a ratio: any accelerated-decay episode would have destroyed the ²⁴⁴Pu inventory outright rather than rescaling it.

Part 6 — What happens when you point several clocks at one object

A single method that gives billions of years can be dismissed as a model artefact. Several methods resting on different physics, agreeing on the same object, cannot.

NGC 6397

Hansen and colleagues fitted the white dwarf cooling sequence of this globular cluster and then compared it with the turnoff age of the same cluster, using the same distance and extinction:

"In each case, good fits are obtained for a wide variety of ages, 12 ± 2 Gyr, where the accuracy is limited, as in prior studies, by the uncertainties in the distance and extinction. When the distance and extinction used in the MSTO method are restricted to the range that fits the white dwarf model … we find a best fit age of 11.6 Gyr, with a 95% confidence lower limit of 10.6 Gyr. Hence the two methods are in complete agreement, as these were the models used to determine the progenitor ages in Figure 13. As a result, this combination represents a successful end-to-end (main sequence to white dwarf) comparison of a set of stellar models with the observed stellar population."⁸

Their final white dwarf constraint, after marginalising over distance, extinction, white dwarf mass, binary fraction, main-sequence model choice, internal composition and hydrogen layer mass, is 11.47 ± 0.47 Gyr. A cruder route in the same paper reaches the same place and a lay reader can follow it: the white dwarf mass at the truncation is about 0.6 solar masses, giving a cooling age of about 11.0 ± 0.5 Gyr; add the progenitor's half-billion-year main-sequence lifetime and the cluster is about 11.5 ± 0.5 Gyr old.

The independently derived distance is a bonus check. Hansen's cooling fit implies a distance modulus of 12.02 ± 0.06, which is 2.53 kpc. Baumgardt and Vasiliev's Gaia-era catalogue gives 2.488 ± 0.019 kpc from five independent routes.⁴⁶ The cooling fit is not being propped up by a bad distance.

The honest disagreement. VandenBerg and colleagues put NGC 6397's turnoff age at 13.00 ± 0.25 Gyr, and that is a roughly three-sigma disagreement with Hansen's 11.47 ± 0.47.⁴⁷ It is a live dispute about which age above 11 billion years is right. It is not a dispute about whether the cluster is older than 10 billion years.

47 Tucanae

Three numbers exist for this cluster, from three methods: 11.75 ± 0.25 Gyr from the turnoff,⁴⁷ 12.0 ± 0.5 Gyr from two eclipsing binaries plus the colour-magnitude diagram,⁹ and 9.9 ± 0.7 Gyr from white dwarf cooling.⁴⁸

Say this plainly. The spread is about two billion years and it is unresolved. It is also a spread of two billion years around eleven billion, not a spread that includes 6,000 years.

The eclipsing-binary result deserves its own paragraph, because it is the least model-dependent number in the whole field. Thompson and colleagues measured the two components of one binary to 0.6 per cent in mass and 0.3 per cent in radius using nothing but Kepler's third law and eclipse geometry:

"We obtain Mp = 0.862 ± 0.005M☉, Rp = 1.183 ± 0.003R☉, Lp = 1.65 ± 0.05L☉ for the primary and Ms = 0.827 ± 0.005M☉, Rs = 1.004 ± 0.004R☉, Ls = 1.14 ± 0.04L☉ for the secondary. … By comparing the M − R diagram of the two binaries and the color-magnitude diagram of 47 Tuc to Dartmouth model isochrones we estimate the age of the cluster to be 12.0 ± 0.5 Gyr"⁹

And the distance that falls out of the same analysis, 4.55 ± 0.03 kpc, agrees within one sigma with Baumgardt and Vasiliev's independent 4.521 ± 0.031 kpc.

M4

Three clocks again, on the closest globular cluster to the Sun. The white dwarf cooling age is 12.1 Gyr with a 95 per cent lower limit of 10.3.⁴⁹ The turnoff age is 11.50 ± 0.38 Gyr.⁴⁷ And Kaluzny's three eclipsing binaries, using measured masses in the mass–radius plane with no distance at all, give "the most probable age of M4 to be between 11.2 and 11.3 Gyr."²⁸

Then a fourth, of a different kind. Miglio and colleagues detected solar-like oscillations in eight M4 giants with K2 and measured their masses seismically. The isochrone fit to M4's colour-magnitude diagram gives a giant-branch mass of 0.84 solar masses; extrapolating the measured eclipsing-binary masses up the giant branch gives 0.85; and the asteroseismic mass, under all four combinations of constraints, is 0.84.¹⁰

Three routes to one number: stellar-model isochrones, Newtonian orbital mechanics, and the frequencies of sound waves inside the stars. All three say the giants in this cluster weigh about 0.84 solar masses — which is what a 13-billion-year-old population should have left on its giant branch, because anything heavier finished burning long ago.

A caveat on M4's cooling age. Hansen adopted a distance modulus corresponding to 1.72 kpc; the modern Gaia-era value is 1.851 kpc, inside his adopted range but on the far side of it. A larger distance makes the white dwarfs intrinsically brighter and therefore younger, so the modern distance would nudge the cooling age down from 12.1 toward the turnoff value. The conservative statement is that M4 is 11 to 12 billion years old by two independent methods.

The cleanest validation in stellar astronomy is this one.

Silva Aguirre and colleagues circulated a set of solar oscillation frequencies, degraded in quality to match Kepler data, among seven independent modelling teams. "The true identity of this star was concealed from the modellers."⁵⁰ What came back: masses from 0.927 to 1.037 solar masses against a true value of 1.000; radii from 0.973 to 1.012 against a true 1.000; and ages from 4.33 to 5.30 billion years, against the meteoritic value of 4.57.

The Sun's age is known independently by lead-lead dating of calcium-aluminium-rich inclusions in meteorites — radioactive decay in solid rock, with no stellar physics of any kind in it. The ringing of the Sun's outer layers, analysed blind, reproduces that number to within a few per cent.

The residual scatter traces to nameable physics. One pipeline's age came out high because its grid lacked microscopic diffusion; adding it brought the answer to 4.83 Gyr. That is the right kind of failure — identifiable, not arbitrary.

And be careful how much weight this bears. Asteroseismic ages of individual solar-type stars are good to roughly 10 to 20 per cent, not 1 per cent. Tayar and colleagues put a systematic floor of about 5 per cent in mass and 20 per cent in age from model-grid differences alone.⁵¹ Twenty per cent of 12 billion years is 2.4 billion years. Still six orders of magnitude from any young-earth timescale.

Two more honest caveats. A seismic mass becomes an age only through a stellar model, so asteroseismology is not independent of stellar models — what it is independent of is the fitting. Instead of inferring an interior from two surface numbers, you measure the interior directly and use the model only to run a clock forward. And the raw scaling relations drift at low metallicity: Epstein and colleagues found seismic masses for metal-poor halo giants biased high by 0.17 ± 0.05 solar masses, which biases ages young.⁵² Miglio's M4 work is the fix, and with the model-suggested correction applied the metal-poor cluster masses come right.

Part 7 — Where the argument actually bites: nearby objects

Everything so far would still be arguable if all these objects were billions of light years away. A determined reader could say the distances are wrong, or that light-travel-time cosmology rescues it.

They are not all far away, and this is the section that matters.

The distance column is not in dispute

Gaia measures parallax: the shift in a star's apparent position as the Earth swings two astronomical units across its orbit. It assumes no stellar model, no distance ladder, and no cosmology. It is geometry.

Jason Lisle, a young-earth astronomer with a PhD in the field, explicitly accepts parallax distances out to about 30,000 light years. So the distances below are not a point of disagreement between the two sides.

The table

Object	Distance (ly)	Distance method	Age	Age method
The Sun	8.32 light-minutes	AU by definition, checked by radar	4.60 ± 0.04 Gyr	helioseismology
WD J2147−4035	91.1 ± 1.2	Gaia parallax 35.79 ± 0.49 mas	cooling 10.2; total 10.7 ± 0.3 Gyr	white dwarf cooling
WD J2317+1830	123.6 ± 1.4	Gaia parallax 26.40 ± 0.31 mas	cooling 9.5; total 9.7 ± 0.2 Gyr	white dwarf cooling
WD J1922+0233	128.6 ± 1.3	Gaia parallax 25.36 ± 0.27 mas	cooling 9.0 Gyr	white dwarf cooling
local 40-pc white dwarfs	all within 130	Gaia parallaxes	halo 12.5 (+1.4/−3.4); thick disk 8.7 ± 0.1; thin disk 6.8–7.0 Gyr	luminosity function
HD 140283	200.5 ± 0.3	Gaia DR3 parallax 16.26724 ± 0.02603 mas	published range 11.4 to 15 Gyr	isochrone; asteroseismology
HD 84937	241.6 ± 0.8	Gaia parallax	12.08 Gyr (parallax error only)	isochrone
HD 132475	305.7 ± 0.7	Gaia parallax	12.56 Gyr (parallax error only)	isochrone
HD 122563	1,052 ± 11	Gaia parallax	Th/Eu chronometry	radioactive decay
M67 (NGC 2682)	~2,900	Gaia mean cluster parallax	~4.2 ± 0.7 Gyr	isochrone + gyrochronology
M4 (NGC 6121)	6,037 ± 49	multi-method, ~1 %	turnoff 11.50 ± 0.38; WD cooling 12.1 Gyr	two clocks
NGC 6397	8,115 ± 62	multi-method, ~1 %	turnoff 13.00 ± 0.25; WD cooling 11.47 ± 0.47	two clocks

Everything down to and including M67 is inside 6,000 light years. For those objects a 6,000-year-old universe has enough time for the light to arrive with room to spare, and every distant-starlight solution is doing no work whatsoever.

The three objects to think about

The Sun, eight light-minutes away. Its helioseismic age is 4.60 ± 0.04 Gyr, from the frequencies of its own acoustic oscillations.⁴ No meteorite, no decay constant, no assumption about initial daughter isotopes. Accelerated radioactive decay does not touch it, because no radioactive clock is involved — the clock is the sound speed in the core. The honest caveat is that this is the age of the best-fitting standard solar model and inherits that model's input physics; Houdek and Gough say so themselves.

WD J2147−4035, 91 light years away. Surface temperature 3,048 ± 35 K, mass 0.69 ± 0.02 solar masses, cooling age 10.2 Gyr, and with the progenitor's main-sequence lifetime added, a total age of 10.7 ± 0.3 Gyr.³ Elms and colleagues: "WD J2147-4035 has a cooling age of over 10 Gyr which is the largest known for a DZ white dwarf." Its light takes 91 years to reach us.

HD 140283, 200 light years away. The Gaia DR3 parallax is 16.26724 ± 0.02603 milliarcseconds, a precision of 0.16 per cent, with a RUWE of 1.06 confirming a clean single-star solution.¹ That is 200.5 ± 0.3 light years. The photons arriving tonight left around the year 1826.

Its age has been measured eighteen times since 2000, and Lundkvist's compilation prints the whole list.² The values run from 11.38 to about 14.7 billion years. Bond and colleagues got 14.46 ± 0.31 Gyr (parallax error only; ±0.8 total).⁵³ VandenBerg and colleagues got 14.27 ± 0.38.⁵⁴ Creevey and colleagues, using a directly measured interferometric angular diameter, got 13.7 ± 0.7 or 12.2 ± 0.6 depending on the assumed extinction.⁵⁵ Tang and Joyce got the lowest modern value, 12.0 ± 0.5.⁵⁶ And in 2025 Lundkvist and colleagues published the first asteroseismic age, from TESS 20-second-cadence photometry showing solar-like oscillations including mixed modes: 14.2 ± 0.4 Gyr.²

Nobody has ever published an age for this star below about 11.4 billion years. That is the number to use — not the highest one.

Lundkvist's own summary of the spread is the sentence the article wants:

"These studies leave no doubt that the star is old, but the modelling choices have resulted in a spread of age estimates, sometimes placing HD 140283 slightly older than the Universe itself."²

None of the disagreement is about distance. The parallax is nailed to 0.16 per cent. The whole spread lives inside stellar interior physics: a 0.1 magnitude change in assumed extinction moves the age by 1.5 billion years; the convective mixing length is contested; the oxygen abundance and opacity mixture matter. The uncertainty is 11 versus 15 billion years. It is not 11 billion versus 6 thousand.

The clean case at cluster scale

M67 is an open cluster 2,900 light years away — comfortably inside the line. It is not one star but a whole coeval population: hundreds of stars with a turnoff, a giant branch, and a white dwarf sequence. Its age is about 4 billion years by isochrone fitting, and Barnes and colleagues independently got a rotational age of 4.2 Gyr with a standard deviation of 0.7 from K2 rotation periods.⁵⁷

That is roughly 700,000 times the young-earth age of the universe, from an object whose light-travel time is under half that age.

The caveats: gyrochronology is calibrated against the Sun and younger clusters, so it is not fully independent of the solar age; and the load-bearing isochrone citations for M67 come through Barnes's summary rather than from the primary studies read directly.

The globular clusters, stated exactly

No globular cluster lies inside 6,000 light years. M4, the closest of the 183 in Baumgardt and Vasiliev's catalogue, is at 1.851 ± 0.015 kpc, which is 6,037 ± 49 light years.⁴⁶

That is not a weakness of the argument. It is a striking coincidence worth stating exactly. On Ussher's chronology creation was in 4004 BC, which is 6,029 years before AD 2026. M4's light-travel time and the entire young-earth age of the universe are the same number to within one part in a thousand.

So even granting every light-travel fix ever proposed, the M4 we observe tonight is the M4 of about 6,040 years after creation — essentially now, on a young-earth timeline. And that M4 contains a fully developed main sequence with a turnoff at about 0.8 solar masses, a horizontal branch, a red giant branch, and a white dwarf cooling sequence. In a 6,000-year-old universe there has been no time for any star to leave the main sequence at all.

A distant-starlight solution buys a factor of 1.006 at M4. The age gap is a factor of about two million.

The young-earth boundary, drawn by a young-earth astronomer

In 2012 Ron Samec and Evan Figg published a paper in the Creation Research Society Quarterly arguing that binary-orbit decay gives young apparent ages. Their conclusion contains this:

"We observe that all of these objects are local (all within ~1000 parsecs or ~3000 ly) and not cosmological, deep-space objects. This may validate the notion that all nearby objects partook in the same dilation effects as the earth during Creation Week and that they all have ages measured in earth time on the order of 10,000 years, nearly the same as we find for solar system objects such as that of Jupiter and interplanetary dust."⁵

Take that criterion seriously and apply it to the table above. HD 140283 is at 200 light years, fifteen times inside Samec's boundary. WD J2147−4035 is at 91 light years, thirty times inside it. The local white dwarf luminosity function is built from objects within 130 light years. By Samec's own criterion, no time-dilation discount is available for any of them: their clocks ran with Earth's.

And three years later, in 2015, Samec rejected the only remaining option:

"WDs are essentially 'fossil stars'. … I believe that God did not create 'dead' stars."

"However, the idea that God created WDs is analogous to the idea that He created fossils in the ground, so they never actually lived and died. Thus they are a false history"¹⁴

This is not a gotcha and should not be read as one. It is a genuine tension between two honest papers by the same author, three years apart, and it is the sharpest available way to show what the young-earth position on nearby old objects actually costs. Nobody in the surveyed literature has noticed the collision.

One caveat on the record: Samec published a Part II in CRSQ 53(1) in 2016 whose subtitle promises errata to the 2012 paper. That volume was not obtained for this article. The 2012 statement is quoted as printed, and anyone leaning on the numerical results of Part I should read Part II first.

Part 8 — The heavy elements record a sequence

The Big Bang makes hydrogen, helium, and a trace of lithium. Nothing heavier survives; that case is made in Big Bang nucleosynthesis and is not re-argued here. Everything from carbon to uranium was made in stars.

The question for this article is not whether stars can make elements. It is whether the distribution of heavy elements across the sky records a long history.

Different makers, different clocks

Each source of heavy elements runs on a timescale set by stellar evolution, and those timescales are not free parameters.

Core-collapse supernovae need a star above about eight solar masses, whose whole life is a few million years. They arrive fast and make mostly alpha elements — oxygen, magnesium, silicon, sulphur, calcium.

Type Ia supernovae need a white dwarf, which means a star of a few solar masses has to finish its entire main-sequence life first, and then be pushed over the edge by a companion. Maoz, Mannucci and Nelemans state the physical floor:

"The production rate is zero for at least 30 − 40 Myr, until main sequence stars with masses below ∼8 M☉ evolve into the first WDs to emerge from the stellar population."⁵⁸

And the observed delay-time distribution goes as roughly 1/t out past ten billion years — meaning Type Ia supernovae are still going off now, from stars that formed ten billion years ago. This is fitted to measured supernova rates in galaxy clusters, in field ellipticals, against host colour and Hubble type, and against the volumetric rate versus redshift.

AGB stars make the main s-process, and nothing arrives until a low-mass star has finished the main sequence, the red giant branch and core helium burning and reached the thermally-pulsing phase — hundreds of millions to billions of years.

Kobayashi, Karakas and Lugaro state the consequence in one line:

"the [α/Fe]–[Fe/H] relation in the Milky Way Galaxy is explained by the delayed enrichment of Fe from SNe Ia, which have a longer timescale than core-collapse supernovae. Therefore, the [α/Fe] ratios can be used to constrain star formation timescales in other galaxies."⁵⁹

The knee, and why it is the centrepiece

Here is the argument in five sentences. Core-collapse supernovae arrive fast and make alpha elements. Type Ia supernovae arrive late and make iron. So a young population sits on a high, flat alpha-to-iron plateau. When Type Ias switch on, iron floods in without a matching alpha supply and the ratio starts falling. Plot alpha-to-iron against iron and you get a plateau followed by a downturn — the knee — and the iron abundance at the knee measures how much iron the galaxy had built by the time its first Type Ias exploded.

Beatrice Tinsley predicted this in 1979, before the data existed:

"The excess [O/Fe] ratios found in halo stars can be explained if a significant fraction of iron comes from SN I, and oxygen from more massive stars. Since SN I precursors are assumed here to have lifetimes of < 10⁸ years, the excess only appears in stars formed within about 10⁸ years of the start of star formation."⁶⁰

Tolstoy, Hill and Tosi state the mechanism and the crucial property:

"SNIa have a longer time scale than SNII and as soon as they start to contribute they dominate the iron enrichment and [α/Fe] inevitably decreases. After that, no SFH can ever again result in enhanced [α/Fe], unless coupled with galactic winds removing only the SNIa ejecta and not that of SNII. This is seen as a 'knee' in a plot of [Fe/H] vs. [α/Fe]. … The knee position indicates the metal-enrichment achieved by a system at the time SNIa start to contribute to the chemical evolution. This is between 10⁸ and 10⁹ yrs after the first star formation episode."¹⁶

"After that, no SFH can ever again result in enhanced [α/Fe]" is an irreversibility statement. The knee is a one-way door in a galaxy's history, and its position records when the door was passed.

Now look at where the knee sits in different galaxies:

System	[Fe/H] at the knee
Galactic bulge	−0.37 ± 0.09
Milky Way halo (model-derived)	≈ −1.0
Sagittarius dwarf	−1.5 to −1.2
Fornax dwarf	< −1.5
Sculptor dwarf	≈ −1.8
Large Magellanic Cloud	≲ −2.2
Small Magellanic Cloud	≲ −2.2

Nearly two orders of magnitude, and the positions track each galaxy's independently measured star-formation history. Tolstoy and colleagues: Sculptor "stopped forming stars 10 Gyr ago, and the knee occurs at [Fe/H] ≈ −1.8," while Sagittarius "has steadily formed stars over a period of 8−10 Gyrs, and only stopped forming stars about 2−3 Gyr ago" and its knee sits nearly a full dex higher.

There is a second, non-obvious result in the same diagram. The pre-knee plateau looks the same everywhere — in dwarf galaxies and in the Milky Way halo alike — because it is set by massive-star yields alone. Only the knee position varies, because that is set by how fast each system was building iron when its Type Ias arrived. Two distinct behaviours in one plot, both doing what the delay-time argument says they should.

Two honesty items that must not be buried.

Nidever and colleagues measured the knees of the Magellanic Clouds with 3,600 red giants and found them at ≲ −2.2, more metal-poor than dwarfs a hundred to a thousand times less massive.¹⁷ The correlation between knee position and luminosity is real — Pearson r = −0.705 across eleven Milky Way dwarfs — but the two most massive satellites break it badly. Their reading is that the knee tracks star-formation efficiency, not mass, and mass is only a proxy that fails for these two.

And Kobayashi's group does not attribute the knee purely to the Type Ia delay time. In their models the knee position "is determined not by the lifetime but by the metallicity dependence of SN Ia progenitors."⁵⁹ That is a real internal disagreement in the field, and both readings still require a long history: a metallicity-dependent progenitor efficiency is still a statement about a sequence of stellar generations.

The metallicity floor

If every heavy element came from a star, the first generation formed from gas with no heavy elements at all. Two observational facts follow, and they pull in opposite directions.

No zero-metallicity star has ever been found. Not one, in decades of searching.

Stars exist with iron abundances a ten-millionth of the Sun's. The record holder is SMSS J031300.36−670839.3 at [Fe/H] below −7.1 after three-dimensional and non-LTE corrections.⁶¹ Six others sit between −4.7 and −5.7.⁶²

Why the floor is at −7 rather than at minus infinity is the physical claim: metal-free gas has no efficient cooling agents, so it cannot fragment into small pieces, so the first stars were massive and died in a few million years. Only a later generation, formed from gas already seeded with a little metal, could contain stars small enough to still be burning.

Be honest about the shape of that argument. "The first stars are missing because they were all short-lived" is, on its own, unfalsifiable: the theory predicts you cannot see the thing it predicts. It becomes a real scientific claim only because it makes other checkable predictions — that surviving stars should bottom out near a critical metallicity, that their abundance patterns should look like the yield of one or a few supernovae, and that Population III light should be findable at high redshift.

The second of those is checkable now and it works. The elemental pattern in the most iron-poor stars can be fitted by the calculated yield of one Population III supernova of a specified mass and explosion energy. That is an unusual kind of test, because a supernova yield model has few free parameters and predicts a dozen element ratios at once.

The third is the open one. As of 2026 the honest status is candidates, not confirmations. Venditti and colleagues, reviewing the whole search: the combination of JWST spectroscopy, time-domain searches, lensing surveys and stellar archaeology "is yielding a growing number of observational candidates and narrowing the allowed parameter space for the first stars."⁶³ Visbal and colleagues state it flatly: "Up to this point, there have been no unambiguous direct detections of Population III (Pop III) stars."⁶⁴ The star formation article says the same, and this one does not contradict it.

Metallicity rises with cosmic time, and this needs almost no theory

Light from a distant object left it long ago. If the universe's gas was more metal-poor in the past, distant gas should be more metal-poor than nearby gas. That is a measurement, made by looking at absorption lines that intervening gas clouds imprint on quasar spectra.

Rafelski and colleagues, combining 47 new damped Lyman-alpha systems with 195 from earlier surveys:

"We find the metallicity of DLAs decreases with increasing redshift, improving the significance of the trend and extending it to higher redshifts, with a linear fit of −0.22 ± 0.03 dex per unit redshift from z = 0.09 − 5.06."¹⁸

That is a seven-sigma trend across five units of redshift, using 242 systems. Gas at redshift 5 is roughly ten times poorer in metals than gas nearby.

Péroux and Howk's later compilation gives the binned numbers: hydrogen-weighted mean metallicity of −0.16 at mean redshift 0.55, falling to −1.03 at mean redshift 4.54.¹⁹ Their summary: "The H i-weighted mean metallicity of neutral gas in the Universe increases slowly from high to low redshift, rising by an order of magnitude from z ≈ 5 to 0." The dust-to-gas ratio moves in parallel by the same amount, which is an independent tracer since dust is made of heavy elements.

Two honesty items. The scatter among individual systems is about 0.5 dex and does not shrink at high redshift, so this is a trend in the mean, not a track any single cloud follows. And one bin, at mean redshift 2.70, sits above the smooth trend — with the largest error bars in the table. The trend is a factor-of-ten rise with a bumpy middle, not a straight line.

One overstatement to avoid. At fixed stellar mass, JWST finds galaxies at redshift 4 to 10 only mildly more metal-poor than galaxies at redshift 2 to 3 — about 0.1 to 0.2 dex. The mass–metallicity relation itself barely moves.⁶⁵ Saying "JWST shows the early universe was metal-poor" without that qualification is an overstatement. The sharp measurement is the total metal content of the universe's gas, which is what the quasar-absorption compilations give.

What a young-earth model has to supply

Stated fairly. On a 6,000-year timescale there are two options, and both have been proposed in print by young-earth authors.

Option A — direct creation. Every element heavier than lithium was made by fiat, in place, with an abundance pattern that mimics what thirteen billion years of stellar processing would produce. Vernon Cupps states it as clearly as anyone: "any paradigm based on the veracity of this account does not allow for any elements on Earth to have come from stars."⁶⁶ Jonathan Henry states it too, and explicitly rejects the alternative: "A naturalistic process impossible over eons is less likely over days, and to say that God accomplished the naturalistic process quickly is to verge on a kind of 'theistic naturalism'."⁶⁷

Here is the bill Option A has to pay, all at the moment of creation: a one-way separation between alpha and iron-peak elements; a turning point at a different iron abundance in every galaxy, spanning two orders of magnitude; turning points that correlate with each galaxy's independently measured star-formation history; an identical pre-knee plateau everywhere while the knees differ; a metallicity floor at [Fe/H] ≈ −7 rather than at zero, with no zero-metallicity star anywhere; abundance patterns in individual extremely metal-poor stars each fitted by the calculated yield of one supernova; a radial iron gradient across the Milky Way disk; a tight age–metallicity sequence in the old disk; and a universe whose neutral gas is ten times more metal-poor at redshift 5 than today.

Every one of those is a pattern indexed by time or by a system's history. A creation event produces one composition, not nine correlated patterns. To produce them it would have to have been designed to look exactly like a long history — which is a theological claim, not a scientific one, and this article says so in those words rather than pretending data refutes it.

Option B — accelerated nucleosynthesis. The same processes ran, enormously faster. This needs a physical mechanism, since reaction rates are set by measured cross-sections and temperatures. It needs an answer for the delay-time structure, which is the harder problem: the knee does not depend on nucleosynthesis being slow, it depends on core-collapse and Type Ia supernovae happening at different times relative to each other. Accelerating everything by the same factor preserves the ratio and still requires the full sequence of stellar generations. Accelerating them by different factors is a free parameter per process, tuned to the answer. It needs a heat budget. And as of 2026 there is no published calculation.

Faulkner, in a 2014 review in Answers Research Journal, names both options and grades his own side honestly:

"one could posit that God created the elemental abundances pretty much as they exist today. However, many will find such an answer unsatisfactory, particularly if there is no clear standard of design of the elemental composition. … There have been a few initial attempts to account for chemical abundances in the creation literature in such a physical way. In the initial concept of his white hole cosmology, Humphreys (1994) proposed rapid primordial nucleosynthesis of elements. Boudreaux (2012) and Davies (2013) have proposed similar things. However, all of these suggestions are at best preliminary. I encourage further development of ideas of this type."⁶⁸

Twelve years on, nothing found in a 2026 search changes that.

Two creationist arguments in this area, answered

Cupps on iron-iron fusion. He calculates that fusing two iron-56 nuclei needs 91 MeV, corresponding to about 10¹² K, while the hottest stars measured are around 4 × 10⁴ K, and concludes "it is not possible for the heavy elements to form in known stable stars and nebula."⁶⁶

The arithmetic is fine and it is beside the point. Nobody claims heavy elements are made by fusing iron with iron. The s- and r-processes are neutron capture. A neutron is uncharged. There is no Coulomb barrier. That is the entire reason the field settled on neutron capture in 1957. Cupps does go on to discuss neutron capture, so he knows this; but a reader who stops at his headline calculation gets a false impression. His follow-up question — where do the neutrons come from — has a specific answer: ¹³C(α,n)¹⁶O in the helium intershell of AGB stars for the main s-process, ²²Ne(α,n)²⁵Mg in massive stars for the weak component, and neutron-star matter itself for the r-process. Both alpha-capture reactions are measured in laboratories.

Cupps's strongest point, and it deserves conceding as stated: "there is no way observationally to determine whether any heavy nuclei observed are primordial (i.e., original) or manufactured later."

For a single object, taken alone, that is true. Look at one star's gold and nothing about the atom tells you when it was made. But no argument above rests on a single object. The evidence is the pattern across many objects, ordered by galaxy, by radius, by stellar age and by redshift. The knee is not a measurement of one star. A radial abundance gradient is not a measurement of one star. A tenfold rise in the metallicity of neutral gas from redshift 5 to today is not a measurement of one star. The objection defeats a claim nobody makes.

Faulkner's "no predictive power" charge, answered with a date. Faulkner writes that nucleosynthesis theory's power "is entirely explanatory" and that "A theory that explains anything and everything explains nothing."⁶⁸

The narrower version of his worry is fair, and this article has already conceded cases where the field patched a model to fit. But the blanket charge is false, and the counterexample has a date on it. Matteucci and Greggio in 1986 predicted silicon and magnesium overabundances in halo stars before those measurements existed — at that time only oxygen data existed — and François confirmed both in the same year.⁶⁹ A second case: the time-delay model predicted that the Galactic bulge, having formed fast, should knee at higher metallicity than the solar neighbourhood, and Gaia-ESO and APOGEE later measured exactly that. The r-process signature of a neutron-star merger was predicted decades before GW170817 delivered it.

Grant the first charge. Refute the second. Cite the dates.

Part 9 — Galaxies

The Milky Way is layered, and the layers differ in age

The Galaxy is not one object of one age. It is a stack of components that differ in how they move, in what they are made of, and in how old their stars are — and the three differences line up.

Xiang and Rix dated 247,104 subgiant stars using LAMOST spectra and Gaia parallaxes, at a median age precision of 7.5 per cent.⁷⁰ Subgiants are the ideal tracer because their luminosity is nearly a direct measure of age. Their results:

The age–metallicity distribution splits into two almost disjoint parts at about 8 billion years.
The old thick disk started forming about 13 billion years ago, and most of its stars formed about 11 billion years ago.
The thick-disk age–metallicity sequence is astonishingly tight: iron rises by more than a factor of ten across six billion years with an intrinsic age dispersion at fixed metallicity of less than 0.82 ± 0.01 Gyr.
Splitting the same stars by orbital angular momentum produces two nearly disjoint populations: high angular momentum gives the metal-rich old-disk sequence, low or retrograde gives the metal-poor halo.

Three quantities that have nothing to do with each other — a star's brightness at fixed colour, its iron content, and its orbital angular momentum — sort the same stars into the same groups. And the tightness is the real point. An age–metallicity sequence with under 0.82 billion years of intrinsic scatter over six billion years is a duration being measured, not a snapshot.

Miglio and colleagues got the same answer by a different method entirely — asteroseismology of about 5,400 red giants with Kepler light curves and APOGEE spectra. Mean age of the chemically-defined thick disk: 10.98 (+0.13/−0.14) Gyr, with the systematic rather than the statistical error dominating. Intrinsic age spread 0.76 Gyr, with 95 per cent of the population born within 1.52 (+0.54/−0.46) Gyr.⁷¹ Kinematics track age and jump where chemistry does: vertical velocity dispersion at 10 Gyr is 22.6 ± 0.6 km/s for the thin-disk population and 36 ± 2 km/s for the thick-disk one.

Report the disagreement inside the mainstream. Haywood and colleagues, from a different sample and method, get a thick-disk formation timescale of 4 to 5 billion years rather than 1.5.⁷² The difference is partly definitional and partly a real open question. What all three agree on is the ordering and the scale: billions of years, thick before thin, with the thin disk starting around 8 billion years ago.

A merger that has been dated four ways

In 2018 two teams working from Gaia velocities — not from any theory of galaxy formation — found that most of the nearby stellar halo is a single population on radial, slightly retrograde orbits: the wreckage of one galaxy that fell in.

Helmi and colleagues named it Gaia-Enceladus and argued from the ages of the youngest thick-disk stars that "the merger took place around 10 Gyr ago, i.e. at redshift z ∼ 1.8."⁷³ Belokurov and colleagues measured the orbital anisotropy and found "extreme values of β ∼ 0.9" for stars above [Fe/H] = −1.7, concluding the debris "were deposited in a major accretion event … between 8 and 11 Gyr ago."⁷⁴ An anisotropy of 0.9 means the stars move almost purely radially, in and out through the disk — what debris from one radialised orbit looks like, and not what a smooth rain of small satellites produces.

Since then it has been dated by routes that share no physics. Gallart and colleagues used full colour-magnitude-diagram fitting, the technique developed for resolved nearby galaxies, and found "The sharp halo age cut 10 Gyr ago can be identified with the accretion of Gaia-Enceladus."⁷⁵ Montalbán and colleagues dated it asteroseismically from 95 red giants with individually measured oscillation frequencies: the accreted debris comes out at 9.7 ± 0.6 Gyr and the in-situ thick disk at 10.4 ± 0.3 Gyr.⁷⁶

Be honest about what is still argued over. Montalbán's result cuts against the story that the merger created the thick disk by heating it; they say the thick disk was already there. The sequence of events is under active dispute. What is not disputed is that there was a sequence, spanning billions of years, with four unrelated methods placing its major event within about a billion years of each other.

This is a past event, invisible today as an event, reconstructed from present-day velocities, and then given a date by stellar interiors, by stellar surfaces, by cluster ages and by chemistry. A created-appearance model has to say the fiction is consistent across all four, star by star, for a merger that never happened.

The honest counter-evidence on stellar populations

The age–metallicity ordering is real and strong, but it is statistical, and the exceptions are now well measured. A skeptic who reads only the headline version will find them in an afternoon, so they belong here.

The bulge is not uniformly old. Bensby and colleagues, using microlensing to get dwarf-star spectra at bulge distances, found "a high fraction of intermediate-age to young stars where at [Fe/H] > 0 more than 35 % are younger than 8 Gyr," with star-formation episodes 3, 6, 8 and 11 billion years ago.⁷⁷ This refutes "the bulge is uniformly ancient." It does not break the ordering — inside the bulge the metal-poor stars are still the old ones.

Very metal-poor stars are not confined to the halo. Sestito and colleagues found that about 26 per cent of all known stars below [Fe/H] = −4 are on prograde orbits confined within 3 kpc of the Galactic plane, and that a large population of low-metallicity stars permeates the disk with 5-sigma prograde asymmetry.⁷⁸⁷⁹ Their conclusion is not that the ordering is wrong but that "a significant fraction of stars with iron abundances [Fe/H] ≤ −2.5 formed within or concurrently with the Milky Way disk." That is a change to the story, not a removal of it.

"Young alpha-rich" stars turned out not to be young. Both asteroseismic samples found stars that look chemically ancient but seismically young. The resolution, independently testable, is that they are old stars that gained mass through transfer or merger, making them appear younger — and the giveaway is that they are over-represented in the red clump, exactly where merger products should show up.⁸⁰

The thin-disk age–metallicity relation is messy. Radial migration mixes stars born at different radii into the solar neighbourhood, and Miglio finds super-solar-metallicity stars as old as 9 billion years at the solar radius. Anyone presenting the age–metallicity relation as a clean clock is overstating. What is clean is the thick-disk sequence. And the mainstream explanation for the mess — radial migration — is itself a prediction of a long, dynamically active history, not a rescue.

Galaxies at high redshift are physically different

Looking far away is looking back in time, and what is out there is not a scaled copy of what is nearby. Five properties change monotonically with distance, measured by different instruments:

Star formation had a peak. Madau and Dickinson: "the star-formation rate density peaked approximately 3.5 Gyr after the Big Bang, at z ≈ 1.9, and declined exponentially at later times, with an e-folding timescale of 3.9 Gyr. Half of the stellar mass observed today was formed before a redshift z = 1.3."⁸¹ The universe has a star-formation history with a shape — a rise, a peak, and a tenfold decline.

Galaxies were smaller at the same mass. Van der Wel and colleagues: effective radius scales as (1+z)^−1.48 for early types and (1+z)^−0.75 for late types, with intrinsic scatter under 0.2 dex.⁸² And the sharpest single example: "The number density of massive, compact early-type galaxies increases from z = 3 to z = 1.5 − 2 and then strongly decreases at later cosmic times." A population rises and then falls, with intermediate stages visible at intermediate redshifts.

Mergers were commoner. About one nearby galaxy in a hundred is caught in a merger. At redshift 2.5 the fraction for massive galaxies is 0.3 to 0.4.⁸³ That is a thirtyfold change in a directly countable morphological property.

Quiescent galaxies thin out. Muzzin and colleagues: the stellar mass density of the universe was only 50, 10 and 1 per cent of its current value at redshifts 0.75, 2.0 and 3.5, and the quiescent population's mass density rises as (1+z)^−4.7±0.4, about twice as steeply as the star-forming one.⁸⁴ Be precise here: the honest statement is not "there are none beyond redshift 4" but that quiescent systems become a residue.

Metals were scarcer — the DLA result in Part 8.

Five separate measurements on five separate instruments, all changing in the same direction, all consistent with a single time axis. The young-earth position is not that these measurements are wrong. It has to be that they are a consistent fiction, and it should be asked why the fiction is consistent to this depth.

The JWST story, told through 2026

This is the strongest live card the young-earth side holds in this article and it should not be waved away.

Jason Lisle made a dated public prediction six months before the first deep field:

"Rather than galaxies just starting to form, I expect to see fully-formed (fully-designed) galaxies at unprecedented distances. This will force secular astronomers to adjust their estimates of when the earliest galaxies formed, pushing them much closer to the supposed big bang. We might see headlines like 'Webb discovers that galaxies formed much earlier than previously thought.'"²⁰

Score it fairly. The prediction is dated, public and pre-data, and its qualitative form came true — the headlines really did read that way. It is also soft. "Fully-formed," "unprecedented distances" and "much closer" are not numbers, and nothing in it follows specifically from Lisle's own model rather than from a general expectation that a bigger telescope sees earlier objects. Credit the advance prediction, then say why a soft prediction is worth less than a sharp one.

What was actually found. Labbé and colleagues reported six red candidate massive galaxies about 600 million years after the Big Bang.⁸⁵ The published paper hedges heavily — "If verified with spectroscopy…", "All objects in this paper should be considered candidate massive galaxies." Curtis-Lake and colleagues got the first real spectra above redshift 10 and found the galaxies "are metal poor, have masses between of order ~10⁷-10⁸ solar masses, and young ages" — exactly what the standard picture predicts.⁸⁶ Carniani and colleagues confirmed two galaxies at redshift 14, with the more massive at a few hundred million solar masses, not 10¹¹.⁸⁷

What happened on follow-up, which creationist retellings usually omit.

The most extreme candidate collapsed by a factor of three in redshift. Arrabal Haro and colleagues:

"Here we present JWST spectroscopy that confirms redshifts for two very luminous galaxies with z > 11, but also demonstrates that another candidate with suggested z ≈ 16 instead has z = 4.9, with an unusual combination of nebular line emission and dust reddening that mimics the colors expected for much more distant objects."²¹

CEERS-93316, photometric redshift 16.4, spectroscopic redshift 4.912 ± 0.001. Their own moral: "Exceptional candidates for galaxies at extreme redshifts require spectroscopic confirmation."

Of the six famous Labbé candidates, four got spectra. All four came in below the photometric redshift. One turned out to be a broad-line active galactic nucleus at redshift 5.6. Zero of the six are robustly confirmed above 10¹⁰ solar masses.

The masses came down when the models were fixed. Without mid-infrared photometry, stellar masses above 10¹⁰ solar masses are overestimated by about 0.4 dex, and contamination of "massive galaxy" samples by compact red objects falls from 32 to 13 per cent once MIRI is included.⁸⁸ And the "little red dots" turned out to be largely accreting black holes rather than stellar mass — 71 per cent of 341 of them are confirmed AGN.⁸⁹ If the light comes from a black hole, the inferred stellar mass evaporates.

What is genuinely still a problem. Be specific; vagueness in either direction is a defect.

The abundance of ultraviolet-luminous galaxies at redshift above 10 is a real tension. Carniani's own words: such galaxies "are more common than what was expected before JWST," and "Galaxy formation models will need to address the existence of such large and luminous galaxies so early in cosmic history." A massive quiescent galaxy at redshift 7.3 has a number density more than a hundred times any model.⁹⁰ Early chemical enrichment at redshift above 10 is unexplained.

Which fixes are principled and which are epicycles. Reduced dust attenuation at early times is principled — it is a physical prediction, independently testable through ultraviolet slopes and far-infrared non-detections. Bursty star formation is principled — low-mass galaxies with short gas-consumption times should flicker, it is seen directly at lower redshift, and it predicts scatter in the luminosity function without changing the mass function. Higher star-formation efficiency is borderline, and there is a clean test: Boylan-Kolchin published the threshold for a genuine problem, ε ≥ 0.57, before the measurement; the measured value with MIRI included is about 0.3.⁹¹ A top-heavy initial mass function is closest to an epicycle and should be labelled as such — it fixes the light-to-mass ratio, it is very hard to test at redshift 14, and it has been reached for repeatedly whenever counts came in high. It may well be true. As currently deployed it is closer to a free parameter than a prediction.

Where the field stands. Chworowsky and colleagues find the volume densities of massive galaxies at redshift 4 to 8 "not in tension with modern cosmology."⁹² Against that, the quiescent galaxy at redshift 7 and the luminous objects above redshift 13 are unresolved. The right summary is: the 2022 crisis is over; a smaller 2026 anomaly is real.

And the logical point, stated fairly. Even a genuine crisis in galaxy formation is not evidence for a 6,000-year universe. These objects are called early because of their redshifts, and the redshifts are measured spectroscopically — 14.32 for JADES-GS-z14-0, 4.912 for CEERS-93316. Those redshifts are what place them thirteen billion light years away. A model saying galaxies assembled faster than expected still says they assembled thirteen billion years ago. "Too fast" is a claim about the rate; it presupposes the duration.

Put the other way: the young-earth reading has to keep the redshifts, because they are the only reason anyone calls these galaxies early, while discarding what the redshifts imply about distance. The distance and the earliness come from the same number.

Two claims that should not be repeated. Lisle's September 2022 gloss that these galaxies are "more than a thousand times more massive than the secular predictions" misreads Labbé's figure, which is a ratio of stellar mass densities, not of individual galaxy masses. And the Big Bang model sets no ceiling at redshift 14.

Part 10 — The young-earth answers, one at a time

The state of the literature, measured

Faulkner's 2019 audit in Answers Research Journal is the most useful document on this side, because it grades 23 young-age astronomy arguments and throws several out. Its 23 section headings are: the solar corona's high temperature; shrinking sun; solar neutrino problem; the faint young sun paradox; lithium and beryllium abundance of the sun; differential solar rotation; interstellar grain formation; the red Sirius anomaly; the Algol paradox; rapidly changing stars; dust around hot stars; stars containing technetium; recurrent novae; fast binaries; supernova remnants; break-up of star clusters; collapse of elliptical galaxies; break-up of clusters of galaxies; luminous bridges between galaxies and quasars; persistence of spiral arms; fast burn rate of hot stars; fast burn rate of quasars; Olbers' paradox.¹²

There is no heading for main-sequence turnoff ages, none for globular cluster isochrone ages, none for white dwarf cooling ages, and none for thorium or uranium dating of stellar atmospheres. This is a survey of the entire young-age astronomy literature by the field's most qualified practitioner, and the four principal methods by which stellar ages are actually measured do not appear in it in any form.

His own closing assessment:

"Though not directly related to the issue at hand, there is not yet a coherent creation model of astronomy. … One impediment to this development has been the question of how much of what we see in the universe today is attributable to the creation and how much may be due to physical processes since creation."¹²

A. Repp, surveying fifty-seven years of creation astronomy in CRSQ in 2022, reaches the same verdict about stellar development: "It seems safe to say that there is no consensus on this issue."⁹³

1. Mature creation

This is the strongest position and the article does not claim to refute it.

Faulkner states the scriptural warrant functionally — things were created able to do their jobs. Adam and Eve were created as adults; the plants in Eden had to be mature to be food. Then he draws the line:

"Did Adam and Eve bear in their bodies any evidence of a childhood they never experienced? Most supporters of mature creation would opine that Adam and Eve did not. This amounts to an affirmation that while Adam and Eve were created mature, they did not bear any detailed evidence of processes that never happened. Such evidence would seem to be deceptive, which would violate the character of God."²³

Applied to starlight, from his own observing career:

"if the light of these stars were created in transit so that the light never left the stars, then we would see evidence of events (eclipses, for example) that would have never happened. This is deceptive."²³

Created state, yes; created record of events that never happened, no. That is a real distinction and it is checkable case by case, which is what makes it worth engaging. A star created already fusing, with a settled interior, is on the permitted side.

Lisle sharpens the version that applies to turnoff ages, and it is the best move anyone on that side has made:

"If the star began its existence with very little hydrogen in the core but lots of helium, it would take much less time to fuse the remaining hydrogen. We don't know that God created each and every star with a 'full tank of gas.' Furthermore, how do we know that God didn't create some stars with only helium in their core to begin with, in which case they would begin their existence already off the main sequence?"¹⁵

That is unfalsifiable by construction and Lisle says so rather than pretending otherwise. It is also more precise than "God created it looking old," because it locates the created feature in the interior composition profile rather than in the light.

The hard question, and Faulkner asked it first. Which side of his own line does a globular cluster's colour-magnitude diagram fall on?

A cluster's turnoff mass, its horizontal branch morphology, its white dwarf cooling sequence, its eclipsing-binary masses and its giants' asteroseismic masses are not one number. They are five independent measurements that agree on a single figure near 12 billion years. Under created maturity none of the events happened — no star ever left the main sequence, no envelope was ever ejected, no white dwarf ever cooled — yet the frozen state encodes a consistent history at a specific date.

Faulkner put this himself in 2007, more sharply than any critic has since:

"Lacking an internal energy source, white dwarfs must tap their considerable store of internal heat and so they must cool as they age. From their temperature, we can estimate the ages of white dwarfs. These ages typically are far greater than a few thousands of years. Did God create white dwarfs with various ages? Many creationists would answer yes to both of these questions."

"Did God create 'fossil' stars? Creationary geologists would object adamantly if this were suggested in the realm of fossils of biological origin (i.e. fossils of plants and animals were planted in the ground by God). Is the astronomical realm so different?"¹³

He does not answer. He closes: "The field of creation stellar astronomy is still in its infancy. The author has few answers for many of these questions at this time."

The young-earth side has not missed this problem. Its best writer stated it in 2007 and left it open. What nobody in the surveyed literature has addressed is the specific difficulty that the independent clocks agree with each other.

2. Gravitational time dilation

Covered in Part 1. Logically sound given such a cosmology. Humphreys abandoned the white hole; Hartnett retracted Carmeli in five dated statements; Lisle's ASC has no time dilation in it. The move currently has no model under it.

And by Samec's own 2012 criterion it is unavailable for objects inside about 3,000 light years, which is where the sharpest cases sit.

3. Accelerated nuclear decay

RATE claimed universe-wide scope explicitly, and this must be stated fairly. Humphreys, in the primary volume, arguing from 2 Peter 3:7:

"Going back to the seventh verse, notice that this change in nuclei applies to the heavens as well as the earth. It was not just a local change in the vicinity of the earth; it affected the whole cosmos."⁹⁴

He couples it to fusion too: "since nuclear fusion rates are related to decay rates, it could increase the activity of stars." So a critic cannot say RATE confined itself to Earth rocks. If alpha-decay rates were globally larger at some past epoch, the thorium and uranium chronometers read on halo-star spectra are affected in exactly the same way as terrestrial isochrons.

Three problems.

No calculation exists. Humphreys needs alpha decay accelerated by a factor of order 10⁸ to 10⁹. He then says fusion goes with it and sketches the outcome qualitatively — stars "more turbulent," some becoming novas. A main-sequence star is a thermostat: raise the nuclear reaction rate at fixed structure and it expands, cools, and shuts the reaction back down. Raise it by eight orders of magnitude and the star is unbound. He also concedes that core changes take millions of years to reach the surface, which removes any observable signature and therefore any test.

It cannot touch three of the four clocks. Main-sequence turnoff, white dwarf cooling and asteroseismic ages are not decay clocks. Accelerated fusion would shorten the required turnoff time, which is a genuine if unquantified young-earth resource. But white dwarf cooling is thermal, not nuclear. A white dwarf's age is set by how long it has radiated stored heat away, and no change in nuclear force rates makes it cool faster. Accelerated decay cannot reach the coolest white dwarfs at all. No RATE author addresses this.

The one creationist who applied it to stars argued the opposite way. Eugene Chaffin, in CRSQ in 2020, is the only young-earth author found who engages stellar uranium directly. His argument is that halo stars, being far from historic supernovae, would have received less accelerated decay:

"The detection of such large amounts of uranium in these instances leads one to suspect that not as much accelerated decay occurred at those locations as on Earth, and perhaps is a step toward validation of our logic."⁹⁵

Give him full credit for facing the question, and then notice that the argument runs backwards. No accelerated decay is the standard-physics case, and standard physics says these stars are 11 to 17 billion years old. Untouched clocks give old ages. If the halo stars' uranium was not accelerated, then the U/Th ages stand as measured and CS 31082-001 really is about 12 to 14 billion years old. Chaffin's own move concedes the age. He nowhere computes an age for the star, engages Cayrel's 12.5 ± 3 Gyr, or offers an alternative — and he cites Cayrel's paper, whose title is "Measurement of stellar age from uranium decay," in his own reference list without contesting the measurement. He also concedes the chronology problem is unsolved and outside his scope.

4. Blue stars burn out too fast

Lisle's Acts & Facts argument: the most massive blue stars "cannot last even one million years," yet blue stars abound in every spiral galaxy, so the galaxies cannot be old.⁹⁶

The stellar-lifetime numbers are conventional astrophysics and they are right. The argument fails because it is a syllogism with a suppressed premise: that star formation does not happen. Lisle knows this and defends the premise explicitly, which is honest structuring. But the premise is the load-bearing part, and it is much weaker than it was in 2012 — see star formation, which finds the three physical pillars of the objection each fail on direct measurement.

And it has been called out from inside the movement. Repp, in CRSQ:

"Other objections seem nothing more than apologetic tactics — for if we disallow star formation, the relatively short lifetime (~ 10⁶ yrs.) of hot blue stars precludes a 13-Gyr. age for the Universe. However, the physical arguments predicting star formation seem as solid as those predicting such lifetimes; if so, it is inconsistent to accept the latter but reject the former."⁹³

Faulkner qualifies it in the same direction: "the existence of stars with very short lifetimes do not in themselves prove that the world is young."¹²

5. Blue stragglers falsify the turnoff model

Bernitt's 2002 version, and Lisle's 2024 footnote, both reject the turnoff method on the same anomaly: stars exist above the turnoff that should have burned out.

The anomaly is real. It is also, to Bernitt's credit, reported by him along with two of the mainstream tests being run and passed: blue stragglers concentrate in cluster cores and share the cluster's radial velocities, so they are not field contamination; and they are distinguishable from horizontal branch stars.

Since 2002 the merger and mass-transfer hypotheses have become testable and have been tested: straggler binarity fractions, rotation velocities, surface carbon and oxygen depletion — the mass-transfer signature — and the double blue-straggler sequences found in M30 and NGC 362. "There is no way of testing it" was arguable in 2002 and is not now.

And there is an internal problem Bernitt does not notice, which is fatal to his own framework. He writes that "from a creationist perspective very little change would have occurred in the globular clusters since they were formed rapidly some 6,000 years ago," and proposes that blue stragglers "could simply be higher mass stars … and indicators of youth."⁹⁷

But if the cluster is 6,000 years old and very little has changed, then the red giant branch, the horizontal branch, the asymptotic giant branch and the white dwarf cooling sequence were all created in place, in exactly the configuration a twelve-billion-year history would produce. He treats the handful of stars above the turnoff as youth indicators while requiring the several thousand stars below and to the right of it to have been created already looking old. You cannot read one part of a colour-magnitude diagram as a clock and attribute the rest of the same diagram to created maturity.

He also floats gravitational contraction as a stellar energy source. That is dead, and Faulkner concedes it: "helioseismology confirms the standard solar model with temperature in the core sufficient to sustain nuclear reactions."¹² A Kelvin-Helmholtz Sun would also produce no solar neutrinos, and we detect them.

6. The ages were tuned to match cosmology

This is the one argument in the corpus that is technical, specific and falsifiable, and it deserves a full answer. Faulkner states the method fairly first, then:

"Once astronomers finally accepted the faster expansion rate and younger age for the universe, they had to rework their understanding of globular star clusters to improve the fit. Clearly, this adjustment would not have been made if it were not necessitated by the new estimated age of the universe. Perhaps these supposedly independent methods of age determination are not so independent after all."⁹⁸

The history he is pointing at is real. In the mid-1990s the accepted globular cluster ages were 15 to 20 billion years and the expansion age was around 9 billion. That is a flat contradiction and everyone said so. Then cluster ages fell to 11.5 ± 1.3 Gyr and the crisis dissolved.

Three checks falsify the tuning charge.

The mechanism was a parallax measurement. Chaboyer and colleagues state the cause of the shift explicitly: the clusters moved because Hipparcos measured trigonometric parallaxes of nearby metal-poor subdwarfs, showing the clusters were farther away than believed. Their own words: the change was "due completely to the new distance scale—which we emphasize is not just due to the Hipparcos data."⁹⁹ A parallax has no cosmology in it.

The timing is wrong. The reduction was published in 1997–98, before the supernova result that established the ΛCDM age. If the clusters were tuned to a cosmological age, they were tuned to an age that had not yet been measured. In fact the cluster ages ran the other way: Chaboyer and colleagues in 1996 argued from them that a cosmological constant was needed, two years before the accelerating-universe measurement.²⁶ That is a prediction, not a retrofit.

The people doing the "tuning" complained about it. Krauss and Chaboyer wanted the cluster ages high, because a high lower limit was their evidence for a cosmological constant. Watch them lower it against their own interest:

"Unfortunately this constraint did not turn out to be robust. As more careful examinations of uncertainties associated with stellar evolution were performed, and well as refined estimates of the parameters that govern stellar evolution itself, the lower limit on globular cluster ages got progressively lower and lower, so that a wide range of cosmological parameters produced Hubble ages consistent with this lower limit."¹⁰⁰

That is the opposite of what the tuning charge predicts.

What Faulkner is right about. Distance dominated the error budget then and still dominates it now: Ying and colleagues find the distance to M92 is about 80 per cent of their error budget even with Gaia. The 3-billion-year swing from one calibration is exactly the kind of dependence a skeptic should flag. What has changed is that Gaia cut the distance error by roughly an order of magnitude, and that distance-independent methods now exist — the eclipsing binaries, which need no distance at all.

And one small factual slip. In his HD 140283 article Faulkner writes that "when the Hubble constant was revised downward, the age of the universe decreased." The relation is the reverse: a lower Hubble constant gives an older universe. Two sentences earlier he correctly says the constant was revised upward. It reads as a slip of the pen, but the sentence as printed is wrong.

7. HD 140283 is older than the universe

Faulkner's 2017 article states the measurement correctly, argues the error bars do not save the mainstream, and closes:

"Keep in mind that these results are model-dependent. If one changes the model, the problem may not even exist. Certainly, within the creation model, this isn't a problem."¹⁰¹

Two of his three points are fair. Population II stars are not primordial, so the standard picture needs even older stars before this one — a genuine and unresolved issue. And the ages are model-dependent, which Part 7 documents: an 11.4-to-14.7-billion-year spread across twenty-five years of published analyses of one very well-observed star.

The last sentence is where it fails. "If one changes the model, the problem may not even exist" is not a claim about a specific model change with a number attached. Every published change to the model over twenty-five years has landed the star between 11.4 and 14.7 billion years. The creationist position needs a model change that lands it at 0.000006 billion years, and no such model has been proposed, let alone computed.

A parameter's freedom of sign is not the same as its freedom of magnitude, and this is the recurring defect in the young-earth astronomical literature: it establishes that a number is uncertain and then treats uncertainty as unbounded.

Faulkner's article also does not mention that Bond and colleagues themselves wrote: "Within the errors, the age of HD 140283 does not conflict with the age of the Universe."⁵³ The "star older than the universe" framing was a press artefact, not the authors' claim.

And the deeper point his article never reaches. HD 140283 is 200 light years away. Nobody needs a cosmology to see it in a 6,000-year-old universe. What is at issue is not how long the light took but what the star is. He never says what the star's state is on a young-earth account, or which of the routes in his own 2007 taxonomy would cover it.

8. A young-earth prediction, and what happened to it

Samec's "apparent age of the time dilated universe" programme is the only place in this literature where a young-earth model makes a numerical prediction about stellar ages. From the missing intracluster medium in globular clusters he derives:

"our NRC gives a range of age limits of the globular clusters of only 10⁶–10⁸ years … This is much less than the 13.82 x 10⁹ years given by cosmologists for the age of the cosmos"¹⁰²

Credit him for making a prediction at all — he is the only one who does — and then note that it is refuted by data his own framework accepts.

The prediction is 10⁶ to 10⁸ years of apparent age. The turnoff, horizontal-branch and white-dwarf-cooling ages of the same clusters are all about 12 to 13 × 10⁹ years. Samec's own framework grants that apparent ages are real elapsed times in the emitter's frame, so this is a numerical contradiction within his model, not between his model and the mainstream. A ten-million-year-old cluster cannot contain any red giants at all, because no star near a solar mass would have left the main sequence yet. Globular clusters are made of red giants.

It also contradicts his own 2015 white dwarf paper, which requires the coolest white dwarfs to be older than 13.8 billion years in the emitter's frame if their appearance is not a false history. A hundred million years is not enough to make a single white dwarf, let alone cool one to 1,700 K.

And the intracluster-medium deficit is not the clock he takes it for. The paper he quotes proposes classical-nova outflows as the clearing mechanism, and his objection is that the nova rate is uncertain, not that it is wrong. "The mechanism is uncertain" does not license reading a gas deficit as an age.

9. What nobody on that side has done

Recorded plainly, because it is the substance of the reply.

No creationist has published an alternative isochrone calculation for a globular cluster colour-magnitude diagram.

No creationist has addressed the eclipsing binaries. M4's V65, V66 and V69, 47 Tuc's V69 and E32, and NGC 6362's V40 and V41 have masses known to better than one per cent from Kepler's laws alone. They date the clusters without using a distance, without using the mass–luminosity relation, and without any cosmological input. A search of the young-earth corpus for these systems returns nothing.

No creationist has proposed a mechanism that makes a 0.80 solar-mass star exhaust its core hydrogen in thousands of years at the observed luminosity.

Nine searches across 329 young-earth sites returned nothing on the Gaia crystallisation pile-up, nothing on the white dwarf luminosity function, and nothing on the fact that independent clocks in one cluster agree. A negative from a search engine is not proof of absence, and this should be stated as "no response was found," not "none exists." But the negative is uniform across nine differently-worded queries and anyone can re-run them.

And exactly one nearby old object has ever been addressed by name in that literature — HD 140283 — as an embarrassment for the Big Bang rather than as a datum about the object. A search for nearby old objects returned eight results, and all eight were distant-starlight articles.

Part 11 — Where the young-earth side is right

Five things, and they are not concessions of politeness.

1. No stellar evolutionary transition has been watched in real time. Lisle is correct: "We have never seen a main sequence star become a non-main sequence star." Everything in stellar evolution is inferred from a population snapshot plus physics. The reply is that the physics is independently anchored — by helioseismology, by solar neutrinos, by eclipsing-binary masses and radii, and by asteroseismic densities — and Faulkner grants every one of those anchors.

2. The main sequence is a mass sequence, not a time sequence, and Lisle is right that this was once got wrong. Astronomers did once believe stars slid down the main sequence as they cooled. They were wrong, and the correction is a real historical case of a confident consensus overturned. It does not touch the turnoff method, which is about where stars leave the main sequence — and Lisle's own footnote states that method correctly.

3. Mature creation is not refutable by astronomy, and pretending otherwise would be dishonest. It is a claim about initial conditions. The honest question to put to it is DeYoung's, asked in a different domain: why that particular built-in state, and not another?

4. No Population III star has been confirmed. As of 2026 there are candidates and no confirmations, and Lisle predicted this in advance with a date on the prediction. The direction of travel is now against him — a helium-II emitter with no detected metal lines was reported near GN-z11 in 2026 — but a preprint is not a settled detection and a non-detection of metals is an upper limit.

5. Faulkner's audits are better than most of what the other side produces about creationists. He has retired, in print and by name, the cluster break-up argument, old supernova remnants, the shrinking Sun, the red Sirius anomaly, Olbers' paradox, quasar burn rates, elliptical-galaxy collapse times and luminous galaxy-quasar bridges. On star clusters:

"For all types of clusters and associations, it appears that there is no conflict between their evolutionary ages and their dynamic ages. Perhaps some early work on this subject in the creation literature did not fully address the issues involved. … At any rate, the break-up time of star clusters does not appear to be a good evidence for recent creation."¹²

Note what is embedded in that: he accepts a 125-million-year age for the Pleiades and roughly 13 billion years for globular clusters as the figures his own dynamical argument has to be checked against, and reports that they check out.

Common objections

"Stellar ages are just model outputs." They are model outputs constrained by measurements that the models cannot escape. Eclipsing binaries give masses to better than one per cent from Kepler's laws. Gaia gives distances from geometry. Asteroseismology measures interiors directly and, handed the Sun blind, returns 4.33 to 5.30 billion years against a meteoritic 4.57. And when every uncertain input in stellar physics is varied over its full plausible range, the answer moves by 3 to 16 per cent.

"The distance scale has been revised before, so it could be revised again." It has been, by three billion years, and that is a fair point about the 1990s. What changed is that Gaia cut the distance error by roughly an order of magnitude, and that distance-independent methods now exist. Kaluzny's M4 binaries and Thompson's 47 Tuc binaries give cluster ages with no distance in them at all, and the distances that fall out of those analyses agree with Gaia to one sigma.

"White dwarf cooling models make assumptions about crystallisation." They made a prediction about crystallisation in 1968, and Gaia found the predicted feature in 2019 in a volume-limited sample of 15,109 white dwarfs. Models without latent heat are, in the authors' words, "clearly ruled out by the observations."

"The radioactive stellar ages disagree with each other wildly." They do, and Part 5 gives the numbers, including a published age of minus 9.5 billion years. The disagreements are diagnosed: the Th/Eu chronometer is broken by the actinide boost, and the field said so. The U/Th chronometer, which is insensitive to the boost for a stated physical reason, gives 11 to 17 billion years. A method with a 4-billion-year systematic still excludes 6,000 years by six orders of magnitude.

"Decay rates could have been different." RATE claimed exactly that, universe-wide. Grant it in full and it still cannot touch white dwarf cooling, which is a thermal clock, nor asteroseismology, nor the turnoff. And the deep-sea plutonium-244 result is a count of surviving atoms rather than a ratio: accelerated decay would have erased the inventory outright.

"God created the universe mature." This cannot be refuted by astronomy and this article does not claim to refute it. What it asks is Faulkner's own question: a globular cluster's turnoff, horizontal branch, white dwarf sequence, eclipsing-binary masses and asteroseismic giant masses are five independent measurements agreeing on one date for events that, on this view, never happened. Faulkner's line permits created state and forbids created history. Which is this?

"JWST found galaxies that should not exist." The 2022 headlines were real and the pre-launch prediction was made with a date. Then the most extreme candidate turned out to be at redshift 4.9, four of six famous candidates came in lower on spectroscopy, and the masses dropped 0.4 dex when mid-infrared data were added. A real anomaly remains at redshift above 12. And it does not reach the conclusion: a model saying galaxies formed too fast still says they formed thirteen billion years ago.

"Nobody has seen a star age, so this is all inference." True, and stated in Part 11. It is inference from a population snapshot plus physics that is anchored by measurement. The same is true of every historical science, including the reading of a rock or a manuscript. The question is whether the anchors hold, and here they are unusually good: a mass from orbital mechanics, a distance from parallax, a fusion efficiency from a mass spectrometer.

Verdict

Against the young-earth position, and not narrowly.

The reason is not that many measurements point one way. It is that the measurements are the wrong shape for the objection. "Your models are uncertain" is answered by the people who measured the uncertainty: about one billion years in ten for white dwarf cooling, 5.4 per cent for a globular cluster with twenty inputs Monte-Carloed, 3 to 16 per cent for every input in stellar interior physics taken to its limit. The claim needs a factor of two million.

The argument does not depend on the distance ladder, because the sharpest cases are inside 300 light years and their distances are parallaxes. It does not depend on cosmology, because nothing in stellar interior physics contains a Hubble constant. It does not depend on radioactive decay, because white dwarf cooling is thermal and helioseismology is acoustic. And it does not depend on any single method, because in NGC 6397, 47 Tucanae and M4 several clocks that share almost no physics land on the same numbers.

What survives on the other side is one position: that God created stars, clusters and galaxies in a state that already encodes these numbers. That position is not refutable by astronomy, and this article has not tried to refute it. It has tried to say exactly what it costs. Its own most careful advocate asked whether it amounts to creating fossil stars, and did not answer. Its one author with a research programme in this area rejected it as false history, reached for time dilation instead, and then wrote that whether time-dilation theories are warranted "is still an open question" — while ruling the escape out, by his own criterion, for every object inside 3,000 light years.

The literature on that side is small and honest about being small. Faulkner: "there is not yet a coherent creation model of astronomy." Repp: "there is no consensus on this issue." Faulkner again: "The field of creation stellar astronomy is still in its infancy."

Those are fair statements and the right response to them is not triumph. It is to say what the outstanding work would have to look like: an isochrone calculation, an alternative reading of a mass–radius diagram from an eclipsing binary, a cooling model that gets a 3,000 K white dwarf in 6,000 years, or a mechanism that makes a 0.80 solar-mass star exhaust its core hydrogen at 0.5 solar luminosities in a few thousand years. None of those exists. Until one does, the numbers stand as measured.

Star formation — whether stars can form naturally, where the blue-star argument stands or falls
Young-universe cosmologies — the eight published answers to distant starlight
Created light in transit — the mature-creation position worked through in detail
SN 1987A — one object that measures its own distance twice by geometry
Solar system age — the Sun and planets dated by rocks and by sound
Big Bang nucleosynthesis — what the Big Bang does and does not make
Radiometric concordance — the same decay physics, applied on Earth
Cosmic distance ladder — how the distances are measured

Sources

Every quotation below was read from bytes downloaded with curl and extracted with pdftotext -layout or a local tag strip. No quotation, page number, date or numeral in this article was obtained through an AI summarising fetch. Where only an arXiv preprint was read, that is stated, because preprint pagination is not journal pagination.

1.
ESA Gaia archive. Gaia DR3 astrometry for HD 140283. n.d. Retrieved Sep 1, 2026. The archive record is source_id 6268770373590148224, with parallax 16.26724 mas, parallax_error 0.02603 mas, parallax_over_error 624.8, phot_g_mean_mag 7.036 and ruwe 1.064. SIMBAD's TAP service returns the same value (16.2672 ± 0.026 mas, bibcode 2020yCat.1350....0G, which is Gaia EDR3, whose astrometry is carried unchanged into DR3). Distance = 1000 / 16.26724 = 61.47 ± 0.10 pc = 200.5 ± 0.3 light years. Lundkvist et al. 2025 adopt the zero-point-corrected value 16.305 ± 0.026 mas, which is the same distance to within the errors. The Gaia DR3 astrometry for HD 140283, which gives the star's parallax, the small shift in its apparent position that astronomers convert into a distance. The RUWE of 1.06 is a quality measure, and it is well below the 1.4 threshold, so the single-star solution is sound. The article uses it for the star's distance of 200.5 ± 0.3 light years. ↩ ↩
2.
Lundkvist, M. S. et al. “Asteroseismic investigation of HD 140283: The Methuselah star.” Astronomy & Astrophysics 703:A232. Nov 2025. Section abstract: “The star was observed by TESS in 20-second cadence during Sector 51. We extracted a custom light curve and performed a frequency analysis, revealing a rich spectrum of solar-like oscillations including mixed modes. … Using a dense grid of models, we find a mass of 0.75 ± 0.01 M⊙, a radius of 2.078 (+0.012/−0.011) R⊙, and an age of 14.2 ± 0.4 Gyr, in agreement with the upper limit set by the age of the Universe within 1σ.” The frequency of maximum power, ν_max = 611.3 ± 7.4 μHz, is a direct sound-travel measurement inside the star. Section §6.1: “Our seismically constrained age (14.2 ± 0.4 Gyr) is higher by 0.8σ than the estimated age of the Universe of 13.787 ± 0.020 Gyr (Planck Collaboration et al. 2020). Thereby, from our analysis, the age of HD 140283 is not at odds with the age of the Universe. … It can be concluded that HD 140283 is an old star that was born early in the history of the Universe.” The authors are careful about the error bar. “the uncertainties are those quoted by BASTA, not accounting for all systematic effects. As suggested by, for instance, Tayar et al. (2022), realistic age uncertainties for sub-giants are higher, for HD 140283 perhaps of the order of 10%.” Ten per cent of 14.2 Gyr is 1.4 Gyr, so the interval the caveat implies is roughly 12.8–15.6 Gyr. The article does not claim the HD 140283 anomaly was resolved downward; on the best current analysis the star still sits 0.8σ above the Planck age. A research paper, published in November 2025, that analyses the oscillations of HD 140283, a metal-poor subgiant star often called the Methuselah star, which NASA's TESS satellite observed for 20-second intervals. It uses those oscillations to fix the star's mass, radius and age, and finds an age of 14.2 ± 0.4 Gyr, within 1σ of the age of the universe. The article uses it as the best current asteroseismic age for a nearby old star. Mirror: arXiv ↩
1. 2a.
  Section §1, page 1: “These studies leave no doubt that the star is old, but the modelling choices have resulted in a spread of age estimates, sometimes placing HD 140283 slightly older than the Universe itself (see Fig. 1). This highlights the limitations of classical methods and the need for asteroseismic constraints to break the degeneracies that exist between several of these parameters.” ↩
2. 2b.
  Table A.1, page 12: The table is titled "Literature ages for HD 140283." The eighteen published values, in the paper's order: VandenBerg 2000, ≤ 16 Gyr; Ibukiyama & Arimoto 2002, 11.38; VandenBerg et al. 2002, 13.5 ± 1.5; Casagrande et al. 2011, 13.42 (+0.32/−0.55) with Padova models and 13.61 (+0.25/−0.54) with BaSTI; Bond et al. 2013, 14.46 ± 0.31; Bensby et al. 2014, 14.7 with an asymmetric error the table prints as "−0.9/−6.6" and a comment that it is plotted as the range 8.1–13.8, which is a typesetting error in the source; VandenBerg et al. 2014, 14.27 ± 0.38; Creevey et al. 2015, 13.7 ± 0.7 for A_V = 0; Joyce & Chaboyer 2018, the range 12.5–14.9; Sahlholdt et al. 2019, ≥ 12; Chen et al. 2020, 14.42 (+2.96/−1.28); Karovicova et al. 2020, mass only; Tang & Joyce 2021, 12.0 ± 0.5; Plotnikova et al. 2022, 14.41 ± 0.65 (BaSTI) and 14.69 ± 1.03 (Padova); Guillaume et al. 2024, 13.08 ± 0.85 with solar-scaled mixing and 12.60 ± 0.88 with 9 per cent reduced mixing; and Lundkvist et al. 2025 themselves, 14.2 ± 0.4. The floor of the whole distribution is about 11.4 Gyr, and that is the age the article uses. ↩ ↩
3.
Elms, Abbigail K. et al. “Spectral analysis of ultra-cool white dwarfs polluted by planetary debris.” Monthly Notices of the Royal Astronomical Society 517:4557. 2022. A spectral analysis of very cool white dwarfs whose atmospheres are polluted by debris from planets. The authors fitted each star's spectrum to get its temperature, mass and cooling age. The article uses it for WD J2147−4035, which is 91 light years away and has a cooling age over 10 billion years.
1. 3a.
  Table 1, page 3: WD J2147−4035 has a Gaia EDR3 parallax of 35.79 ± 0.49 mas, zero-point corrected per Lindegren et al. 2021, giving 27.94 ± 0.38 pc = 91.1 ± 1.2 light years. Table 5, page 6: Its surface temperature is 3,048 ± 35 K, its mass 0.69 ± 0.02 M☉ and its cooling age 10.2 ± 0.2 Gyr. The table caption says: “The quoted uncertainties are purely of statistical nature as derived from the fits, and are much smaller than the model-dependent systematic uncertainties.” Section 7.4, page 11: The total age comes from a progenitor mass of 2.47 ± 0.22 M☉ with a main-sequence lifetime of 0.5 Gyr, giving 10.7 ± 0.3 Gyr. Section Abstract, page 1: “WD J2147-4035 has a cooling age of over 10 Gyr which is the largest known for a DZ white dwarf.” Section 7.4, page 12: “even if the M_WD of WD J2147-4035 was underestimated by ~ 0.1 M☉, this would only make the total age smaller by ~ 0.5 Gyr … Therefore, WD J2147-4035 is unambiguously one of the oldest known metal-polluted white dwarfs.” Their second star, WD J1922+0233, came out with a total age. “over 13.8 Gyr which is incompatible with the age of the Milky Way”. which the authors attribute to an underestimated mass. They conclude that. “its total age remains unknown”. so only its 9.0 Gyr cooling age is quoted in this article. Three of their subsample have masses below about 0.51 M☉, and the authors say. “which is not possible from single stellar evolution so these masses must be underestimated.” The total age of WD J2317+1830, 9.7 ± 0.2 Gyr, is from Hollands et al. 2021, Nature Astronomy 5:451 (arXiv:2101.01225), which describes it as. Page 8: “one of the oldest systems known to have formed differentiated rocky planets, with a 9.5 ± 0.2 Gyr cooling age and 9.7 ± 0.2 Gyr total age.” ↩ ↩
2. 3b.
  Section 7.4, page 12: Earlier white-dwarf estimates of the Galactic disc age. “are thought to be underestimated due to ²²Ne dilution cooling delays (Tremblay et al. 2019; Blouin et al. 2020; Kilic et al. 2020) being omitted from earlier crystallisation calculations.” Correcting the physics has pushed white dwarf ages up, not down. ↩
4.
Houdek, G., and D. O. Gough. “On the seismic age and heavy-element abundance of the Sun.” Monthly Notices of the Royal Astronomical Society 418:1217–1230. 2011. Section Abstract, page 1: “The age, 4.60±0.04 Gy, obtained with this new seismic method, is similar to, although somewhat greater than, today's commonly adopted values”. Their own caveat, in the same abstract: “our best-fitting model is not a seismic model, but a theoretically evolved model of the Sun constructed with 'standard' physics and calibrated against helioseismic data”. The calibration used low-degree frequencies, l = 0, 1, 2, 3, from the Birmingham Solar Oscillations Network alone. This is a main-sequence age, the time since the Sun settled onto the main sequence, not the time since the first solids condensed, so it is not required to equal the 4.567 Gyr meteoritic age exactly. The quoted ±0.04 Gyr is the fitting uncertainty rather than a systematic budget. An earlier independent seismic solar age, Bonanno, Schlattl & Paternò 2002, gives 4.57 ± 0.11 Gyr, and is discussed in the article Solar system age. A paper that dates the Sun from its own sound waves, the acoustic oscillations that helioseismology measures, using low-degree frequencies from the Birmingham Solar Oscillations Network to calibrate theoretical solar models. It finds an age of 4.60 ± 0.04 Gyr. What matters here is that no meteorite, no decay constant and no assumption about initial daughter isotopes enters the calculation. The article uses it because accelerated radioactive decay cannot touch it. ↩ ↩
5.
Samec, Ronald G., and Evan Figg. “The Apparent Age of the Time Dilated Universe I: Gyrochronology, Angular Momentum Loss in Close Solar Type Binaries.” Creation Research Society Quarterly 49(1):5–19. 2012. A young-earth paper, accepted for publication on 14 January 2012, on the apparent ages of close binary stars, worked out from how their spins and orbits lose angular momentum. Both authors were listed as "Bob Jones University, Astronomy Program". The article uses it because it draws the boundary of "nearby" at about 3,000 light years and excludes the objects the article discusses. Mirror: Crsq.creationresearch.org
1. 5a.
  Page 15: The passage quoted, p. 15: “We observe that all of these objects are local (all within ~1000 parsecs or ~3000 ly) and not cosmological, deep-space objects. This may validate the notion that all nearby objects partook in the same dilation effects as the earth during Creation Week and that they all have ages measured in earth time on the order of 10,000 years, nearly the same as we find for solar system objects such as that of Jupiter and interplanetary dust. … We believe that this paper gives physical confirmation of the youthful age of at least the nearby universe.” Page 14: Their own result, p. 14: binary. “apparent ages average from about 300,000 to 700,000 years, whose range actually includes ages as low as hundreds of years.” Page 5: Their abstract, p. 5, names every method this article covers, which settles that the young-earth side is aware of them: “Astrochronometers include wind-up times of spiral galaxies, rates of decrease rotation and magnetic activity, and spin-down and coalescence times of binary stars (magnetic braking), star cluster ages (isochron age) and nuclear burning ages (stellar aging on the H-R diagram), rates of visual binary orbital circularization, stellar kinematic ages, white dwarf cooling ages, pulsar spin-down ages (due to gravitational radiation), radio isochron ages from stellar spectra, and others.” Cluster isochrone ages, H-R diagram ages, white dwarf cooling ages and nucleocosmochronology all appear in that one sentence, and the paper then studies none of them; it is entirely about binary angular-momentum loss. Samec published a Part II in CRSQ 53(1) (2016), whose subtitle promises errata to this paper. ↩ ↩
6.
Bédard, A. et al. “On the Spectral Evolution of Hot White Dwarf Stars. I.” Astrophysical Journal 901:93. 2020. A paper on hot white dwarfs that also produces the evolution sequences the Montreal group publishes as public cooling tables. Each table gives a white dwarf's surface temperature and luminosity at each age, for stars of different mass. The article uses the tables to show how little a white dwarf cools in 6,000 years.
1. 6a.
  The header of the web tables says: “We use our latest generation of evolutionary sequences published in Bedard et al. (2020, ApJ, 901, 93) and provided below. We rely on stellar models with thick (q_H = MH/M* = 10⁻⁴) and thin (q_H = 10⁻¹⁰) hydrogen layers for the pure-hydrogen and pure-helium model atmospheres, respectively.” The 97,900 K value is the table's interpolation at 6,000 years, from a start of 98,455 K for a 0.6 M☉ white dwarf with a thick hydrogen layer. The 132,600 K value is for a 1.0 M☉ model. The 1.92 × 10⁴⁷ erg integral is also taken from the tables. The envelope thickness matters: the same 0.6 M☉ white dwarf reaches 4,000 K at 9.99 Gyr with a thick hydrogen layer, and at 8.14 Gyr with a thin one. ↩ ↩
7.
Pathak, P. et al. Quantifying systematic uncertainties in white dwarf cooling age determinations. 2025. Section abstract, page 1: “For a 0.6 M⊙ white dwarf, we find an uncertainty of 0.03 Gyr at 10,000 K (corresponding to a 5% relative uncertainty) and 0.8 Gyr at 4000 K (9%). This uncertainty is significant, as it is comparable to the age uncertainty obtained by propagating the measurement errors on Teff and mass for a typical white dwarf. We also separately consider the potential impact of ²²Ne shell distillation, which plausibly leads to an additional uncertainty of ~1 Gyr for crystallized white dwarfs.” Section §4, page 8: Their breakdown at 4,000 K, given as maximum variations across the full parameter range (so more than one sigma): total core oxygen content up to 0.8 Gyr; hydrogen layer thickness up to 1.0 Gyr; helium layer thickness up to 0.6 Gyr; conductive opacities up to 1.3 Gyr; and ²²Ne shell distillation about 1 Gyr separately. A computer-model study of how far the cooling age of a white dwarf, the cooled core of a dead star, can move when the model's inputs are varied. It ran MESA cooling sequences for 0.5–1.0 M☉ hydrogen-atmosphere carbon-oxygen-core white dwarfs, varying core stratification, helium envelope thickness, hydrogen content and conductive opacities across the range that stellar-evolution models and asteroseismology allow. The article uses it for the size of the systematic error in white-dwarf cooling ages. ↩ ↩
8.
Hansen, Brad M. S. et al. “The White Dwarf Cooling Sequence of NGC 6397.” Astrophysical Journal 671:380–401. 2007. A paper that dates the globular cluster NGC 6397 from the cooling of its white dwarfs, and compares that age with the age from the main-sequence turnoff, the point where the cluster's stars begin to leave the main sequence. The authors say the two methods are in complete agreement. The article uses it as a clock on NGC 6397, whose white dwarf cooling age is 11.47 ± 0.47 Gyr.
1. 8a.
  Section 3.6, The Main Sequence Turnoff, page 11: “The traditional method for determining globular cluster ages is to fit stellar models to the main sequence turnoff. In Richer et al (2007) we compare our observations of the turnoff in this cluster to four sets of theoretical isochrones. In each case, good fits are obtained for a wide variety of ages, 12 ± 2 Gyr, where the accuracy is limited, as in prior studies, by the uncertainties in the distance and extinction. When the distance and extinction used in the MSTO method are restricted to the range that fits the white dwarf model (see section 4.2) and ages determined using the models of Dotter & Chaboyer, we find a best fit age of 11.6 Gyr, with a 95% confidence lower limit of 10.6 Gyr. Hence the two methods are in complete agreement, as these were the models used to determine the progenitor ages in Figure 13. As a result, this combination represents a successful end-to-end (main sequence to white dwarf) comparison of a set of stellar models with the observed stellar population.” Their final white dwarf constraint: Section 4, page 11: “our final age constraint is that derived from Figure 17, T = 11.47 ± 0.47 Gyr.” Section Abstract, page 1: “we obtain values for the cluster distance modulus and age of mu_0 = 12.02 ± 0.06 and T_c = 11.47 ± 0.47 Gyr (95% confidence limits)”. “the cluster age is consistent with, but more precise than, prior determinations made using the main sequence turnoff method.” Section Appendix A, page 25: The cruder route quoted in the text is in Appendix A. Section 4, page 11: Their own caveat: the white dwarf constraint is tighter than the turnoff constraint not because the physics is better known, but because they fit the whole cooling sequence in colour and magnitude rather than one localised feature, which breaks the distance-extinction degeneracy. ↩ ↩
9.
Thompson, I. B. et al. “The Cluster AgeS Experiment (CASE). VIII. Age and distance of the globular cluster 47 Tuc from detached eclipsing binaries.” Monthly Notices of the Royal Astronomical Society 492(3):4254–4267. 2020. Section Abstract: “We obtain Mp = 0.862 ± 0.005M⊙, Rp = 1.183 ± 0.003R⊙, Lp = 1.65 ± 0.05L⊙ for the primary and Ms = 0.827 ± 0.005M⊙, Rs = 1.004 ± 0.004R⊙, Ls = 1.14 ± 0.04L⊙ for the secondary. … We obtain 4.55±0.03 and 4.50±0.07 kpc, respectively – values compatible within 1σ with recent estimates based on Gaia DR2 parallaxes. By comparing the M − R diagram of the two binaries and the color-magnitude diagram of 47 Tuc to Dartmouth model isochrones we estimate the age of the cluster to be 12.0 ± 0.5 Gyr, and the helium abundance of the cluster to be Y≈0.25.” Masses to 0.6 per cent and radii to 0.3 per cent, from Kepler's third law and eclipse geometry, with no distance in the measurement. The distance that falls out afterwards agrees within one sigma with Baumgardt & Vasiliev's independent 4.521 ± 0.031 kpc. A follow-up the reader should know about: Ying et al. 2025 (ApJ 987: 52) ran the same twenty-one-parameter Monte Carlo isochrone grid twice, once against the colour-magnitude diagram and once against the eclipsing binaries alone. The distance-modulus and reddening columns of the binary rows read 'NA'. For 47 Tuc it obtained 11.76–11.89 Gyr from the diagram against 11.36 ± 0.81 Gyr from the binaries, and for NGC 6362 it obtained 11.61 ± 0.98 against 11.83 ± 0.52. A 2020 paper in a series from the Cluster AgeS Experiment that dates the globular cluster 47 Tucanae, a tight ball of old stars, from two eclipsing binaries, pairs of stars whose eclipses reveal the mass and size of each star. The method uses no distance, so it is an independent check on the cluster's age. The article sets its 12.0 billion year age beside the turnoff age and the white dwarf cooling age for the same cluster. Mirror: arXiv ↩ ↩ ↩
10.
Miglio, A. et al. “Detection of solar-like oscillations in relics of the Milky Way: asteroseismology of K giants in M4 using data from the NASA K2 mission.” Monthly Notices of the Royal Astronomical Society 461:760. 2016. A research paper that detected solar-like oscillations, rhythmic pulsations of the star's surface caused by sound waves inside it, in eight red giant stars in the globular cluster M4. It used the mass each star's oscillations give, and the article uses it because that mass agrees with the masses from the stellar models and from the eclipsing binaries.
1. 10a.
  Section §3.1.1, page 4: The isochrone fit to M4's colour-magnitude diagram with BaSTI models gives. “an age of 13 Gyr and a M_ISO,RGB = 0.84 M⊙”. Extrapolating the measured masses of M4's turnoff eclipsing binaries (Kaluzny et al. 2013) up the giant branch gives M_EB,RGB = 0.85 M☉, and the asteroseismic mass, under all four combinations of seismic and non-seismic constraints in their Table 2, is 0.84 M☉. Section abstract: “for RGB stars we find excellent agreement with the expected masses from isochrone fitting, and with a distance modulus derived using independent methods. The offset with respect to independent masses is lower, or comparable with, the uncertainties on the average RGB mass (4 - 10%).” Three routes give one number: stellar-model isochrones, Newtonian orbital mechanics, and the frequencies of sound waves inside the stars. This paper also fixes the metal-poor scaling-relation bias reported by Epstein et al. 2014. ↩ ↩
11.
Cayrel, R. et al. “Measurement of stellar age from uranium decay.” Nature 409:691–692. Feb 15, 2001. “²³⁸U (half-life 4.5 Gyr) is in principle a more precise age indicator, but even its strongest spectral line, from singly ionized uranium at a wavelength of 385.957 nm, has previously not been detected in stars. Here we report a measurement of this line in the very metal-poor star CS31082-001, a star which is strongly overabundant in its heavy elements. The derived uranium abundance, log(U/H) = −13.7 ± 0.14 ± 0.12 yields an age of 12.5 ± 3 Gyr, though this is still model dependent.” Its log g is 1.5 ± 0.2, and its abundances are log ε(Fe) = 4.60 ± 0.06, log ε(Th) = −0.96 ± 0.03 and log ε(U) = −1.70 ± 0.10. Osmium and iridium are about 1/9 of solar. The team detected fourteen Th II lines, used eleven, and ten of those. “appear to be first detections.” Their own summary: “Any age between 11.1 and 13.9 Gyr is compatible with the various determinations associated with their error bars. We consider the median value 12.5 Gyr as our best present estimate for the age of CS31082-001, with a conservative standard error of 3 Gyr.” Why the uranium-to-thorium ratio beats the alternatives, in their words: “the initial production ratio (U/Th)₀ is in principle much less affected by theoretical uncertainties than any of the individual (U/r)₀ because of the proximity of the two elements in their mass numbers.” Their closing concession: “The accuracy of this uranium dating technique is at present limited by incomplete knowledge of a few critical physical data, in particular oscillator strengths and production ratios of the elements produced by the r-process.” A short letter, published 15 February 2001, that reports the first measurement of uranium in a very metal-poor halo star, CS31082-001, and uses the ratio of uranium to thorium to date it. Its star parameters are V = 11.7 and T_eff = 4,825 K, and the star is very poor in iron, about 1/800 of the solar amount. The team's best age is 12.5 billion years, with a conservative standard error of 3 billion years, and they say the method is still model dependent. The article uses it as one independent clock for the age of an old halo star. ↩ ↩
1. 11a.
  Page 2: Preprint page, right-hand column. The decay law is ρ = ρ₀ exp(−ln2 × Δt/t₁ᐟ₂). The three working formulae are printed as equations (1), (2) and (3), with coefficients 46.7, 14.8 and 21.8 respectively, with Δt in Gyr and r a stable third-peak r-process element (osmium or iridium). Half-lives as stated there are ²³²Th 14.05 Gyr and ²³⁸U 4.468 Gyr. Their own gloss on the steepness: “an error of 0.1 dex in a Th abundance propagates as a time equal to the age of the Solar System, neglecting other sources of error.” ↩
12.
Faulkner, Danny R. “An Evaluation of Astronomical Young-Age Determination Methods 2: Solar, Stellar, Galactic, and Extragalactic.” Answers Research Journal 12:329–349. 2019. The second of two parts that evaluate astronomical arguments creationists have used for a young universe, the first part being on the solar system. The author evaluates 23 arguments, recommends dropping some of them, and calls others strong. The article uses it as the most useful creationist audit of these arguments, and because it has no heading for main-sequence turnoff ages, white dwarf cooling ages or stellar radioactive dating.
1. 12a.
  Section Fast Burn Rate of Hot Stars: “Unfortunately, Slusher relied upon some out-of-date references. … No astronomer today thinks that O and B type stars are primordial. Rather, astronomers generally think that O and B stars formed very recently (they are classed as extreme population I stars). Faulkner (2011) and Lisle (2012) have discussed briefly O and B type stars, but within the context of critiquing ideas of star formation. It is essential that any treatment of O and B stars include discussion of the possibility of star formation by naturalistic means. Therefore, the existence of stars with very short lifetimes do not in themselves prove that the world is young.” ↩
2. 12b.
  Section Break-up of Star Clusters: “Creationists have cited the break-up time of open star clusters as evidence of recent origin (Slusher, 1974; Slusher 1980a, 16; Slusher 1980b, 59). Rybka (1993, 102) did a simple calculation showing that the break-up time was on the order of millions of years, referring to the work of V.A. Ambartsumian. However, it appears that the intended work of Ambartsumian was about very loose groupings of stars called associations. … Consider the Pleiades, the best example of an open cluster. Converse and Stahler (2010) have modeled the dynamics of the Pleiades and found that it will dissipate in approximately one billion years. … However, the estimated evolutionary age of the Pleiades is about 125 million years, so there is no conflict between the evolutionary age and this maximum dynamic age. Most astronomers think that globular clusters are very old, forming shortly after the big bang. Gnedin and Ostriker (1997) found that globular clusters disrupt in time approximately that of the big bang age of the universe. … Once again, there does not appear to be a conflict between the dynamic ages of globular clusters and their evolutionary ages. For all types of clusters and associations, it appears that there is no conflict between their evolutionary ages and their dynamic ages. Perhaps some early work on this subject in the creation literature did not fully address the issues involved. … At any rate, the break-up time of star clusters does not appear to be a good evidence for recent creation.” ↩
3. 12c.
  Section Abstract: “I evaluate 23 astronomical arguments for recent origin for the sun, stars, and galaxies that biblical creationists have used. Some of the arguments are found to be wanting, and I thus recommend discontinuing their use. Other arguments for young age appear to be strong.” The paper has no heading for main-sequence turnoff ages, globular cluster isochrone ages, white dwarf cooling ages or stellar radioactive chronometry. Its companion paper on the solar system is Faulkner's "An Evaluation of Astronomical Young-Age Determination Methods 1: The Solar System," ARJ 12 (2019), at https://answersresearchjournal.org/astronomical-age-solar-system/. Faulkner keeps the spiral-arm persistence argument rather than retiring it. Anyone writing that Faulkner told creationists to drop the spiral argument would be misquoting the paper. Page 345: “the persistence of spiral structure probably is a good argument against the supposed long age of the universe. However, any discussion of spiral structure as an indication of recent origin ought to include at least a mention of the recognition of the winding dilemma by astronomers and their attempts to explain it.” ↩ ↩
4. 12d.
  Elsewhere in the same paper, the author accepts stellar structure theory generally: “We have a good understanding of the physical structure of stars. We also understand that stars derive their energy from thermonuclear reactions in their cores. … These processes take considerable time, but, given enough time, the gross properties of stars must change.” Section Solar neutrino problem: “Indeed, helioseismology confirms the standard solar model with temperature in the core sufficient to sustain nuclear reactions. One may complain that the standard solar model is based upon assumption of the evolutionary timescale, as Henry (2004) said in a response to his paper on solar oscillations (Faulkner and Samec 2004). However, this is not entirely accurate — for one can believe in a recent creation and the standard solar model. The standard solar model can sustain the sun for billions of years, but that does not mean that the sun necessarily is billions of years old. The standard solar model results in a very stable sun, more stable than a homogeneous, low temperature-core sun. Hence, the standard solar model could be interpreted as a design feature of special creation.” The last sentence states the young-earth position on stellar ages in miniature: the physics is granted, and what is denied is that the stars have been running for as long as they could have. ↩
5. 12e.
  The paragraph just before the conclusion says: “I discourage the use of those young-age indicators that do not survive scrutiny. … Since many of the arguments for recent creation that I have evaluated have turned out to be poor arguments, it might cause some to despair. However, we must be vigilant to ensure that our arguments for recent creation are good. Furthermore, there remain good arguments for recent origin.” Section Conclusion, page 348: “Though not directly related to the issue at hand, there is not yet a coherent creation model of astronomy. A creationary model of astronomy likely would facilitate development of new arguments for recent origin. One impediment to this development has been the question of how much of what we see in the universe today is attributable to the creation and how much may be due to physical processes since creation. Different views on this exist with the creation astronomy community. Hopeful development of a creation model of astronomy will clarify this situation.” ↩
13.
Faulkner, Danny R. “A Review of Stellar Remnants: Physics, Evolution, and Interpretation.” Creation Research Society Quarterly 44(2):76–84. 2007. A review article in a creationist journal, by a creationist astronomer at the University of South Carolina Lancaster. It surveys how white dwarfs and neutron stars are observed and modelled, then sets out possible creationist responses. Its closing section says that many questions are posed but few definitive answers are offered. The article uses it because it concedes that white dwarfs must cool as they age, and that their ages are far greater than a few thousand years.
1. 13a.
  Page 83: “Lacking an internal energy source, white dwarfs must tap their considerable store of internal heat and so they must cool as they age. From their temperature, we can estimate the ages of white dwarfs. These ages typically are far greater than a few thousands of years. Did God create white dwarfs with various ages? Many creationists would answer yes to both of these questions.” The next quotation is about neutron stars, on the same page. “Did God create 'fossil' stars? Creationary geologists would object adamantly if this were suggested in the realm of fossils of biological origin (i.e. fossils of plants and animals were planted in the ground by God). Is the astronomical realm so different?” Page 76: The methodological statement. “we must be very careful that we do not mistakenly 'throw the baby out with the bath water' by dismissing some of the conclusions that are based upon good observations and physics. As difficult as it may be, we must separate the evolutionary speculations from the well-established ideas.” Page 84: The closing remarks, and then the statement of the theological cost. “We have posed many questions here, but offered few definitive answers. The field of creation stellar astronomy is still in its infancy. The author has few answers for many of these questions at this time, and he hopes that this brief discussion can lead to further discussion and research.” “Many creationists view the aging and death processes of stars, as well as stellar remnants, such as black holes, as part of the curse. … If so, then when did these aging and dying processes occur? If after the Fall, then how did the light get here? If before the Fall, then is that consistent with a creation that was declared 'very good?'” ↩ ↩
2. 13b.
  Page 83: The whole paragraph, of which the article quotes the middle. “One extreme approach would be to note the vast time required to accomplish these processes and simply deny that any of these have happened. This attitude would obligate us to provide alternate explanations for these various objects. … The other extreme would be to embrace much of these explanations, but on our terms. For instance, some creationists have proposed that there was a time of rapid change sometime in the past. The accelerated nuclear decay proposed by the RATE project offers the possibility of rapid stellar development early in the universe. The Humphreys white hole cosmology allows for vast periods of time to have elapsed elsewhere in the universe while only a few days passed on earth during the Creation Week. The Humphreys cosmology might allow for much of the stellar development briefly described here. A very small minority of recent creationists believe that the Creation Week refers to the creation of the earth and its biosphere rather than the entire universe. This view would allow the acceptance of virtually all of evolutionary stellar astronomy.” ↩
14.
Samec, Ronald G. “The coolest white dwarf: older than the age of the universe?” Journal of Creation 29(2):8–9. 2015. A short creationist Perspectives piece by Ronald Samec on a 2014 report of the coolest known white dwarf, the dead core of a star, which is the companion of a pulsar (PSR J2222-0137). He accepts the standard cooling physics without dispute and answers the age problem with time dilation rather than by disputing the physics. The article uses it as the clearest example of a creationist who says the hard part out loud.
1. 14a.
  Page 8: The subject of the piece, Kaplan et al., 'A 1.05 M☉ Companion to PSR J2222-0137: The Coolest Known White Dwarf?', ApJ 789(2): 119 (2014). He states the cooling physics without dissent: “the star steadily loses its outer gasses and ends its life as a ball of compact degenerate electron gas or a White Dwarf (WD). It may begin this stage with a very high temperature, say 50,000 K and it steadily cools over time following a well-known cooling rate.” Page 9: The passage quoted in the article: “WDs are essentially 'fossil stars'. They have used up their fusion fuels and have lost their atmosphere and are left with degenerate electrons, slowly cooling over time. I believe that God did not create 'dead' stars. On the other hand, if time dilation has operated, then the stars actually existed as normal stars and they have lived out their lives and ended as WDs. The information is scientifically accurate and no false history is conveyed by their appearance.” His rejection of a mature creation on the same page: “However, the idea that God created WDs is analogous to the idea that He created fossils in the ground, so they never actually lived and died. Thus they are a false history and any information they could lend to such creation topics as the Flood history and their historical appearance in the fossil record are moot.” His verdict on the cosmologies of his own side, on the same page: “The 13.8 Ga scenario is obviously wrong even in creation time dilation based scenarios. Creationists have a real job sorting it out. One of the questions that need answering is: 'Are time dilation theories warranted?' This is still an open question.” Two of his own claims in the same piece should be flagged. His headline '33.8 Ga' figure is the pulsar's spin-down age, a well-known upper-limit estimator rather than a real age. His summary that the white dwarf is 'probably older than the Milky Way' misstates the argument of the paper he is discussing. Neither affects the passages quoted here. ↩ ↩
15.
Lisle, Jason. “Stellar Astronomy: Part 4 — Of Dwarfs and Supergiants.” Biblical Science Institute. Mar 22, 2024. Part 4 of Lisle's series on stellar astronomy. It grants that the masses of binary stars can be measured from Newton's laws, then argues that the ten-billion-year fuel estimate assumes a starting composition that God need not have given. The article uses it for that answer to the main-sequence age argument.
1. 15a.
  “Most stars are part of a binary system in which two or more stars orbit their common center of mass. By measuring their orbital period and their distance from each other, we can compute the mass of the stars from Newton's laws.” “Based on the size of the sun and its energy output, it would theoretically take ten billion years for it to convert all its core hydrogen into helium. However, this assumes that its core composition was initially identical to its surface composition. If the star began its existence with very little hydrogen in the core but lots of helium, it would take much less time to fuse the remaining hydrogen. We don't know that God created each and every star with a 'full tank of gas.' Furthermore, how do we know that God didn't create some stars with only helium in their core to begin with, in which case they would begin their existence already off the main sequence?” In footnote 4, Lisle writes: “Many astronomers would point to cluster turn-off points as evidence in favor of the standard model of stellar evolution. … Since early type stars exhaust their fuel more quickly than late type stars, they would indeed leave the main sequence sooner. Thus, secular astronomers use this as a tool to estimate the age of the cluster. However, even clusters that are assumed to be very old often contain a handful of blue main sequence stars known as 'blue stragglers,' which is contrary to the predictions of the model. Secular astronomers suppose that such stragglers formed by stellar mergers and are therefore recent, but there is no way to test this.” In the surveyed young-earth literature this footnote is the only place where the turnoff method is stated correctly and then answered. The body text first grants the mass-measuring apparatus, and argues that the main sequence is a mass sequence rather than a time sequence, which is true and is not what the turnoff argument claims. The body text and the footnote are in tension, and the 2008 time-dilation move has disappeared from this treatment. ↩ ↩ ↩
16.
Tolstoy, Eline et al. “Star Formation Histories, Abundances and Kinematics of Dwarf Galaxies in the Local Group.” Annual Review of Astronomy and Astrophysics 47:371–425. 2009. A 2009 review of the star-formation histories of dwarf galaxies in the Local Group, the cluster of galaxies that includes the Milky Way, by Tolstoy, Hill and Tosi. It explains that the ratio of alpha elements, which are built mainly by massive stars, to iron falls once the slower iron-making supernovae start to contribute, so the drop works as a clock for a galaxy's early history. The article uses it, with a second source, for the claim that heavy elements record a sequence rather than a snapshot. Mirror: arXiv
1. 16a.
  Section 4.1.1, alpha elements, pages 28–29: The passage on the alpha-to-iron ratio, [α/Fe], which is the ratio the review uses to trace the star-formation timescale. SNII are massive stars, and SNIa are intermediate-mass binary systems with mass transfer: “The ratio of α-elements to iron, [α/Fe], is commonly used to trace the star-formation timescale in a system, because it is sensitive to the ratio of SNII (massive stars) to SNIa (intermediate mass binary systems with mass transfer) that have occurred in the past. SNIa have a longer time scale than SNII and as soon as they start to contribute they dominate the iron enrichment and [α/Fe] inevitably decreases. After that, no SFH can ever again result in enhanced [α/Fe], unless coupled with galactic winds removing only the SNIa ejecta and not that of SNII. This is seen as a 'knee' in a plot of [Fe/H] vs. [α/Fe]. … The knee position indicates the metal-enrichment achieved by a system at the time SNIa start to contribute to the chemical evolution. This is between 10⁸ and 10⁹ yrs after the first star formation episode.” The measured positions of the knee: “At present the available data only cover the knee with sufficient statistics to quantify the position in the Scl dSph, a system which stopped forming stars 10 Gyr ago, and the knee occurs at [Fe/H]≈ −1.8. … This means that the metal-poor population has formed before any SNIa enrichment took place, which means on a timescale shorter than 1 Gyr.” “The Sgr dSph has enhanced [α/Fe] up to [Fe/H] ≈ −1.0, which is significantly more metal-rich than the position of the knee in the Scl dSph. This is consistent with what we know of the SFH of Sgr, which has steadily formed stars over a period of 8−10 Gyrs, and only stopped forming stars about 2−3 Gyr ago.” “The knee is constrained to be below [Fe/H] < −1.5 [in Fornax]. … it appears that the position of the knee correlates with the total luminosity of the galaxy, and the mean metallicity of the galaxy.” Page 29: The second, non-obvious result, which the article uses: “The abundance ratios observed in all dSphs for stars on the metal-poor side of the knee, tend to be indistinguishable from those in the MW halo. From this small sample it seems that the first billion years of chemical enrichment gave rise to similar enrichment patterns in small dwarf galaxies and in the MW halo.” ↩ ↩
17.
Nidever, David L. et al. “The Lazy Giants: APOGEE Abundances Reveal Low Star Formation Efficiencies in the Magellanic Clouds.” Astrophysical Journal 895:88. 2020. Section abstract: “At the metal-poor end, we capture the increase of [α/Fe] with decreasing [Fe/H], and constrain the 'α-knee' to [Fe/H] ≲ −2.2 in both MCs, implying a low star-formation efficiency of ∼0.01 Gyr⁻¹. The MC knees are more metal poor than those of less massive Milky Way (MW) dwarf galaxies such as Fornax, Sculptor, or Sagittarius. One possible interpretation is that the MCs formed in a lower-density environment than the MW.” Section §7, page 16: “We find that the MCs α-element abundance trendlines (and their knees) are more metal poor than Fornax, Sculptor, and Sgr, even though the MCs are a factor of ∼100–1000 more massive in stellar mass”. The correlation across the lower-luminosity dwarfs, as the paper states it: “has a Pearson correlation coefficient of −0.705, which indicates a strong anti-correlation. The MCs clearly fall off this correlation.” Their reading is that the knee tracks star-formation efficiency, for which mass is only a proxy, and the proxy fails for these two. Section §1, page 3: The authors also caution that. “the definition of such a feature is often ill-defined”. The knee positions collected in the article should be read as a spread of order 2 dex rather than a precision ladder, because the values come from different alpha elements, instruments and definitions. A survey of about 3,600 red giant stars in the Large and Small Magellanic Clouds, which are nearby galaxies, that tracks how the ratio of α-elements to iron changes with metallicity, using an order of magnitude more red giants than previous work. It finds the knee, where that ratio starts to fall, at lower metallicity in the Clouds than in the Milky Way's dwarf galaxies, and reads this as low star-formation efficiency rather than a matter of mass. The article uses it because the knee does not follow galaxy mass. ↩ ↩
18.
Rafelski, Marc et al. “Metallicity Evolution of Damped Lyα Systems out to z ∼ 5.” Astrophysical Journal 755:89. 2012. Section abstract, page 1: “We present chemical abundance measurements for 47 damped Ly-α systems (DLAs), 30 at z > 4 … we combine our new metallicity measurements with 195 from previous surveys to determine the evolution of the cosmic metallicity of neutral gas. We find the metallicity of DLAs decreases with increasing redshift, improving the significance of the trend and extending it to higher redshifts, with a linear fit of −0.22 ± 0.03 dex per unit redshift from z = 0.09 − 5.06. The metallicity 'floor' of ≈ 1/600 solar continues out to z ∼ 5, despite our sensitivity for finding DLAs with much lower metallicities. However, this floor is not statistically different from a steep tail to the distribution. We also find that the intrinsic scatter of metallicity among DLAs of ∼ 0.5 dex continues out to z ∼ 5.” Two honesty points follow from the paper's own wording: the 0.5 dex scatter means this is a trend in the mean, not a track that any individual cloud follows, and the 'floor' should not be quoted as a hard claim. The authors also note a connection worth having: “the metallicity distribution and the α/Fe ratios of z > 2 DLAs are consistent with being drawn from the same parent population with those of halo stars”. In plain terms, absorption lines in quasar spectra and stellar spectra in our own Galactic halo are landing on the same abundance patterns. A measurement paper that determined the chemical abundances, the metal content, of 47 damped Lyman-alpha systems, which are clouds of gas that absorb the light of distant quasars, and combined them with 195 measurements from earlier surveys. It finds that metallicity falls with redshift. The article uses it for the trend in the metallicity of neutral gas. ↩ ↩
19.
Péroux, Céline, and J. Christopher Howk. “The Cosmic Baryon and Metal Cycles.” Annual Review of Astronomy and Astrophysics 58:363–406. 2020. A review of the cycle of gas, stars and heavy elements across cosmic time, which tabulates the metallicity of neutral hydrogen gas absorbers in bins of redshift. The article uses it for the factor-of-ten rise in the metallicity of neutral gas from redshift 5 to today, and for the bumpy middle of that rise.
1. 19a.
  Table 1: The hydrogen-weighted mean metallicity of neutral gas absorbers by redshift bin: −0.16 (+0.14/−0.11) at mean z = 0.55 from 18 absorbers; −0.40 ± 0.09 at z = 1.64 from 40; −0.56 (+0.10/−0.11) at z = 2.26 from 62; −0.41 (+0.21/−0.30) at z = 2.70 from 57; −0.79 (+0.11/−0.12) at z = 3.21 from 38; −0.82 (+0.14/−0.17) at z = 3.95 from 21; and −1.03 (+0.12/−0.20) at z = 4.54 from 14. The z = 2.70 bin sits above the smooth trend and carries the largest error bars in the table, so the trend is a factor-of-ten rise with a bumpy middle, not a straight line. Section §3.4: “The H i-weighted mean metallicity of neutral gas in the Universe increases slowly from high to low redshift, rising by an order of magnitude from z ≈ 5 to 0.” Section §4.1: “H i-weighted DTG increases by ∼ 1 dex from z ≈ 5 to today, in parallel to the increase in mean metallicity.” ↩ ↩
20.
Lisle, Jason. “The James Webb Space Telescope.” Biblical Science Institute. Jan 21, 2022. A pre-launch article by Jason Lisle, from January 2022, that predicted what the James Webb telescope would find: fully formed galaxies at unprecedented distances, and heavy elements in them rather than only the first stars. The article uses it as the dated prediction that the telescope's results are tested against.
1. 20a.
  “What do creationists expect to find? I cannot speak for all creation scientists. But I personally expect that the JWST will find quite a different situation. Rather than galaxies just starting to form, I expect to see fully-formed (fully-designed) galaxies at unprecedented distances. This will force secular astronomers to adjust their estimates of when the earliest galaxies formed, pushing them much closer to the supposed big bang. We might see headlines like 'Webb discovers that galaxies formed much earlier than previously thought.'” The second prediction, in the next paragraph: “Furthermore, I expect the signal of some heavy elements in these galaxies. That is, I don't expect to see evidence of genuine Population III stars – those with no heavy elements at all. … Note that some secularists might try to 'move the goalpost' by redefining Population III stars as having a very low fraction of heavy elements. But I predict it won't be zero as required by the big bang model.” The second prediction is now under pressure. Maiolino et al. (arXiv:2603.20362v4, March 2026, preprint) report a He II λ1640 emitter three kiloparsecs from GN-z11 at z = 10.6 with no metal lines detected, and argue "Population III stars are the most plausible explanation." A preprint is not a settled detection, and a non-detection of metals is an upper limit, but the direction of travel is against Lisle's prediction. The January 2022 prediction and Lisle's September 2022 scoring of it ("Creation Cosmology Confirmed!") are two separate documents deserving separate verdicts. The September piece quotes the Labbé v1 preprint by its v1 title and abstract, and its claim that these galaxies are "more than a thousand times more massive than the secular predictions" misreads Labbé's figure, which is a ratio of stellar mass densities rather than of individual galaxy masses. ↩ ↩
21.
Arrabal Haro, Pablo et al. “Confirmation and refutation of very luminous galaxies in the early universe.” Nature 622:707–711. 2023. Table 1, pages 6–7: Preprint page numbers. Table 1 gives CEERS-93316 at z = 4.912 ± 0.001, Maisie's Galaxy at z = 11.416 ± 0.005 and CEERS2-588 at z = 11.043 ± 0.003. Section Abstract: “Here we present JWST spectroscopy that confirms redshifts for two very luminous galaxies with z > 11, but also demonstrates that another candidate with suggested z ≈ 16 instead has z = 4.9, with an unusual combination of nebular line emission and dust reddening that mimics the colors expected for much more distant objects.” Section The example of CEERS-93316: The paper's moral, quoted in the article. “Exceptional candidates for galaxies at extreme redshifts require spectroscopic confirmation”. A spectroscopy paper that checked the most extreme galaxy candidates from the James Webb Space Telescope. It confirmed two galaxies above redshift 11, and it showed that a candidate suggested at redshift about 16 was really at redshift 4.9, where dust and unusual glowing gas made it look much farther away. The article uses it because the most extreme candidate in the early-galaxy story turned out to be at that lower redshift. ↩ ↩
22.
Lisle, Jason. “Stellar Evolution and Millions of Years.” Answers in Genesis Feedback. Feb 8, 2008. A reply by Jason Lisle, dated 8 February 2008 on the Answers in Genesis website, to a reader's question: how can white dwarfs lie below the main sequence if stars are only 6,000 years old? Lisle gives several answers, and one of them is relativistic time dilation, which in the models of Russ Humphreys and John Hartnett lets stars age for millions of years while only thousands of years pass on Earth. The article quotes that time-dilation answer as the young-earth route to long stellar ages.
1. 22a.
  The fifth of Lisle's moves in the reply, and the one the article quotes: “In addition to all the above caveats, we should also consider the various creation cosmogony scenarios involving relativistic time dilation, such as the ones promoted by Russ Humphreys and John Hartnett. In these models, stars can literally age millions of years while only thousands of years elapse on Earth — in which case the secular conjectures regarding stellar aging may be close to the right idea. Such time dilation is the result of Einstein's physics — the well-tested General Theory of Relativity.” The reader's question is worth recording, because it is precisely this article's question. The letter asks: “How do you explain the presence of white dwarfs below the Main Sequence when stars that have masses of 0.5 to 1.4 solar masses should all be on the Main Sequence if only 6,000 years old? … A claim that God created them as degenerate white dwarfs cannot be accepted on scientific grounds, since the H-R diagram clearly shows a progression from Main Sequence to red giants and finally white dwarfs.” Lisle's other moves in the same reply: stellar evolution "has never been observed or directly tested — nor could it be, if it really takes millions of years"; that not all progressions are time-progressions, with the genuine historical precedent that the main sequence itself was once wrongly read as a time progression; a mature-creation move, "God created the first trees already bearing fruit"; and a rapid-collapse move resting on the red Sirius story, "there is no empirical way to show that this must take millions of years." Faulkner later rejected the Sirius argument with white dwarf cooling physics, in ARJ 12 (2019): 336: "newly produced white dwarfs are extremely hot (typically in excess of 100,000 K), and they exponentially cool. This cooling process is well-understood physically. Sirius B is far too cool (25,000 K) to be a white dwarf that is merely 2,000 years old." Faulkner adds a methodological rebuke on the same page: "it tacitly accepts an idea of stellar evolution, something that creationists generally are loathe to do. We ought not to selectively choose which parts of stellar evolution theory we wish to accept." ↩
23.
Faulkner, Danny R. “Solving the Light Travel Time Problem.” Answers in Depth. Feb 23, 2021. A popular web article, published 23 February 2021, that surveys the creationist answers to the light travel time problem, meaning the question of how distant starlight can have reached Earth in a few thousand years. It covers mature creation, accelerated light speed, the white hole cosmology, Carmeli's cosmology, the anisotropic synchrony convention and the dasha proposal, and it reports that Humphreys has abandoned the white hole model. The article uses it for that report, and for two passages in which the author rejects mature creation as deceptive.
1. 23a.
  “For instance, I have more than four decades of experience observing eclipsing binary stars. … However, if the light of these stars were created in transit so that the light never left the stars, then we would see evidence of events (eclipses, for example) that would have never happened. This is deceptive.” The consequence the author draws: “Since in the mature creation solution, the light from distant objects never left those objects, none of those distant objects need exist at all. Consequently, in the mature creation solution, much of the universe could be an illusion.” ↩
2. 23b.
  Faulkner reports that Humphreys abandoned the white hole cosmology. A fuller survey of all eight young-earth answers to distant starlight, including Hartnett's five dated retractions of the Carmelian cosmology (2015-02-07, 2019-04-19, 2021-03-11, 2021-06-23 and 2024-12-06), is summarised in the article Young-universe cosmologies. ↩
3. 23c.
  The mature-creation position is stated at its strongest first. “Creationists who appeal to mature creation to solve the light travel time problem reason that if most astronomical bodies were not visible at least by the end of the creation week, then the astronomical bodies could not fulfill their functions given in the day-four account (Genesis 1:14-19). … Adam and Eve were created as mature adults. … Likewise, the plants in the Garden of Eden (and elsewhere) would have looked mature too.” “Our bodies have subtle evidence of things that occurred during our growth and maturing processes. … Did Adam and Eve bear in their bodies any evidence of a childhood they never experienced? Most supporters of mature creation would opine that Adam and Eve did not. This amounts to an affirmation that while Adam and Eve were created mature, they did not bear any detailed evidence of processes that never happened. Such evidence would seem to be deceptive, which would violate the character of God.” ↩
24.
Eker, Z. et al. “Interrelated main-sequence mass-luminosity, mass-radius and mass-effective temperature relations.” Monthly Notices of the Royal Astronomical Society 479:5491. 2018. Section Abstract: “Absolute parameters of 509 main-sequence stars selected from the components of detached-eclipsing spectroscopic binaries in the Solar neighbourhood are used to study mass-luminosity, mass-radius and mass-effective temperature relations.” A study of 509 main-sequence stars, the ordinary stars like the Sun, taken from detached eclipsing binaries (pairs of stars that pass in front of each other, so their masses and radii can be read off). One power law does not fit the whole range, so the authors use six mass segments. Their fit for the 0.72 to 1.05 solar-mass range, made to 53 eclipsing-binary components, is log L = 5.743 log M − 0.007. With M = 1.000 it returns L = 0.984 L☉, which reproduces the Sun to 1.6 per cent without the Sun being in the fit. The article uses it as a reason stellar models are not free to be wrong. ↩
1. 24a.
  Table 4, page 15: Preprint page. The fitted exponents and scatters, by mass range in solar masses, are: 0.179–0.45, α = 2.028, σ = 0.076 dex, N = 22; 0.45–0.72, α = 4.572, σ = 0.109, N = 35; 0.72–1.05, α = 5.743, σ = 0.129, N = 53; 1.05–2.40, α = 4.329, σ = 0.140, N = 275; 2.4–7, α = 3.967, σ = 0.165, N = 80; 7–31, α = 2.865, σ = 0.152, N = 44. The textbook rule L ∝ M³·⁵ is a rough average. The real exponent runs from 2.0 to 5.7, and the scatter is 0.08 to 0.17 dex, a factor of 1.2 to 1.5. ↩
25.
Ying, Jundan Nie et al. “The Absolute Age of M92.” Astronomical Journal 166(1):18. 2023. A 2023 paper that dates the globular cluster M92 by building 20,000 sets of theoretical stellar models (isochrones) with twenty input parameters varied across their plausible ranges, a Monte Carlo test. It finds an age of 13.80 ± 0.75 billion years. The article uses it for the age of M92 and for an error budget dominated by distance. Mirror: arXiv
1. 25a.
  Section Abstract: “We find the age of M92 to be 13.80 ± 0.75 Gyr. The 5.4% error in the absolute age is dominated by the uncertainty in the distance to M92 (∼ 80% of the error budget); of the remaining parameters, only the total metallicity, α element abundance, and treatment of helium diffusion contribute significantly to the total error.” Section 6, page 12: Their conclusion: “The fact that the distance to M92, and not stellar physics, dominates the uncertainty points to the importance of precise and accurate distance measurements for further improvements in absolute age measurements.” Two things follow. Convective overshoot does not make the list of contributors above one per cent: only distance modulus, metallicity, α-enhancement and helium diffusion do. The 13.80 ± 0.75 Gyr figure is uncomfortably close to the Planck age of the universe, so it should be reported with its error bar and alongside VandenBerg et al. 2013's 12.75 Gyr for the same cluster. Different groups get 12.75 to 13.8 Gyr, and the article presents neither as settled. ↩
2. 25b.
  Table 1, page 3: The twenty varied inputs and their assumed ranges: [Fe/H] normal at −2.30 ± 0.10; primordial helium uniform over 0.244–0.249; [α/Fe] normal at 0.40 ± 0.1; mixing length uniform over 1.0–2.5; heavy-element diffusion uniform over 0.5–1.3; helium diffusion uniform over 0.5–1.3; surface boundary condition trinary across Eddington, Krishna Swamy and Hauschildt; low-temperature opacities uniform over 0.7–1.3; high-temperature opacities normal at 1.0 ± 0.03; plasma neutrino losses normal at 1.0 ± 0.05; conductive opacities normal at 1.0 ± 0.20; convective envelope overshoot uniform over 0–0.2 pressure scale heights; convective core overshoot uniform over 0–0.2; and seven nuclear reaction rates with their measured uncertainties, including ¹⁴N(p,γ)¹⁵O at 3.32 ± 0.11 keV-barns. The multiplicative parameters vary about their standard values. Notice the sizes: low-temperature opacities are allowed to be wrong by 30 per cent, conductive opacities by 20 per cent, and the mixing length is allowed to run over a range spanning everything anyone has ever proposed. Twenty thousand isochrone sets were built by Monte Carlo over that space, and the answer moved by 5.4 per cent. ↩
26.
Chaboyer, Brian et al. “A Lower Limit on the Age of the Universe.” Science 271:957–961. 1996. A Monte Carlo study of the uncertainties in the models used to date the oldest globular clusters. It ran 1,000 sets of stellar isochrones with the inputs varied within their uncertainty ranges, and fitted the 18 oldest clusters. Its one-sided 95 per cent lower bound on their mean age is about 12.1 billion years. The article uses it as the 1996 argument that cluster ages pointed toward a cosmological constant, two years before the accelerating-universe measurement.
1. 26a.
  Section Conclusions, page 13: Preprint page. “we find that the age of globular clusters in our galaxy is inconsistent with a flat, matter dominated universe unless h < 0.54, and for a nearly empty, matter dominated universe unless h < 0.80. If the value of h is definitely determined to be larger than either of these values, some modification, such as the addition of a cosmological constant, would seem to be required.” Published in 1996, two years before the supernova results that established cosmic acceleration. The statement of the conflict as it then stood, from page 1 of the preprint: Page 1: “if the age of the oldest globular clusters in our galaxy is really 16 Gyr, as various best fits estimates recently suggest, then this will be inconsistent with the Hubble age if h > .6. Consistency with the flat matter dominated universe estimate is impossible if h > .4.” ↩
2. 26b.
  Section Conclusions, page 13: Preprint page. “The explicit effect of the largest single common observational uncertainty, that in Mv(RR), increases the net width of the distribution by approximately ±0.6 Gyr (i.e. ≈ ±5%). Note that simply varying Mv(RR) over its full 2σ range, keeping all other parameters fixed, would produce a ±16% change in GC ages estimates. For comparison, the next most significant input parameter uncertainties in this same sense are [α/Fe] (±7% effect), mixing length (±5% effect), and diffusion, ¹⁴N p reaction rate, and primordial Helium abundance, each of which would affect age estimates at the ±3% level if allowed to vary over its entire range, keeping all other parameters fixed.” Chaboyer et al. 1998 record that this run built. “over 4 million stellar models”. The Monte Carlo behind these numbers is described in the abstract: Section Abstract: “Utilizing estimates of the uncertainty range (and distribution) in the input parameters of stellar evolution codes we produced 1000 Monte Carlo realizations of stellar isochrones, with which we could fit the ages of the 18 oldest globular clusters.” ↩
3. 26c.
  Page 13: Preprint page. “In particular, at the lower age limits the rise is steeper than Gaussian, reflecting the fact that essentially all models give an age in excess of 10 Gyr.” Their headline result on the same page: “at the one sided 95% confidence level … a lower limit of approximately 12.1 Gyr can be placed on the mean value of these 18 Globular clusters. (The symmetric 95% range of ages about the mean value of 14.56 Gyr is 11.6-18.1 Gyr.)” ↩
27.
Christensen-Dalsgaard, Jørgen. “Helioseismology.” Reviews of Modern Physics 74:1073. 2002. Page 31: Preprint page. “The values tend to be in the range 0.24 - 0.25. … It is striking, in all these cases, that the values obtained are substantially below the initial value Y0 = 0.27 - 0.28 required to calibrate the models to have the present solar luminosity. This confirms the importance of settling of helium which reduces the envelope helium abundance during evolution.” A review of helioseismology, the study of the Sun's interior from the frequencies at which its surface oscillates. The helium abundance in the Sun's convection zone is measured from the equation-of-state signature in those frequencies, not assumed. The article uses it for the Sun's helium content, and so for its starting hydrogen fraction, about 0.71. ↩
28.
Kaluzny, J. et al. The cluster ages experiment (CASE). V. Analysis of three eclipsing binaries in the globular cluster M4. 2013. A research paper that weighed three pairs of eclipsing stars in the globular cluster M4, pairs that pass in front of each other as they orbit, and used their measured masses to date the cluster. It compares the stars' masses and sizes with two sets of theoretical isochrones (model sequences for stars of one age) and finds the most probable age of M4 to be 11.2 to 11.3 Gyr (billion years). The article uses it as the age that needs no distance.
1. 28a.
  “We find that the distance to M4 is 1.82±0.04 kpc — in good agreement with recent estimates based on entirely different methods. We compare the absolute parameters of V66 and V69 with two sets of theoretical isochrones in mass-radius and mass-luminosity diagrams, and for an assumed [Fe/H] = -1.20, [α/Fe] = 0.4, and Y = 0.25 we find the most probable age of M4 to be between 11.2 and 11.3 Gyr. CMD-fitting with the same parameters yields an age close to, or slightly in excess of, 12 Gyr.” The mass–radius route uses no distance at all; the distance falls out afterwards as a by-product, and agrees with the Gaia-era value. ↩
2. 28b.
  “The measured masses of the primary and secondary components (Mp and Ms) are 0.8035±0.0086 and 0.6050±0.0044 M⊙ for V65, 0.7842±0.0045 and 0.7443±0.0042 M⊙ for V66, and 0.7665±0.0053 and 0.7278±0/0048 M⊙ for V69.” The preprint prints one uncertainty as 0/0048, a typographical artefact for 0.0048. These are among the most precisely weighed stars outside the solar neighbourhood, and 0.60–0.80 M☉ is exactly the turnoff-mass regime the whole method depends on. The related NGC 6362 result, Kaluzny et al. 2015 (CASE VII, arXiv:1508.04894v1), gave masses of (0.8337±0.0063, 0.7947±0.0048) M☉ for V40 and (0.8215±0.0058, 0.7280±0.0047) M☉ for V41, and reported "the most probable age of V40 to be 11.7±0.2 Gyr, compatible with the age of the cluster derived from CMD fitting (12.5±0.5 Gyr)." It also said "V41 seems to be markedly younger than V40. If independently confirmed, this result will suggest that V41 belongs to the younger of the two stellar populations recently discovered in NGC 6362," which shows the multiple-population problem inside a single binary-pair analysis. The first CASE result, Thompson et al. 2001 on ω Centauri (AJ 121: 3089, arXiv:astro-ph/0012493), was much weaker: it gave age ranges of 9.1–16.7 and 12.9–18.5 Gyr, and its authors said "Major improvements in the accuracy in estimated age and distance can be made with better measurements of the masses of the components." The method's early precision was poor, which makes the later results more credible rather than suspiciously tidy. ↩
29.
Tremblay, Pier-Emmanuel et al. “Core crystallization and pile-up in the cooling sequence of evolving white dwarfs.” Nature 565(7738):202–205. 2019. A 2019 paper that counts the white dwarfs, the dead cores of Sun-like stars, within 100 parsecs of the Sun, using Gaia distances, and finds that they pile up at one brightness. The pile-up is where the cores crystallise, locking their carbon and oxygen ions into a solid while the star releases heat, which delays its cooling. The article quotes its statements that white dwarfs cool predictably, and that the pile-up is a direct observation of a transition predicted over fifty years earlier. Mirror: arXiv
1. 29a.
  Page 1: Page numbers in these notes are those of the arXiv preprint, not of Nature. The first sentence of the abstract: “White dwarfs are stellar embers depleted of nuclear energy sources that predictably cool over billions of years at the expense of their leaking thermal reservoir of ions.” Page 6: Their own statement of what remains model-dependent, in the final paragraph of the main text: “The descending branch of the empirical white dwarf luminosity function is heavily impacted by phase separation and quantum effects in Debye cooling, necessitating the understanding of these processes when relying on stellar remnants for age-dating stellar populations.” ↩
2. 29b.
  Page 1: The prediction, which the paper's abstract states: “It is predicted that a first-order phase transition occurs during white dwarf cooling, leading to the crystallisation of the non-degenerate carbon and oxygen ions in the core that releases a significant amount of latent heat and delays the cooling process by about one billion years.” The prediction is H. M. van Horn, 'Crystallization of White Dwarfs', Astrophysical Journal 151 (1968): 227–238, which the paper cites as its reference 3. Page 2: The first section, on the observational consequence: “The observational implication of this, predicted over fifty years ago, is an isolated crystallisation sequence in the colour versus absolute magnitude Hertzsprung-Russell (H-R) diagram, yet no direct observational evidence existed until now to characterise this event.” The sample: white dwarfs within 100 parsecs from the Gentile Fusillo et al. Gaia DR2 catalogue, of which 15,109 had photometry and astrometry good enough to derive temperature, surface gravity and mass, at a median parallax precision of 1.5 per cent. 1,309 have SDSS spectroscopy giving atmospheric composition (p. 3). Page 3: The feature: “A third separate 'transverse' sequence is visible at fainter absolute magnitudes. Unlike the aforementioned bifurcation, this transverse feature is inconsistent with a single-mass cooling track.” Alternatives ruled out on pp. 3–4: post-main-sequence mass loss and binary mergers cannot produce a tight mass-versus-temperature correlation; only about 8 per cent of the pile-up stars are magnetic above 2 MG; and cross-matching with Galex, 2MASS, WISE, Pan-STARRS and SDSS shows the stars are underluminous at all wavelengths, so they are genuinely high-mass rather than an atmospheric artefact. The name 'Q branch' comes from a follow-up paper by Cheng, Cummings & Ménard, not from this one. Page 5: The interpretation: “The crystallised sequence is not a cooling track but a mass-dependent pile-up across the H-R diagram resulting from the white dwarfs spending more time at this location as they release their latent heat.” ↩
3. 29c.
  Page 5: They built the luminosity function of white dwarfs with Gaia masses between 0.9 and 1.1 M☉ and compared it with population simulations under three treatments of crystallisation: “The case without latent heat release by crystallisation is clearly ruled out by the observations. When latent heat is included in the modelling, there is, comparatively, a substantial increase of the predicted number of white dwarfs in the range of luminosity of the observed peak. The Gaia luminosity function is best reproduced when ¹⁶O sedimentation is allowed to occur along with the release of latent heat.” Page 6: Their own stated limit: “The model bump is not perfectly modelled, as the observed feature is narrower and of higher magnitude”. than predicted. Their assumed model white dwarf was a ¹²C/¹⁶O core, 50/50 by mass and homogeneously distributed, with envelope fractions M_H/M_WD = 10⁻⁴ and M_He/M_WD = 10⁻², and a Coulomb coupling parameter at freezing of Γ = 175. ↩
4. 29d.
  Page 6: “We report direct evidence that a first-order phase transition really occurs in high-density Coulomb plasmas, a theory that cannot be tested in laboratories because of the extreme densities involved.” ↩
30.
Bedin, Luigi R. et al. “Reaching the End of the White Dwarf Cooling Sequence in NGC 6791.” Astrophysical Journal Letters 624:L45. 2005. Section Abstract, page 1: Preprint page. “The white dwarf luminosity function shows a well defined peak at mF606W = 27.4, and a bending to the blue in the color-magnitude diagram. If this peak corresponds to the end of the white dwarf cooling sequence the comparison with theoretical isochrones provides a cluster age estimate of ~2.4 Gyr, in sharp contrast with the age of 8-9 Gyr inferred from the main-sequence turn-off. If the end is at fainter magnitudes, the peak at mF606W = 27.4 is even more enigmatic. We discuss possible causes, none of them very convincing.” Section 3, page 5: “The fact that the TO and WD ages agree for M67 makes the disagreement in NGC 6791 even more puzzling.” A study of the faint white dwarfs in the cluster NGC 6791, whose main-sequence turnoff puts its age at 8 to 9 billion years. It found a peak in the white dwarf luminosity function, a count of white dwarfs by brightness. If the peak marks the end of cooling, it gives an age of about 2.4 billion years, and the authors found none of the possible causes for the peak very convincing. The article quotes it to show the two clocks in one cluster disagreeing. ↩
31.
Bedin, Luigi R. et al. “Reaching the End of the White Dwarf Cooling Sequence of NGC 6791.” Astrophysical Journal 678:1279. 2008. Section Abstract, page 1: Preprint page. “The brighter peak previously observed in the white dwarf luminosity function (WDLF) is now better delineated, and the second, fainter peak that we suggested earlier is now confirmed. … The discrepancy between the age from the WDs and that from the main sequence turnoff remains, and we have an additional puzzle in the second peak in the WDLF. Canonical WD models seem to fail — at least at ~25%-level — in reproducing the age of clusters of this metallicity.” Section 8.2, pages 23–24: Preprint pages. The candidate physics estimates are these. A revised ¹²C(α,γ)¹⁶O rate gives. “less than 1 Gyr”. Diffusion of ²²Ne in the liquid phase, citing Deloye and Bildsten 2002, gives. “~1 Gyr”. Separation of ²²Ne during crystallisation, citing Segretain 1996, gives ~0.4 Gyr. A second study of the same cluster, NGC 6791, which confirmed the two peaks in its white dwarf luminosity function and checked the cooling models against them. The authors admit that the standard white dwarf models fail, by about a quarter, to reproduce the cluster's age, and they list physical effects that could account for the gap. The article quotes that admission. ↩
32.
Salaris, M. et al. “²²Ne distillation and the cooling sequence of the old metal-rich open cluster NGC 6791.” Astronomy & Astrophysics 686:A153. 2024. A model paper on the white dwarfs of NGC 6791, an old, metal-rich open cluster, which adds the sinking of neon-22 to the cooling models. Its models give a white-dwarf luminosity function that matches the faint end of the observed one at ages consistent with the cluster's eclipsing binaries and its turnoff. The article uses it for the revision of the cooling-delay estimate.
1. 32a.
  Section Introduction, page 2: “the very recent model calculations by Salaris et al. (2022) and Bauer (2023) have shown that the cooling delay induced by neon diffusion is smaller than previously calculated, leading Bauer (2023) to question whether the latest generation of WD models would still be able to lead to WD ages consistent with the cluster age determined from the main sequence.” Section abstract, page 1: “We also show how our models produce luminosity functions (LFs) that can match the faint end of the observed WD LF in NGC 6791, for ages consistent with the range determined from a sample of cluster's eclipsing binary stars, and the main sequence turn-off. Without the inclusion of distillation the theoretical WD cooling sequences reach too faint magnitudes compared to the observations.” Section §1, page 2: The cluster's independent parameters from eclipsing binaries (Brogaard et al. 2011, 2012, 2021) are [Fe/H] = +0.29 ± 0.03 (random) ± 0.08 (systematic), age 8.3 ± 0.3 Gyr, E(B−V) = 0.14 ± 0.02, and (m−M)_V = 13.51 ± 0.06. The neon abundance in the white dwarf cores is about 3 per cent by mass. The intermediate step is García-Berro et al. 2010 (Nature 465: 194–196), "A white dwarf cooling age of 8 Gyr for NGC 6791 from physical separation processes," which raised the white dwarf age from 6.4 ± 0.2 to 8.0 ± 0.2 Gyr by including ²²Ne sedimentation with carbon-oxygen phase separation, as Campos et al. 2016 (arXiv:1512.03114, §1, preprint p. 2) summarise it. ↩
33.
Campos, F. et al. “A comparative analysis of the observed white dwarf cooling sequence from globular clusters.” Monthly Notices of the Royal Astronomical Society 456:3729. 2016. An independent reanalysis of the white dwarf cooling sequences of globular clusters, the dense star clusters that orbit the galaxy. Its ages for NGC 6397 and 47 Tuc are higher than earlier analyses of the same kind of data, and the authors say the disagreement raises questions about the method. The article uses it to show that the cooling clock has weaknesses that its own authors admit.
1. 33a.
  Page 1: Preprint page. “This effect alone, however, does not completely reproduce the observational data — the blue turn retains some mystery. In this context, we discuss several other potential problems in the models. These include possible partial mixing of H and He in the atmosphere of white dwarf stars, the lack of a good physical description of the collision-induced absorption process and uncertainties in the opacities at low temperatures.” Their reanalysis gives NGC 6397 an age of 12.93 (+0.37/−0.21) Gyr and 47 Tuc an age of 10.95 (+0.21/−0.15) Gyr, both higher than Hansen's values from the same class of data, a method-to-method disagreement of 1 to 1.5 Gyr which they say. “brought up questions about the reliability of ages determined through the white dwarf cooling sequence.” Section 4.1.2, page 11: Their own quoted model uncertainty is. “of the order of 0.50 Gyrs”. ↩
34.
Kilic, Mukremin et al. “The Ages of the Thin Disk, Thick Disk, and the Halo from Nearby White Dwarfs.” Astrophysical Journal 837:162. 2017. A study that dates the thin disk, the thick disk and the halo of the Milky Way from nearby white dwarfs, the cooled remnants of stars, using a sample within 40 parsecs of the Sun. It fits the white dwarfs' luminosity function (how many there are at each brightness) and gives ages for each population. The article uses it for the faint end of that sample, where the statistics are thin.
1. 34a.
  Section abstract, page 1: “Simultaneously fitting for both disk components, we derive ages of 6.8-7.0 Gyr for the thin disk and 8.7 ± 0.1 Gyr for the thick disk from the local 40 pc sample. … The faint end of the luminosity function for the halo white dwarfs is less constrained, resulting in an age estimate of 12.5 (+1.4 / -3.4) Gyr for the Galactic inner halo.” Section §3, page 5: On the thinness of the statistics: the 40 pc sample has only 84 objects fainter than bolometric magnitude 15, and even the deep proper-motion catalogue has 311. Section §4, page 8: “this luminosity function is not deep enough to show a cutoff at the faint end, and is consistent with an arbitrarily large halo age of 15.0 Gyr … within 2 sigma.” The halo number is genuinely soft on the upper side; the lower bound is not, being at or above 8.1 Gyr at three sigma. Section §3, page 4: The largest systematic in older disk work is stated bluntly: fitting the 40 pc sample as thin disk only gives 8.7 ± 0.1 Gyr, and the authors write of that result: “this age measurement is clearly wrong since it ignores the contribution of thick disk stars and it assumes solar metallicity for all objects in the sample.” Two related results. Kilic et al. 2019 (arXiv:1810.03536) identify 142 field halo white dwarfs from 156,044 Gaia candidates and state "The latter provides a firm lower limit of 10.3 Gyr for the age of the inner halo based on the well-understood physics of white dwarfs", with a total inner-halo age of 10.9 ± 0.4 Gyr (internal errors only). Harris et al. 2006 (AJ 131: 571, arXiv:astro-ph/0510820) locate the disk luminosity-function drop at bolometric magnitude 15.3 but explicitly decline to convert it to an age, because "the small number of low-luminosity stars in the sample and their unknown atmospheric composition prevent quantitative conclusions about this decline." ↩
35.
Cowan, John J. et al. “The Chemical Composition and Age of the Metal-Poor Halo Star BD +17° 3248.” Astrophysical Journal 572:861–879. 2002. Section Abstract: “We tentatively identify U II at 3859 Å in the spectrum of BD +17° 3248 which makes this the second detection of uranium in a very metal-poor halo star. … we determine an average cosmochronological age for BD +17° 3248 of 13.8 ± 4 Gyr, consistent with that found for other similar metal-poor halo stars.” Table 4: The pair-to-pair scatter within this single star is: Th/Eu 10.0 Gyr, Th/Pt 10.3 Gyr, Th/Ir 21.7 Gyr, Th/U at least 13.4 Gyr, U/Ir at least 15.5 Gyr, U/Pt at least 12.4 Gyr. A useful check they perform: ages computed from measured solar-system r-process ratios, rather than theoretical production ratios, average 13.2 Gyr, which they note. “is not inconsistent with the average based upon theoretical”. ratios, so substituting measurement for theory changes the answer here by less than 1 Gyr. The uranium identification is stated by them as tentative. A study of the chemical composition of BD +17° 3248, a very metal-poor halo star, including a tentative identification of uranium in its spectrum. Its average age, from six pairs of radioactive clocks, is 13.8 ± 4 billion years. The article uses it for that age, which it gives as 13.8 ± 4 Gyr from six chronometer pairs. ↩
36.
Frebel, Anna et al. “Discovery of HE 1523−0901, a Strongly r-Process Enhanced Metal-Poor Star with Detected Uranium.” Astrophysical Journal Letters 660:L117. 2007. “For the first time, we are able to employ several different chronometers, such as the U/Th, U/Ir, Th/Eu and Th/Os ratios to measure the age of a star. The weighted average age of HE 1523−0901 is 13.2 Gyr.” Restricting to the uranium-based chronometers gives 13.4 Gyr, with an observational uncertainty of 0.8 Gyr and a model-atmosphere uncertainty of 0.9 Gyr, and a production-ratio uncertainty that is. “much reduced to 1.8 Gyr.” The authors note the observational bar: “The strongest U line at optical wavelengths is very weak, requiring very high quality spectroscopic data.” The steepness of the conversion: HE 1523−0901 comes out 1.5 Gyr younger than CS 31082-001 from. “only a 0.07 dex difference in the observed U/Th ratio.” Their cross-check on the earlier star gives log ε(U/Th) = −0.93, “in very good agreement with the published value of −0.89.” A short letter announcing a metal-poor halo star, HE 1523−0901 (V = 11.1, [Fe/H] = −2.95, [r/Fe] = 1.8), with uranium detected in its spectrum. The authors measured it with VLT/UVES at a resolving power of 75,000 and dated it with several chronometers at once. The article uses it as the first star where several different chronometers could be used together, giving an age of 13.2 Gyr. ↩
1. 36a.
  Table 1, page 5: The derived age of the single star HE 1523−0901, from one spectrum and one set of measured abundances, under fifteen chronometer-and-production-ratio combinations. The ages run: Th/Eu 9.5, 11.7 and 13.3 Gyr under the Sneden 2003, Schatz 2002 and Cowan 2002 production ratios respectively; Th/Os 10.7; Th/Ir 12.1 and 17.8; U/Eu 13.2; U/Os 12.9; U/Ir 13.3 and 14.8; and U/Th 12.2, 12.4, 13.1, 13.4 and 13.9 under five different theoretical production ratios. Every difference in that column comes from the choice of theoretical production ratio, not from the data. The production-ratio uncertainty term is 5.6 Gyr for every Th/r pair and 1.6–2.2 Gyr for every uranium-based pair, dwarfing the observational, temperature, gravity and microturbulence terms. The authors state the reason U/Th survives: “the U/Th PR should be less sensitive to theoretical r-process model uncertainties compared with ratios of elements [far apart in mass]”. “∼ 2 Gyr is roughly the age spread caused by the different PRs for U/Th.” ↩
37.
Holmbeck, Erika M. et al. “The R-Process Alliance: 2MASS J09544277+5246414, the Most Actinide-Enhanced R-II Star Known.” Astrophysical Journal Letters 859:L24. 2018. The paper cautions that the three thorium lines disagree among themselves: “span an abundance range of 0.23 dex”. Among themselves, on the 46.7 Gyr-per-dex conversion, that spread is ±5 Gyr of internal scatter by itself. Only three Th II lines and one U II line were measured. Section Abstract: “With an enhancement of [Th/Eu] = +0.37, 2MASS J09544277+5246414 is also the most actinide-enhanced r-II star yet discovered, and only the sixth metal-poor star with a measured uranium abundance ([U/Fe] = +1.40). Using the Th/U chronometer, we estimate an age of 13.0±4.7 Gyr for this star.” A letter from the R-Process Alliance on one star, 2MASS J09544277+5246414, which is strongly enriched in r-process elements (those made by rapid neutron capture), including thorium and uranium. The authors date it with the thorium-to-uranium ratio, which gives 13.0 ± 4.7 Gyr. The article uses it for that age, and for the star's place as the sixth metal-poor star with a measured uranium abundance. ↩
1. 37a.
  Section 3, page 7: “Using the production ratios from Schatz et al. (2002), the Th/U age is 13.0±4.7 Gyr, commensurate with its low metallicity. Uncertainties on the age contain only the measured abundance uncertainty and do not include those on the production ratios. The Th/Eu and U/Eu ages are −9.5 Gyr and 5.8 Gyr, respectively. These ages can be made to agree with the Th/U age of 13.0 Gyr by boosting the initial production ratios of thorium and uranium by a factor of 3.1. In a similar analysis, Schatz et al. (2002) found a 'boost factor' of 2.5 for CS 31082–001.” Their description of the anomaly: “about 30% of r-process-enhanced stars exhibit Th/Eu abundance ratios a factor of 2–3 higher than the majority of other r-process-enriched stars. … The mechanism for producing this 'actinide boost' is neither well-studied nor well-understood and suffers from low number statistics.” And: “in actinide-boost stars, using the Th/Eu ratio leads to unphysical values, since the measured Th/Eu ratio may be higher than the theoretical initial production ratio. To our knowledge, none of the existing r-process nucleosynthesis models can account for the abundance pattern observed in actinide-boost stars.” The reason U/Th survives: “the Th/U chronometer can still be used to estimate stellar ages since the actinide boost affects thorium and uranium proportionally.” ↩
38.
Hill, V. et al. “First stars I. The extreme r-element rich, iron-poor halo giant CS 31082-001.” Astronomy & Astrophysics 387:560–579. 2002. A detailed study of the giant star CS 31082-001, which is very rich in r-process elements and poor in iron. Its authors test whether one production ratio for the heavy elements applies to all stars, and find that it breaks down in the actinide region, where the thorium and uranium lie. The article uses it for the authors' verdict on chronometers based on actinides, and for their re-measured atomic data.
1. 38a.
  The verdict quoted in the text is that any chronometer. “based on ratios of an actinide to any stable element from the second r-process peak is doomed to failure. The Th/Eu ratio in CS 31082-001 is an extreme example of such a failure.” Section Abstract: “If CS 31082-001 and CS 22892-052 have similar ages, as expected for two extreme halo stars, this implies that the production ratios were different by about 0.4 dex for the two objects. Conversely, if the Th/Eu production ratio were universal, an age of 15 Gyr for CS 22892-052 would imply a negative age for CS 31082-001. Thus, while a universal production ratio for the r-process elements seems to hold in the interval 56 ≤ Z ≤ 72, it breaks down in the actinide region.” Section 6: In their section 6: “use of the conventional Th/Eu chronometer (Cowan et al. 1999) leads instead to a slightly negative(!), or at most a T-Tauri like age for CS 31082-001.” ↩
2. 38b.
  On the effect of re-measured oscillator strengths: “The change in the U/Th abundance ratio is quite significant, revising the log(U/Th) from the previous value of −0.74 ± 0.15 to −0.94 ± 0.11. Using the same initial production ratio as in Cayrel et al. (2001), this leads to an age of almost 17 Gyr, 4.3 Gyr greater than that originally published.” Adopting an updated production ratio (Goriely & Arnould 2001, mean 0.50 ± 0.02), they land on: “this leads to an age of 14.0 ± 2.4 Gyr for the Th and U in CS 31082-001 (where the error refers only to the uncertainty on the observed U/Th ratio).” Their candid statement of what that error bar does and does not cover: “At this point, it is impossible to assign a reliable error estimate to this age, given the lack of observational constraints on the production ratio of U/Th from other species that might have experienced the same neutron exposure.” ↩
39.
Shah, Shivani P. et al. “Uranium Abundances and Ages of R-process Enhanced Stars with Novel U II Lines.” Astrophysical Journal. 2023. A 2023 paper by Shah and colleagues that re-measures the uranium in four old, metal-poor stars using two new spectral lines, and re-derives their ages from the radioactive decay of uranium and thorium. That clock does not depend on the stellar models that the other dating methods need. The article uses it because it shows how far the ages of the same stars move when the atomic data and the nuclear theory change. Mirror: arXiv
1. 39a.
  Section Introduction: The authors' summary of the wider problem: “More generally, the application of the Th/Eu chronometer to RPE stars has led to a large range in ages from 21 Gyr to -9 Gyr, even though these stars are metal-poor (Ji & Frebel 2018; Holmbeck et al. 2018).” On how small the sample is: “U abundance has been reliably determined for only five RPE stars.” Their own stated limitation: “we have not taken into account the systematic uncertainties associated with using just one set of theoretical PRs (from Schatz et al. 2002). Other r-process models have predicted slightly varying zero-age abundance ratios for U/Th and U/Eu.” ↩
2. 39b.
  Table 5: Their homogeneous re-analysis adds two new U II lines, at 4050 Å and 4090 Å, to the canonical 3859 Å line and re-derives ages for four stars. U/Th ages, with the systematic, statistical and production-ratio errors separated: J0954+5246, 11.1 ± 6.4 (±5.7 ±1.9 ±2.2); J2038−0023, 13.5 ± 4.8; HE 1523−0901, 16.6 ± 5.1; CS 31082−001, 11.1 ± 4.0. U/Eu ages: 12.0 ± 4.3, 11.3 ± 3.1, 14.2 ± 3.8, and 7.3 ± 2.6 respectively. “The resulting total uncertainties on the ages of the sample stars are on the order of ∼ 4 − 6 Gyr for the U/Th ages and ∼ 3 − 4 Gyr for the U/Eu ages.” On the CS 31082-001 discrepancy: “For CS 31082-001, U/Eu age is ∼ 4.0 Gyr lower than the U/Th age due to its actinide-boost nature.” The central U/Th value for HE 1523−0901, 16.6 Gyr, is formally older than the universe, and the paper's Figure 8 plots the Planck age as a dashed line against its points. Frebel's 13.2 Gyr for that star is given alongside it, because reporting only the older-than-universe value would be selective. The error bar of 16.6 ± 5.1 covers 13.8 comfortably. ↩
40.
Goriely, S., and M. Arnould. “Actinides: How well do we know their stellar production?” Astronomy & Astrophysics 379:1113–1122. 2001. A paper that asks how well the stellar production of the actinides, the heaviest elements used as clocks, is known. It shows that if one production ratio applied to every star, two stars would come out at very different ages, one of them about 36 Gy old. The article uses it because it puts the same absurdity the other way round: under a universal production ratio, one star would be about 36 Gy old.
1. 40a.
  Section 4, page 6: “In particular, the Th/Eu ratio in CS 31082-001 is about 3.2 times larger than in CS 22892-052. Hence, under the universality assumption, CS 22892-052 (with [Fe/H]=-3.1) predates CS 31082-001 (with [Fe/H]=-2.9) by 24 Gy, and would thus be about 36 Gy old.” The authors add an independent strike from lead: “if the universality of the Pb/Th ratio is assumed, the observed Pb/Th values turn out to be discrepant by a factor of about 10, at least if the two stars have roughly the same age. … either the universality assumption is invalid, and a specific actinide-producing r-process has to be called for, or the Pb in CS 22892-052 is largely of s-process origin.” ↩
41.
Holmbeck, Erika M. et al. “Actinide-rich and Actinide-poor r-Process Enhanced Metal-Poor Stars do not Require Separate r-Process Progenitors.” Astrophysical Journal. 2019. The key result, in the paper's words: “actinide-boost stars do not indicate the need for a unique and separate r-process progenitor. Rather, small variations of neutron richness within the same type of r-process event can account for all observed levels of actinide enhancements. The very low-Y_e, fission-cycling ejecta of an r-process event need only constitute 10–30% of the total ejecta mass to accommodate most actinide abundances of metal-poor stars.” The tie to a witnessed event: “our empirical Y_e distributions of ejecta are similar to those inferred from studies of GW170817 mass ejecta ratios, which is consistent with neutron-star mergers being a source of the heavy elements in metal-poor, r-process enhanced stars.” A paper that tests whether the extra thorium and uranium in actinide-boost stars need their own kind of r-process source. It finds that small variations in neutron richness within one class of r-process event can produce all the observed actinide enhancements. The article uses it as a real physical answer to the actinide-boost problem, not a rescue. ↩
42.
Watson, Darach et al. “Identification of strontium in the merger of two neutron stars.” Nature 574:497–500. 2019. Section Abstract: “Here we report the identification of the neutron-capture element strontium in a re-analysis of these spectra. The detection of a neutron-capture element associated with the collision of two extreme-density stars establishes the origin of r-process elements in neutron star mergers, and demonstrates that neutron stars comprise neutron-rich matter.” The authors' caution about the derived masses: “These numbers should be treated with some caution as this is the derived mass assuming spherical symmetry, a fixed photospheric velocity, and no correction for light travel time effects. They must be interpreted as lower limits to the total amount of material ejected.” A 2019 paper that re-analyses the spectra of the 2017 collision of two neutron stars, extremely dense stars, and identifies strontium in them, a heavy element made by neutron capture. The authors say this establishes that neutron-star mergers make r-process elements. The article uses it as the identification that fills the gap its opening paragraph describes, namely evidence of freshly made r-process elements in an astrophysical site. Two further identifications, from other papers that were not read for this article, are Sneppen et al. 2023 (A&A 675: A194, a 760 nm P Cygni line attributed to yttrium) and Hotokezaka et al. 2023 (MNRAS 526: L155–L159, a tellurium line). On the published record the tellurium identification is Hotokezaka's, not Sneppen's, so both should be checked before assigning either claim. Mirror: arXiv ↩
1. 42a.
  Section Opening paragraph: “Half of all the elements in the universe heavier than iron were created by rapid neutron capture. The theory for this astrophysical 'r-process' was worked out six decades ago and requires an enormous neutron flux to make the bulk of these elements. Where this happens is still debated. A key piece of missing evidence is the identification of freshly-synthesised r-process elements in an astrophysical site.” ↩
43.
Pian, E. et al. “Spectroscopic identification of r-process nucleosynthesis in a double neutron star merger.” Nature 551:67–70. 2017. Section abstract: “we find that the kilonova is characterized by rapidly expanding ejecta with spectral features similar to those predicted by current models. The ejecta is optically thick early on, with a velocity of about 0.2 times light speed, and reaches a radius of ∼ 50 astronomical units in only 1.5 days. … Comparison with spectral models suggests that the merger ejected 0.03–0.05 solar masses of material, including high-opacity lanthanides.” The event is GW170817, on 17 August 2017 at 12:41:04 UT, with the optical counterpart AT 2017gfo in the galaxy NGC 4993 at 40 megaparsecs. A research paper that took spectra of the kilonova, the optical glow from the debris of the neutron-star merger GW170817, and compared them with models of the ejected material. The article uses it for the spectra that identify the merger as a source of heavy elements. ↩
44.
Kasen, Daniel et al. “Origin of the heavy elements in binary neutron-star mergers from a gravitational wave event.” Nature 551:80–84. 2017. “We infer the presence of two distinct components of ejecta, one composed primarily of light (atomic mass number less than 140) and one of heavy (atomic mass number greater than 140) r-process elements. Inferring the ejected mass and a merger rate from GW170817 implies that such mergers are a dominant mode of r-process production in the Universe.” The heavy component is inferred from opacity, because lanthanide-rich ejecta make a red, slow-fading infrared light curve, and not from a thorium or uranium line. No actinide abundance has ever been measured in a kilonova. “The inferred ejecta mass of GW170817, however, should be considered approximate, owing to uncertainties in the radioactive heating rate, the uncertain atomic data used for opacities, and possible viewing angle effects.” A theory paper that models the ejected material from the neutron-star merger GW170817, which was detected through its gravitational waves. It infers two components of ejecta, one rich in light heavy elements and one rich in heavy ones, and concludes that such mergers are a dominant source of the rapid neutron-capture (r-process) elements. The article uses it for the two-component model of the merger's debris. ↩
1. 44a.
  Section closing paragraph: “In the past, the uncertain origin of the heaviest elements was studied indirectly, by analysing fossil traces of these species in the surfaces of old stars. With AT 2017gfo we have now directly glimpsed and spectroscopically dissected a sample of pure r-process matter, big enough to enrich a million such stars.” ↩
45.
Wallner, A. et al. “Abundance of live ²⁴⁴Pu in deep-sea reservoirs on Earth points to rarity of actinide nucleosynthesis.” Nature Communications 6:5956. 2015. Section Abstract: “In particular ²⁴⁴Pu, a radioactive actinide nuclide (half-life = 81 My), can place strong constraints on recent r-process frequency and production yield. Here we report the detection of live interstellar ²⁴⁴Pu, archived in Earth's deep-sea floor during the last 25 My, at abundances lower than expected from continuous production in the Galaxy by about 2 orders of magnitude. This large discrepancy may signal a rarity of actinide r-process nucleosynthesis sites, compatible with neutron-star mergers or with a small subset of actinide-producing supernovae.” Three things follow. The same class of isotopes is being measured two entirely different ways, through a spectrograph kiloparsecs away and as atom counts in Pacific sediment by accelerator mass spectrometry, and both point at rare, specific, neutron-rich events. The rarity is what makes 'time since the last event' a meaningful question rather than a steady-state one. A count of surviving atoms is not a ratio: an accelerated-decay episode would have destroyed the ²⁴⁴Pu inventory outright rather than rescaling it. Related follow-ups: Hotokezaka, Piran & Paul, Nature Physics 11 (2015): 1042; Wallner et al., Science 372 (2021): 742–745. A 2015 paper that counts atoms of plutonium-244, a radioactive element made in rare, neutron-rich events, that arrived on Earth from space and sits in Pacific deep-sea sediment. The atoms were counted by accelerator mass spectrometry, and there are about 100 times fewer than continuous production in the galaxy would predict. The article uses it to show that the heaviest elements come from rare, specific events, so the time since the last such event is a meaningful question. ↩
46.
Baumgardt, Holger, and Eugene Vasiliev. “Accurate distances to Galactic globular clusters through a combination of Gaia EDR3, HST and literature data.” Monthly Notices of the Royal Astronomical Society 505:5957–5977. 2021. A catalogue paper that measured the distances of Milky Way globular clusters, the dense balls of old stars that orbit the galaxy. It combined five independent methods, using Gaia satellite data, Hubble Space Telescope data and older literature. The article uses it to check the distance used in the white dwarf cooling fit, and to show that no globular cluster lies within 6,000 light years of the Sun.
1. 46a.
  Section 4.2.4, page 11: Preprint page. “We derive a distance of D = 1.851 ± 0.015 kpc to M4 (NGC 6121).” Kaluzny et al. 2013 obtained 1.82 ± 0.04 kpc from three eclipsing binaries. “The true distance to NGC 6121 is therefore likely somewhere in the range 1.83 to 1.85 kpc, making M4 another cluster whose distance is determined with an accuracy of about 1%.” The light-year figure is a conversion at 3261.564 light years per kiloparsec, which makes 1.851 kpc 6,037 light years. ↩
2. 46b.
  Section 4.2.5, page 12: Preprint page. “we derive a distance of 2.488 ± 0.019 kpc for NGC 6397, making this cluster the second closest cluster to the Sun.” Five independent routes are combined: direct Gaia EDR3 parallaxes, kinematic distances from Gaia proper-motion dispersions matched to line-of-sight velocity dispersions, the same from HST proper motions, subdwarf main-sequence fitting calibrated by Gaia parallaxes, and HST stellar number counts. Section Abstract: “We find good agreement in the average distances derived from the different methods down to a level of about 2%. Combining all available data, we are able to derive distances to 162 Galactic globular clusters, with the distances to about 20 nearby globular clusters determined with an accuracy of 1% or better.” ↩
47.
VandenBerg, Don A. et al. “The Ages of 55 Globular Clusters as Determined Using an Improved ΔV^HB_TO Method Along with Color-Magnitude Diagram Constraints, and Their Implications for Broader Issues.” Astrophysical Journal 775:134. 2013. A 2013 study that dates 55 globular clusters, tight balls of stars that orbit a galaxy, by the brightness difference between two landmarks on each cluster's colour-magnitude diagram (a plot of stars' brightness against their colour), the horizontal branch and the turnoff, checked against theoretical models. The article uses its ages for NGC 6397, NGC 6121 (M4), NGC 104 (47 Tuc) and NGC 6656 (M22), and for a live dispute over which age, above 11 billion years, is right. Mirror: arXiv
1. 47a.
  Table 2, pages 122–123: The values used in this article: NGC 6397, 13.00 ± 0.25 Gyr at [Fe/H] = −1.99; NGC 6121 (M4), 11.50 ± 0.38 at −1.18; NGC 104 (47 Tuc), 11.75 ± 0.25 at −0.76; NGC 6656 (M22), 12.50 ± 0.50 at −1.70. The range across the whole sample runs from 9.0 ± 0.38 Gyr for Pal 12, a Sagittarius dwarf galaxy cluster, to 13.00 Gyr for five clusters. Read the error bars correctly, because this is the most abused number in the literature. The ± values printed in Table 2 are fitting errors, and the total error is much larger. Quoting the figure without saying which error it is misreports the source, as in: “13.00 ± 0.25 Gyr for NGC 6397”. The abstract gives the total: Section Abstract, pages 1–2: “For well-defined CMDs, the error bar arising just from the 'fitting' of ZAHBs and isochrones is ≈ ±0.25 Gyr, while that associated with distance and chemical abundance uncertainties is ∼ ±1.5–2 Gyr. The oldest GCs in our sample are predicted to have ages of ≈ 13.0 Gyr (subject to the aforementioned uncertainties).” The same abstract reports the age–metallicity relation: “ages are found to vary from mean values of ≈ 12.5 Gyr at [Fe/H] ≲ −1.7 to ≈ 11 Gyr at [Fe/H] ≳ −1. At intermediate metallicities, the age-metallicity relation (AMR) appears to be bifurcated.” A live disagreement with Marín-Franch et al. 2009 and Dotter et al. 2010, who 'argued that the majority of the Galactic GCs are nearly coeval' (preprint p. 51): Page 51: “argued that the majority of the Galactic GCs are nearly coeval”. An unexplained problem of their own (section 3.2, preprint p. 15): their isochrones had to be shifted blueward by 0.01–0.025 mag to fit observed turnoff colours, and 'The cause of such discrepancies is not known.'. The wider modern picture is consistent. Valcin et al. 2020 (JCAP 12 (2020) 002, arXiv:2007.06594v2) fit the full colour-magnitude diagram of 68 clusters in a Bayesian framework and find. “the average age of the oldest globular clusters is tGC = 13.32 ± 0.1(stat.) ± 0.5(sys.)” Valcin et al. 2021 (JCAP 08 (2021) 017, arXiv:2102.04486v2) cut that systematic to 0.23–0.33 Gyr by showing empirically that cluster red giant branches demand a convection depth consistent with the Sun's. ↩ ↩ ↩
48.
Hansen, Brad M. S. et al. “An Age Difference of 2 Gyr between a Metal-Rich and a Metal-Poor Globular Cluster.” Nature 500:51–53. 2013. A Nature paper that dates the metal-rich globular cluster 47 Tucanae by fitting the properties of its white dwarf population, and finds it about 2 Gyr younger than the metal-poor cluster NGC 6397 from the same models. The authors observed one field in the cluster with the Hubble Space Telescope and fitted Monte Carlo models of white dwarf cooling to its luminosity function and colour-magnitude distribution. The article uses it as the white dwarf age for 47 Tucanae, which is 9.9 ± 0.7 Gyr.
1. 48a.
  Section Abstract: “Here we present an age determination of the metal-rich globular cluster 47 Tucanae by fitting the properties of the cluster white dwarf population, which implies an absolute age of 9.9 ± 0.7 Gyr at 95% confidence. This is ∼ 2.0 Gyr younger than inferred for the metal-poor cluster NGC 6397 from the same models.” Page 3: “we obtain an age of 9.7 ± 0.4 Gyr (at 95% confidence) for 47 Tuc, as compared to the corresponding value of 11.7 ± 0.3 Gyr for NGC 6397. Therefore, the relative age difference is 2.0 ± 0.5 Gyr.” The authors state their own tension with the turnoff ages: “Estimates based on fitting main sequence models to the main sequence turnoff report ages between 10 and 13 Gyr, with the range due to a variety of factors including differences in adopted models physics.” Page 4: On the eclipsing binary cross-check: “the age derived from the binary system is 10.39 ± 0.54 Gyr, here citing only the systematic error. This estimate is fully consistent with the WD result given above, suggesting that consistency between the two age measurement techniques is possible, with accurate determination of extrinsic properties such as distance, reddening and metallicity.” ↩
49.
Hansen, Brad M. S. et al. “Hubble Space Telescope Observations of the White Dwarf Cooling Sequence of M4.” Astrophysical Journal Supplement Series 155:551–576. 2004. Section Abstract, page 1: “After taking into account the various uncertainties, we find a best fit age of 12.1 Gyr, with a 95% lower limit of 10.3 Gyr.” A caveat noted in the text: they adopted a true distance modulus varying over 11.18 ± 0.18, i.e. centred on 1.72 kpc (p. 18), whereas the modern Gaia-era value is 1.851 kpc (modulus 11.34), which is inside their range but on the far side of their central value. A larger distance makes the white dwarfs intrinsically brighter, hence hotter and younger, so the modern distance would nudge the cooling age down from 12.1 toward the turnoff value of 11.5. The conservative statement is that M4 is 11–12 Gyr old by two independent methods. The earlier and more widely quoted M4 result is Hansen et al. 2002 (ApJL 574: L155, arXiv:astro-ph/0205087), which says in section 4 (p. 6): Section 4, page 6: “We find an age for the globular cluster M4 of 12.7±0.7 Gyr using the white dwarf cooling sequence … This method is completely independent of the main sequence turnoff age method, which yields an average age of 13.2 ± 1.5 Gyr for the metal poor globular clusters (Chaboyer 2001). Our age for M4 is also in excellent agreement with a third very different stellar chronometer, namely the nuclear decay measurements of uranium and thorium, which yields 12.5 ± 3 Gyr for a metal-poor halo star (Cayrel et al. 2001).” A Hubble Space Telescope study of the white dwarfs in the globular cluster M4, which fits the whole cooling sequence to estimate the cluster's age. The authors fit a two-dimensional colour-magnitude diagram rather than a one-dimensional luminosity function, marginalising over distance, extinction, white dwarf mass, and envelope and core composition. The article uses it as the white dwarf clock on M4, which gives 12.1 Gyr with a 95% lower limit of 10.3 Gyr. ↩
50.
Silva Aguirre, V. et al. “Standing on the shoulders of dwarfs: the Kepler asteroseismic LEGACY sample II – radii, masses, and ages.” Astrophysical Journal 835:173. 2017. A 2017 study that dates a large sample of main-sequence stars, the stage the Sun is in, from the way they vibrate (asteroseismology), using Kepler satellite data. Its key test sent seven independent modelling teams the vibrations of a disguised Sun, without telling them what star it was, and asked for its mass, radius and age. The article uses the test to show what happens when several teams date the same star. Mirror: arXiv
1. 50a.
  Page 12: “The true identity of this star was concealed from the modellers.” Table 2, page 15: What the seven independent pipelines returned for the disguised Sun. Masses, against a true value of 1.000 solar masses (M☉): 0.979, 0.986, 0.978, 1.021, 0.997, 0.927 and 1.037 M☉. Radii, against a true 1.000 solar radii (R☉): 0.992, 0.994, 0.993, 1.006, 0.995, 0.973 and 1.012 R☉. Ages: 4840 ± 367, 4686 ± 393, 4852 (+1181/−1069), 4331 ± 85, 4859 ± 128, 4621 ± 200 and 5297 ± 350 Myr. The table note gives the reference value: “the solar age is determined to be 4.57 Gyr (Bahcall et al. 1995).” Page 12: The one high outlier is diagnosed as a pipeline whose grid lacked microscopic diffusion, the slow sorting of elements by gravity inside a star. Its authors write: “A test including this effect in the YMCM modelling results in an estimated age of tau = 4.83 Gyr, in better agreement with the solar value of 4.57 Gyr.” The Sun's age is known independently, from lead-lead dating of calcium-aluminium-rich inclusions in meteorites. That is radioactive decay in solid rock, and it shares no physics with sound waves in a star. Section Abstract: On the precision now achieved for the wider sample: “Average uncertainties are of the order of ~2% in radius, ~4% in mass, and ~10% in age, making this the best-characterised sample of main-sequence stars available to date.” ↩
51.
Tayar, Jamie et al. “A Guide to Realistic Uncertainties on the Fundamental Properties of Solar-type Exoplanet Host Stars.” Astrophysical Journal 927:31. 2022. Section Abstract: “current uncertainties on measured interferometric angular diameters and bolometric fluxes set a systematic uncertainty floor of ~2% in temperature, ~2% in luminosity, and ~4% in radius. Comparisons between widely available model grids suggest uncertainties of order ~5% in mass and ~20% in age for main sequence and subgiant stars.” The defensible claim, which the article takes from this paper, is that asteroseismic ages of individual solar-type stars are good to roughly 10–20 per cent, not 1 per cent. Twenty per cent of 12 billion years is 2.4 billion years. Two further caveats belong with this one. Chaplin and Miglio 2013 (Annual Review of Astronomy and Astrophysics 51: 353) state the model dependence plainly, on preprint p. 64: Page 64: “estimation of the ages is inherently model dependent. Systematic uncertainties from predictions of main-sequence lifetimes need to be taken into account”. For a 1.4 M☉ model, including convective core overshoot changes the main-sequence lifetime. “of the order of 20 %.” The surface effect, on preprint p. 33: Page 33: “A well-known problem in comparing observed and model-calculated frequencies comes from the hard-to-model near-surface layers of stars,”. That biases the inferred density by a few tenths of a per cent using frequencies, and by 1 to 2 per cent using the large separation, which roughly doubles into mass. What asteroseismology is independent of is not the stellar model but the fitting: instead of inferring an interior from two surface numbers, it measures the interior and uses the model only to run a clock forward. A 2022 guide that asks how exact the measured properties of Sun-like stars that have planets really are, and which errors a careful study must carry. It compares widely used stellar model grids, the lookup tables that turn a star's measured brightness and temperature into its mass and age, and finds how far they disagree. The article uses it to set the honest size of the error bar on asteroseismic ages, which is about 10 to 20 per cent rather than 1 per cent. Mirror: arXiv ↩
52.
Epstein, Courtney R. et al. “Testing the Asteroseismic Mass Scale Using Metal-Poor Stars Characterized with APOGEE and Kepler.” Astrophysical Journal Letters 785:L28. 2014. Section Abstract: “Although our sample is small, standard scaling relations, combined with asteroseismic parameters from the APOKASC Catalog, produce masses that are systematically higher ( = 0.17 ± 0.05 M⊙) than astrophysical expectations. The magnitude of the mass discrepancy is reduced by known theoretical corrections to the measured large frequency separation scaling relationship.” Miglio et al. 2016's M4 work, with the model-suggested Δν correction applied, is the fix. A short research letter that compared masses from asteroseismology, which reads the pulsations of a star to find its mass, for nine metal-poor red giants. Six of the nine are kinematically halo stars, objects whose masses are tightly constrained in advance, because a population 12 billion years old cannot have giants heavier than about 0.9 solar masses. The standard scaling relationships gave masses that were systematically too high, and the authors say known theoretical corrections reduce the gap. The article uses it because a seismic mass that is therefore biased high makes the inferred age of a metal-poor giant come out too young. ↩
53.
Bond, Howard E. et al. “HD 140283: A Star in the Solar Neighborhood that Formed Shortly After the Big Bang.” Astrophysical Journal Letters 765:L12. 2013. Section Abstract, page 1: Preprint page. “Using the Fine Guidance Sensors on the Hubble Space Telescope, we have measured a trigonometric parallax of 17.15 ± 0.14 mas for HD 140283, with an error one-fifth of that determined by the Hipparcos mission. Employing modern theoretical isochrones … we use the precise distance to infer an age of 14.46 ± 0.31 Gyr. The quoted error includes only the uncertainty in the parallax … Uncertainties in the stellar parameters and chemical composition, especially the oxygen content, now contribute more to the error budget for the age of HD 140283 than does its distance, increasing the total uncertainty to about ±0.8 Gyr.” The ±0.31 and ±0.8 are two different quantities and should not be conflated. A Hubble Space Telescope measurement of the distance to HD 140283, a star near the Sun. Combined with modern stellar models, the distance gives an age of 14.46 billion years. The parallax error is one-fifth of the Hipparcos error. Their Hubble parallax has since been superseded by Gaia: Gaia's 16.267 mas disagrees with 17.15 by about six sigma and is five times more precise, and it moves the star further away, which raises rather than lowers the inferred age. The article uses it as the first of the age measurements for this star. ↩
1. 53a.
  Preprint, immediately after the abstract passage. “Within the errors, the age of HD 140283 does not conflict with the age of the Universe, 13.77 ± 0.06 Gyr, based on the microwave background and Hubble constant, but it must have formed soon after the big bang.” The "star older than the universe" framing was therefore a press artefact rather than the authors' claim. ↩
54.
VandenBerg, Don A. et al. “Three Ancient Halo Subgiants: Precise Parallaxes, Compositions, Ages, and Implications for Globular Clusters.” Astrophysical Journal 792:110. 2014. Section Abstract, page 2: “Comparisons of isochrones with the three subgiants on the (log Teff, MV)-diagram yielded ages of 12.08 ± 0.14, 12.56 ± 0.46, and 14.27 ± 0.38 Gyr for HD 84937, HD 132475, and HD 140283, in turn, where each error bar includes only the parallax uncertainty. The total uncertainty is estimated to be ∼ ±0.8 Gyr (larger in the case of the near-turnoff star HD 84937). Although the age of HD 140283 is greater than the age of the universe as inferred from the cosmic microwave background by ∼ 0.4–0.5 Gyr, this discrepancy is at a level of < 1 σ.” Do not present 12.08 ± 0.14 as though 0.14 were the real error bar: it is the parallax contribution alone, and the authors specifically flag HD 84937 as the case with the larger total uncertainty. Two further results from the same abstract follow. The first: “(Stellar models that neglect diffusive processes seem to be ruled out as they would predict that HD 140283 is ∼ 1.5 Gyr older than the universe)”. The diffusion input is not free to be dropped. The second, in the same abstract: “The field halo subgiants appear to be older than globular clusters of similar metallicities: if distances close to those implied by the RR Lyrae standard candle are assumed, M 92 and M 5 are younger than HD 140283 and HD 132475 by ∼ 1.5 and ∼ 1.0 Gyr, respectively.” That is an unexplained 1–1.5 Gyr discrepancy, and the article reports it as such. The authors' statement of why these are the best available ages: “The most accurate ages for the oldest stars are those obtained for nearby halo subgiants, because they depend almost entirely on just the measured parallaxes and absolute oxygen abundances.” A 2014 paper that dates three very old halo subgiant stars, which have just left the main sequence, using measured parallaxes (the tiny shift in each star's apparent position as Earth orbits the Sun, which gives its distance) and chemical compositions. Its ages are among the most precise available for the oldest stars, because they depend almost entirely on those parallaxes and on the oxygen abundances. The article uses it for the age of HD 140283 and for the unexplained gap between these field stars and the globular clusters. Mirror: arXiv ↩
55.
Creevey, O. L. et al. “Benchmark stars for Gaia. Fundamental properties of the Population II star HD 140283 from interferometric, spectroscopic, and photometric data.” Astronomy & Astrophysics 575:A26. 2015. Section Abstract, page 1: Preprint page. “Establishing a reference α = 1.00 and adopting Yi = 0.245 we derive a mass and age of HD 140283: M = 0.780 ± 0.010 M⊙ and t = 13.7 ± 0.7 Gyr (AV = 0.0 mag), or M = 0.805 ± 0.010 M⊙ and t = 12.2 ± 0.6 Gyr (AV = 0.1 mag).” They measured the angular diameter as θ_1D = 0.353 ± 0.013 mas. They also flagged a convection anomaly: “We need to adjust α to much lower values than the solar one (∼ 2) in order to fit the observations, and if AV = 0.0 mag then 0.5 ≤ α ≤ 1.” Their quoted distance of 58 pc predates Gaia. A measurement paper that used the VEGA instrument on the CHARA array, an interferometer that combines light from several telescopes, to measure the angular size of HD 140283 directly. The measured diameter gives a radius of 2.21 ± 0.08 solar radii, and with the star's mass and age derived from it, the age shifts by 1.5 billion years for a 0.1 magnitude change in the assumed dust extinction. That 1.5 billion year shift is the clearest single demonstration that these ages are model-sensitive at the 1 to 2 billion year level, and not at the factor-of-a-million level. The article uses it as one of the measured ages for this star. ↩
56.
Tang, Jing, and Meridith Joyce. “Revised Best Estimates for the Age and Mass of the Methuselah Star HD 140283 Using MESA and Interferometry and Implications for 1D Convection.” Research Notes of the AAS 5:117. 2021. “HD 140283 requires a convective mixing length 10-20% below the solar value.” This is the lowest modern value in the whole published sequence for this star, and it is still 12 billion years. Page 1: “an age of 12.01 ± 0.05 Gyr (1 sigma)”. “When dominant sources of modeling uncertainty are taken into account, we report 0.81±0.05 M⊙ and 12±0.5 Gyr.” A short research note that re-models HD 140283, the Methuselah star, an extremely old and metal-poor star, with MESA, a computer code for stellar evolution, and with new interferometric measurements of its size. The authors give its best age and mass, and they also find that the star's convection, the way heat is carried outward through its outer layers, must be modelled with a shorter mixing length than the Sun's. The article uses it as the lowest modern age for this star, and that age is still 12 billion years. Mirror: arXiv ↩
57.
Barnes, Sydney A. et al. “Rotation Periods for Cool Stars in the 4 Gyr old Open Cluster M67, The Solar–Stellar Connection, and the Applicability of Gyrochronology to at least Solar Age.” Astrophysical Journal 823:16. 2016. Page 1: “rotational age of 4.2 Gyr, with a standard deviation of 0.7 Gyr”. Pages 2, 5: The isochrone age is given on page 2, and again in section 3.3 on page 5, as. “believed to be ~4 Gyr old (Demarque et al. 1992; VandenBerg & Stetson 2004; Bellini et al. 2010).” A study of how fast cool stars spin in M67, a cluster of stars about 4 billion years old. Its spin periods were turned into an age, which the article uses as an independent check on the cluster's age from its turnoff stars. ↩
58.
Maoz, Dan et al. “Observational Clues to the Progenitors of Type Ia Supernovae.” Annual Review of Astronomy and Astrophysics 52:107–170. 2014. A review of the observational evidence on the progenitors of Type Ia supernovae, the explosions that these systems produce. It sets out the delay-time distribution, which is how many of these explosions happen at each time after stars form, and what the observed rates say about it. The article uses it for the delay before the first white dwarfs can explode.
1. 58a.
  Section §3.5, page 41: “A fundamental function for progenitor-question purposes is the distribution of times between star formation and SN Ia explosion, usually called the delay-time distribution (DTD). The DTD is the hypothetical SN rate versus time that would follow a brief burst of star formation.” Section §3.5.1, page 42: “The ∼ t⁻¹ DTD behavior breaks down at short delays. A 'bottleneck' in the process is the production rate of WDs that can serve as SN Ia progenitors. The production rate is zero for at least 30 − 40 Myr, until main sequence stars with masses below ∼8 M⊙ evolve into the first WDs to emerge from the stellar population. The WD production rate then rises to a maximum, and falls as ∼ t⁻¹/² … up to a cutoff time, tc, corresponding to the main-sequence lifetime of the least massive stars whose descendant WDs contribute to a DD-channel SN Ia. Since this main-sequence mass is usually considered to be 2 − 3 M⊙, tc is generally ∼ 1 Gyr.” A t⁻¹ law has equal numbers of events per logarithmic time interval, so as many Type Ia supernovae occur between 1 and 10 Gyr after star formation as between 0.1 and 1 Gyr. Section §3.5.11, page 50: “In summary, SN Ia rate measurements quite consistently indicate a ∼ t⁻¹ DTD.” The t⁻¹ shape is fitted to measured supernova rates in galaxy clusters, in field ellipticals, against host colour and Hubble type, and against the volumetric rate versus redshift (their §§3.5.2–3.5.10). ↩
59.
Kobayashi, Chiaki et al. “The Origin of Elements from Carbon to Uranium.” Astrophysical Journal 900:179. 2020. A model paper that calculates which kinds of stars produced each element from carbon to uranium, and what share each kind supplies. It also lists the model's own failures, such as silver overproduced by a factor of 6 and gold underproduced by a factor of 5. The article uses it for the "knee" in the ratio of α-elements to iron, and for the authors' disagreement about what causes it.
1. 59a.
  Section introduction, page 2, column 2: “the [α/Fe]–[Fe/H] relation in the Milky Way Galaxy is explained by the delayed enrichment of Fe from SNe Ia, which have a longer timescale than core-collapse supernovae. Therefore, the [α/Fe] ratios can be used to constrain star formation timescales in other galaxies.” “Because of the nature of the triple α reactions, elements with A ≥ 12 are produced not during the Big Bang but are instead formed inside stars”. “The α-elements (O, Mg, Si, S, and Ca) are mostly produced in massive stars before being ejected by core-collapse (Type II, Ib, and Ic) supernovae … Conversely, half of the iron-peak elements (Cr, Mn, Fe, Ni, Co, Cu, and Zn) are produced by Type Ia Supernovae”. “The traditional main and strong s-process components (producing elements from Sr to Pb) are produced in the He-rich intershell of low-mass AGB stars … where the neutron source is mainly ¹³C(α,n)¹⁶O. The weak s-process component (from Fe to Sr) is produced instead in massive stars near solar metallicity.” Page 30: The measured source fractions are these. “49% of C, 51% of F, and 74% of N are produced by AGB stars”. “22% of Si, 29% of S, 34% of Ar, and 39% of Ca are come from SNe Ia”. The grammatical slip ("are come from") is in the preprint. “In classical works, most of Fe was thought to be produced by SNe Ia, but the fraction is only 60% in our model.” ↩
2. 59b.
  Page 11, column 2: “The [Fe/H] at which the [α/Fe] starts to decrease depends on the adopted SN Ia progenitor model, and is determined not by the lifetime but by the metallicity dependence of SN Ia progenitors in our models. It is very difficult to reproduce this rapid evolutionary change without the metallicity effect of SNe Ia.” This is a real internal disagreement in the field. Kobayashi's group builds Type Ia progenitors whose white-dwarf winds fail at low metallicity, which suppresses them in metal-poor gas whatever the clock says, while Matteucci's school reads the knee mainly as a timescale. Both readings still require a long history, because a metallicity-dependent progenitor efficiency is still a statement about a sequence of stellar generations. Page 30: The paper prints its own failures: silver is overproduced by a factor of 6, gold is underproduced by a factor of 5, uranium is overproduced. “The under-production of the elements around Ti is a long-standing problem.” ↩
60.
Tinsley, Beatrice M. “Stellar Lifetimes and Abundance Ratios in Chemical Evolution.” Astrophysical Journal 229:1046–1056. May 1, 1979. Page 1046: The abstract: “The delay in iron ejection explains the high [O/Fe] ratios of halo stars, while the absence of similar [C/Fe] ratios (in available data) supports the idea that carbon comes mainly from less massive stars than does oxygen.” Page 1055: Conclusion 5: “The excess [O/Fe] ratios found in halo stars can be explained if a significant fraction of iron comes from SN I, and oxygen from more massive stars (cf. Chevalier 1976). Since SN I precursors are assumed here to have lifetimes of < 10⁸ years, the excess only appears in stars formed within about 10⁸ years of the start of star formation.” Conclusion 4, on the same page: “it is found that the relative abundances of primary elements are essentially functions of time, determined by the progenitor stellar lifetimes.” A 1979 paper by Beatrice Tinsley on how long different stars take to die, and so how fast different elements are returned to a galaxy's gas. It argues that the high oxygen-to-iron ratio of halo stars comes from iron that arrives from SN I, a class of supernova whose progenitors are assumed to live under 10⁸ years, so the excess shows up only in the earliest stars. The article quotes it for that claim, that the excess appears only in stars formed within about 10⁸ years of the start of star formation. The paper was received 2 October 1978 and accepted 8 November 1978. Mirror: Articles.adsabs.harvard.edu ↩
61.
Keller, S. C. et al. “A single low-energy, iron-poor supernova as the source of metals in the star SMSS J031300.36−670839.3.” Nature 506:463–466. 2014. Section Methods: “<−7.7, <−7.5, <−7.3”. These are the one-dimensional LTE limits, which depend on the spectral line used. “the 99.7% upper limit becomes [Fe/H]3D,NLTE = <−7.1”. After the three-dimensional and non-LTE corrections, this is the value in the Nature abstract and the more conservative of the two. The article uses −7.1. Frebel & Norris 2015 prints < −7.30 for the same star from the one-dimensional analysis; these are one measurement analysed two ways, not two stars. A measurement paper on one star, SMSS J031300.36−670839.3, which has almost no iron. Its team reanalysed the star's spectrum with corrections for three-dimensional and non-LTE effects, and argued that one low-energy supernova made the metals found in it. The article uses it for the record-low iron limit. ↩
62.
Frebel, Anna, and John E. Norris. “Near-Field Cosmology with Extremely Metal-Poor Stars.” Annual Review of Astronomy and Astrophysics 53:631–688. 2015. A review article on the most iron-poor stars known and what they show about the earliest stars. It lists the seven most iron-poor stars in a table and discusses why the first stars were probably very massive. The article uses it for that table, and for the absence of the first generation of stars.
1. 62a.
  Table 1, page 32: The table is titled "The seven most Fe-poor stars." SMSS J031300.36−670839.3 at [Fe/H] < −7.30 (Keller et al. 2014); HE 1327−2326 at −5.66; HE 0107−5240 at −5.39; HE 0557−4840 at −4.81; SDSS J1742+2531 at −4.8; SDSS J102915+172927 at −4.73; HE 0233−0343 at −4.68. Six of the seven are enormously carbon-rich, with [C/Fe] from +1.65 to above +4.9. The exception, SDSS J102915+172927 with [C/Fe] < +0.9, the authors flag as a "critical challenge." A second carbon-poor case has since been added: Starkenburg et al. 2018 (MNRAS 481:3838, arXiv:1807.04292v1) report Pristine 221.8781+9.7844 at [Fe/H] = −4.66 ± 0.13 with "an unusually low carbon abundance," and conclude "The existence of a class of ultra metal-poor stars with low(er) carbon abundances suggest that there must have been several formation channels in the early Universe." “critical challenge”. “an unusually low carbon abundance”. “The existence of a class of ultra metal-poor stars with low(er) carbon abundances suggest that there must have been several formation channels in the early Universe.” Section 5.1.1, pages 64–65: The reason given for the absence of Population III stars. “Consensus exists that due to the lack of cooling agents in primordial gas (i.e., metals or dust), significant fragmentation of the available gas was largely suppressed so that these first objects must have been very massive.” Section 5.2, page 72: The critical-metallicity values. “Zcrit ∼ 10⁻³·⁵ Z⊙ for the line cooling (Bromm & Loeb 2003) and Zcrit ∼ 10⁻⁵ Z⊙ for dust cooling.” ↩
63.
Venditti, Alessandra et al. How can we finally see the first light? Status and perspective in the search for Population III stars. Aug 19, 2026. Section Abstract: “Finding the first (Population III or Pop III) stars is one of the fundamental quests of astronomy … Yet their initial mass function, formation sites and feedback remain highly uncertain, as well as the timing and topology of the transition to metal-enriched star formation. The observability of their peculiar spectral features is also debated, due to their short lifetime and faintness. … The combination of JWST spectroscopy, time-domain searches, lensing surveys, stellar archaeology, absorption-line studies, as well as improved simulations, is yielding a growing number of observational candidates and narrowing the allowed parameter space for the first stars, setting the stage for a 'golden era' of Pop III searches.” A growing number of candidates, not a confirmation, and the article states it in exactly those terms. A 2026 review of the search for the first stars, Population III, which formed from primordial gas with almost no metals. It summarises the theory, the main observing strategies, and what has been found so far, which is candidates rather than confirmed detections. The article uses it for that status: the search is yielding a growing number of candidates, not confirmations. Mirror: arXiv ↩
64.
Visbal, Eli et al. LAP1-B is the first observed system consistent with theoretical predictions for Population III stars. 2025. Section 1: “Up to this point, there have been no unambiguous direct detections of Population III (Pop III) stars, defined by their extremely low metallicities.” “the first object to agree with three key theoretical predictions for Pop III stars: (1) formation in extremely low-metallicity halos with virial temperatures ranging from Tvir ∼ 10³ − 10⁴ K, (2) a top-heavy initial mass function, and (3) formation of low-mass clusters with ∼a few × 1000 M⊙ in massive Pop III stars.” A 2025 paper that argues LAP1-B, a z = 6.6 object lensed by the galaxy cluster MACS J0416, is the first object to fit three theoretical predictions for Population III stars, the first stars. Its authors say it is consistent with a cluster of a few thousand solar masses of such stars. The article uses it for the statement that there have been no unambiguous direct detections of Population III stars, and a companion article on star formation states the same conclusion. Mirror: arXiv ↩
65.
Curti, Mirko et al. JADES: Insights on the low-mass end of the mass – metallicity – star-formation rate relation at 3 < z < 10 from deep JWST/NIRSpec spectroscopy. 2024. Section Abstract: “We find evidence for a shallower slope at the low-mass-end of the mass-metallicity relation (MZR), with 12 + log(O/H) = (7.72±0.02) + (0.17±0.03) log(M⋆/10⁸ M⊙) … The evolution in the normalisation is observed to be relatively mild compared to previous determinations of the MZR at z ∼ 3 (∼ 0.1 − 0.2 dex across the explored mass regime). We observe a deviation from the local fundamental metallicity relation (FMR) for our sample at high redshift, especially at z > 6, with galaxies significantly less enriched (with a median offset in log(O/H) of ∼ 0.5 dex, significant at ∼ 5σ) than predicted given their M⋆ and SFR.” Nakajima et al. (arXiv:2301.12825) reach a compatible conclusion: “there is a small evolution of the MZ relation from z ∼ 2 − 3 to z = 4 − 10.” The sharp measurement is the gas-weighted metal content of the universe, which the damped Lyman-alpha result gives. A JWST spectroscopy study from the JADES team of how a galaxy's metal content depends on its mass and star-formation rate, for galaxies at redshift 3 to 10 down to low stellar masses. It finds a shallower slope at low mass, and at fixed mass galaxies at redshift 4 to 10 are only mildly less enriched than those at redshift 2 to 3. The article uses it to avoid overstating the metal-poor early universe, because at fixed mass the relation barely moves. ↩
66.
Cupps, Vernon R. “Stellar Nucleosynthesis: Where Did Heavy Elements Come From?” Acts & Facts, Institute for Creation Research. Dec 29, 2017. “a paradigm based on the biblical narrative rests on a universe purposefully designed by a Creator in which Earth was created four days before the sun, moon, and stars (Genesis 1:1-2, 14-16). Thus, any paradigm based on the veracity of this account does not allow for any elements on Earth to have come from stars.” His companion article, "Did Heavy Elements Come from Supernovas?", Acts & Facts 46(9):9 (2017), https://www.icr.org/article/10185/, adds: “It's clear in this narrative that Earth's elements were created by God on the second day before any stars came into existence”. “Since there is no definitive evidence that any of the elements with more protons than ⁶²Ni were formed in the interiors of supernovas, the deep-time paradigm of heavy element creation is a matter of belief, not fact.” His strongest sentence, which he concedes in the text: “there is no way observationally to determine whether any heavy nuclei observed are primordial (i.e., original) or manufactured later.” A popular article by Vernon Cupps arguing that heavy elements cannot have come from stars, and that God made them directly on Earth. Cupps holds a PhD in nuclear physics from Indiana University and worked at Fermilab, which is why his arguments are worth engaging in detail. The article uses it for the direct-creation option. ↩
1. 66a.
  Cupps's arithmetic, in his own words, from the same article: “In order for two ⁵⁶Fe nuclei to fuse, one of the nuclei must have an energy of at least 91 MeV … in order to overcome the coulomb barrier between them. … 91 MeV corresponds to a star temperature of approximately 1.06 × 10¹² °K … The hottest stars measured to date are the blue hypergiants such as Eta Carinae with temperatures in the neighborhood of 4 × 10⁴ °K, approximately seven orders of magnitude less than the energy required for such fusion reactions to occur. … Clearly, it is not possible for the heavy elements to form in known stable stars and nebula.” The arithmetic is correct, and it applies to a process that no one proposes. His neutron-supply question is: “Where do all the neutrons needed for these two methods come from?” It is answered in Kobayashi, Karakas and Lugaro 2020 (arXiv:2008.04660v1, page 2), which names ¹³C(α,n)¹⁶O as the main s-process neutron source in the helium intershell of AGB stars, and ²²Ne(α,n)²⁵Mg as the weak-component source in massive stars. ↩
67.
Henry, Jonathan. “The elements of the universe point to creation: Introduction to a critique of nucleosynthesis theory.” Journal of Creation 20(2):53–60. Aug 2006. Section What is the creation alternative to NST?: “It is possible to conclude that the 'finished' state of creation included to a large degree the present suite of stable isotopes, without the need for nucleosynthesis to account for them. With respect to origins, this is the creation alternative to NST.” The explicit rejection of accelerated nucleosynthesis, in the same section: “A mistaken alternative is to assume that naturalistic processes can be reconciled with fiat creation by shortening the timescale to fit within a literal Creation Week. A naturalistic process impossible over eons is less likely over days, and to say that God accomplished the naturalistic process quickly is to verge on a kind of 'theistic naturalism'.” Henry and Humphreys are on opposite sides of this within the young-earth camp. Humphreys' 1994 white hole cosmology proposed rapid primordial nucleosynthesis, which is exactly the option Henry calls theistic naturalism. An article in the Journal of Creation, the peer-reviewed journal of Creation Ministries International, that argues the elements point to creation. It focuses on the claim that big bang nucleosynthesis predicted the cosmic hydrogen-to-helium abundance and the temperature of the microwave background. The article uses it because it rejects the idea that naturalistic element-making could be squeezed into a literal Creation Week. ↩
68.
Faulkner, Danny R. “A Discussion of Stellar Nucleosynthesis.” Answers Research Journal 7:147–159. 2014. A discussion paper in the creation literature on how the chemical elements were made. Its author says the creationist attempts to explain the abundances are all preliminary, and argues that nucleosynthesis theory explains the abundances without predicting them. The article uses it because the author calls the creationist attempts "at best preliminary" and says that the theory's power "is entirely explanatory".
1. 68a.
  Section Conclusion, pages 158–159: “Can we develop a creationary model to explain the chemical composition of the universe? Perhaps. Much of the Creation Week was miraculous, or at least outside of the realm that science today is equipped to probe, so one could posit that God created the elemental abundances pretty much as they exist today. However, many will find such an answer unsatisfactory, particularly if there is no clear standard of design of the elemental composition. On the other hand, God could have used physical processes as we now understand them. … Of course, in a biblical framework, this must have happened very rapidly, far more rapidly than the gradual processes described here. There have been a few initial attempts to account for chemical abundances in the creation literature in such a physical way. In the initial concept of his white hole cosmology, Humphreys (1994) proposed rapid primordial nucleosynthesis of elements. Boudreaux (2012) and Davies (2013) have proposed similar things. However, all of these suggestions are at best preliminary. I encourage further development of ideas of this type.” ↩
2. 68b.
  “A good theory must have good explanatory power. … However, the fact that there is variation in composition within the solar system and among stars clearly indicates that this is not an exact science. Any variation is explained in terms of local differences in history … once the process of explaining differences begins, virtually any differences can be explained, all the while defending the basic paradigm. These sorts of musings eventually take on the characteristics of just-so stories that we so often encounter in evolutionary explanations. A theory that explains anything and everything explains nothing.” “But in addition to having explanatory power, a good theory also must have predictive power. … It does not appear that the detailed theory of primordial and stellar nucleosynthesis makes such predictions. That is, the power of nucleosynthesis theory in astronomy is entirely explanatory.” The reference list includes Boyd et al. 2012, "The r-process in metal-poor stars and black hole formation," ApJL 744:L14, so the relevant literature was in hand. The paper does not engage it as a chronometer, and does not discuss metal-poor halo stars, r-process abundance patterns, the [α/Fe] knee or the metallicity floor anywhere in the paper. ↩
69.
Matteucci, Francesca. “Modelling the chemical evolution of the Milky Way.” Astronomy and Astrophysics Review. 2021. A review of how models of the Milky Way's chemical history are built and tested, including the time-delay model that follows how long each kind of star takes to pollute the gas. The article uses it for two cases in which such a model predicted a measurement before it was made.
1. 69a.
  Page 3: “This allowed the authors to predict the evolution of several [α/Fe] ratios (α=O, Si, Mg) resulting positive and around ∼ +0.4 dex in halo stars (with [Fe/H] < −1.0 dex), and decreasing in disk stars until reaching the solar value ([Fe/H]=0 dex). At that time, there were data only for [O/Fe] in halo stars, and so the results for Si and Mg were just predictions. Those predictions were then confirmed by François (1986), who found indeed overabundances of Si and Mg relative to Fe in halo stars.” The prediction is from Matteucci and Greggio, Astronomy & Astrophysics 154 (1986): 279. Page 59: The second case, from the same review: the time-delay model predicted that the Galactic bulge, having formed fast, should knee at higher metallicity than the solar neighbourhood. Gaia-ESO (Rojas-Arriagada et al. 2017) and APOGEE (Zasowski et al. 2019) later measured a bulge knee at [Fe/H] = −0.37 ± 0.09, about 0.06 dex above the thick disk's. ↩
70.
Xiang, Maosheng, and Hans-Walter Rix. “A time-resolved picture of our Milky Way's early formation history.” Nature 603:599–603. Mar 23, 2022. Page 1: Why subgiants: “Subgiant stars, stars sustained by hydrogen shell fusion, can be unique tracers for such purposes, as they are in the brief stellar evolutionary phase that permits the most precise and direct age determination, because their luminosity is a direct measure of their age.” Section Methods: The sample size: “the remaining sample contains 247,104 stars,”. The median relative age precision, printed on the paper's Figure 1, is 7.5 per cent, across 1.5 to 13.8 Gyr. Section Abstract: “the stellar age-metallicity distribution p(τ, [Fe/H]) splits into two almost disjoint parts, separated at τ ≃ 8 Gyr”. “the formation of the Galaxy's old (thick) disk started ∼13 Gyr ago, only 0.8 Gyr after the Big Bang, and two Gigayears earlier than the final assembly of the inner Galactic halo. Most of these stars formed ∼11 Gyr ago, when the Gaia-Sausage-Enceladus satellite merged with our Galaxy.” On the tightness of the age-metallicity relation: “it reveals an exceptionally clear, continuous and tight age-metallicity relation from [Fe/H] ≲ −1 13 Gyr ago all the way to [Fe/H] = 0.5 ∼7 Gyr ago”. The fitted intrinsic age dispersion at fixed metallicity is. “less than 0.82 ± 0.01 Gyr”. across that interval, which implies metallicity scatter at fixed age below 0.22 dex. On the orbital split, cutting at Jφ = 500 kpc km/s gives. “two nearly disjoint”. distributions, the high-angular-momentum set being the tight old-disk sequence at [Fe/H] > −1 and the low or retrograde set. “predominately [Fe/H] < −1.2, reflects the halo.” On the thin disk, honestly: the young side shows. “manifest evidence for stellar radial orbit migration”. and. “a considerably higher [Fe/H] dispersion at a given age.” A 2022 paper that dates 247,104 subgiant stars, which are stars sustained by hydrogen shell fusion, using LAMOST spectra and Gaia parallaxes, and builds a timeline of the Milky Way's formation from their ages. Subgiants are used because their luminosity is a direct measure of their age, which makes their ages among the most precise available for stars this old. The article uses it for the timeline of the Galaxy's assembly that these ages give. Mirror: arXiv ↩
71.
Miglio, A. et al. “Age dissection of the Milky Way discs: Red giants in the Kepler field.” Astronomy & Astrophysics 645:A85. 2021. Section abstract, page 1: “nearly 5400 giants”. Table 1, page 4: The mean age of the chemically defined thick disk (α-rich red giants), reference run R1, is 10.98 (+0.13/−0.14) Gyr. Section §4.4, page 12: “a mean age of the high-α population in our sample of ∼ 11 Gyr, with variations depending on the modelling run of the order of 1 Gyr,”. The systematic error rather than the statistical error dominates. The intrinsic age spread is 0.76 (+0.27/−0.23) Gyr, and. “we can measure the age spread from the posterior samples, and infer that 95% of the population was born within 1.52 (+0.54/−0.46) Gyr.” Section §4.3, pages 11–12: Kinematics: “σz(10 Gyr)low [Mg/Fe] = 22.6 ± 0.6 km s⁻¹ and σz(10 Gyr)high [Mg/Fe] = 36 ± 2 km s⁻¹”. The authors read the difference as. “suggestive of a halt in the star formation (quenching) after the formation of the chemical-thick disc.” Section §4.2, page 12: They also report super-solar-metallicity stars as old as 9 Gyr at the solar radius, which can only be migrants from the inner disk. This is the messiness of the thin-disk age–metallicity relation, reported straight. A study of about 5,400 red giant stars with Kepler light curves and APOGEE spectra, whose masses and ages come from their asteroseismic oscillations. It finds the chemically thick disk to be about 11 billion years old, with most of its stars born within about 1.5 billion years of each other. The sample is about 5,400 giants, not tens of thousands. The article uses it as an asteroseismic age for the thick disk, and these ages come from asteroseismology, not isochrone placement on a colour-magnitude diagram, so they are methodologically independent of Xiang & Rix 2022. ↩
72.
Haywood, Misha et al. “The age structure of stellar populations in the solar vicinity: Clues of a two-phase formation history of the Milky Way disk.” Astronomy & Astrophysics 560:A109. 2013. Section Abstract: “there are two distinct regimes of [α/Fe] versus age which we identify as the epochs of the thick and thin disk formation. A tight correlation between metallicity and [α/Fe] versus age is clearly identifiable for thick disk stars, implying that this population formed from a well mixed interstellar medium, probably initially in starburst and then more quiescently, over a time scale of 4-5 Gyr … We suggest that the youngest thick disk set the initial conditions from which the inner thin disk started to form about 8 Gyr ago.” Their 4–5 Gyr thick-disk formation timescale disagrees with Miglio's 1.52 (+0.54/−0.46) Gyr. The disagreement is partly definitional and partly a real open question. The ordering and the scale, with the thick disk before the thin disk and the thin disk starting about 8 Gyr ago, are common to all three studies compared in the article. This is a genuinely different sample and method from Xiang & Rix 2022 and Miglio et al. 2021. A paper that analyses solar-neighbourhood stars with high-quality chemical measurements, plotting the ratio of alpha elements to iron against age, and that finds two distinct regimes which the authors identify as the formation of the thick disk and then the thin disk. The stars are dated by isochrones, with no asteroseismology. The article uses it because it gives a thick-disk formation timescale of 4 to 5 billion years, rather than 1.5. ↩
73.
Helmi, Amina et al. “The merger that led to the formation of the Milky Way's inner stellar halo and thick disk.” Nature 563:85–88. Oct 31, 2018. Their isochrone fit to the debris gives: “an age range from 10 to 13 Gyr is compatible with the HRD of Gaia-Enceladus.” Section Abstract: “We demonstrate that the inner halo is dominated by debris from an object which at infall was slightly more massive than the Small Magellanic Cloud, and which we refer to as Gaia-Enceladus. … Hundreds of RR Lyrae stars and thirteen globular clusters following a consistent age-metallicity relation can be associated to Gaia-Enceladus on the basis of their orbits. With an estimated 4:1 mass-ratio, the merger with Gaia-Enceladus must have led to the dynamical heating of the precursor of the Galactic thick disk and therefore contributed to the formation of this component approximately 10 Gyr ago.” Page 4: Their dating argument: “the progenitor of the Galactic thick disk was in place when Gaia-Enceladus fell in, which based on the ages of its youngest stars, would suggest that the merger took place around 10 Gyr ago, i.e. at redshift z ∼ 1.8.” A Nature paper that names and dates the merger which built much of the Milky Way's inner halo. It names the accreted object Gaia-Enceladus and argues that the inner halo is dominated by its debris, from an object slightly more massive than the Small Magellanic Cloud. The article uses it for the 10 Gyr date for the merger, which the authors base on the ages of the youngest thick-disk stars. ↩
74.
Belokurov, V. et al. “Co-formation of the disc and the stellar halo.” Monthly Notices of the Royal Astronomical Society 478:611–619. 2018. Section Abstract: At the low-metallicity end of the halo the anisotropy is: “mildly radial with 0.2 < β < 0.4”. For stars more metal-rich than [Fe/H] = −1.7 the anisotropy is: “for stars with [Fe/H] > −1.7 we measure extreme values of β ∼ 0.9.” The authors' conclusion: “the observed acute anisotropy is inconsistent with the continuous accretion of dwarf satellites. Instead, we argue, the stellar debris in the inner halo were deposited in a major accretion event by a satellite with Mvir > 10¹⁰ M⊙ around the epoch of the Galactic disc formation, i.e. between 8 and 11 Gyr ago.” The halo's mean spin is given as 20 < v̄θ (km s⁻¹) < 30. A study of the Milky Way's stellar halo, the sphere of old stars around the galaxy. It used Gaia and SDSS measurements of stars within about 10 kpc of the Sun and found that the more metal-rich halo stars move almost purely radially, in and out rather than around. Its authors argue that one massive satellite galaxy, deposited 8 to 11 billion years ago, built the inner halo. The article uses it because one radial merger, not a steady rain of small satellites, produces this kind of motion. ↩
75.
Gallart, Carme et al. “Uncovering the birth of the Milky Way through accurate stellar ages with Gaia.” Nature Astronomy 3:932–939. Jul 22, 2019. Section Abstract: “We show that the stars in both halo sequences share identical age distributions, and are older than the bulk of thick disc stars. The sharp halo age cut 10 Gyr ago can be identified with the accretion of Gaia-Enceladus.” A research paper in Nature Astronomy by Carme Gallart and six co-authors. They date the Milky Way's halo and thick-disc stars by fitting full colour-magnitude diagrams, a technique developed for resolved nearby galaxies that Gaia's distances made possible for the Milky Way. The article uses it because it dates the merger with Gaia-Enceladus by a method that differs in kind from the isochrone comparison used by Helmi and colleagues. ↩
76.
Montalbán, Josefina et al. “Chronologically dating the early assembly of the Milky Way.” Nature Astronomy 5:640–647. May 17, 2021. Page 3: Ninety-five red-giant-branch stars survive their cuts, each with individually measured radial-mode oscillation frequencies fitted in the AIMS Bayesian code, at an average age precision of 11 per cent. Page 8: Their table, preprint p. 8, reports mean ages by group: accreted debris (low [Mg/Fe], eccentricity above 0.7), 9.7 ± 0.6 Gyr with an intrinsic spread of 0.8 (+0.9/−0.4) Gyr; low [Mg/Fe] with eccentricity below 0.7, 8.2 ± 0.8 Gyr; and the in-situ high-[Mg/Fe] thick disk, 10.4 ± 0.3 Gyr with a spread of 0.5 (+0.4/−0.3) Gyr. Section conclusion, page 8: Their conclusion, p. 8, is that the high-[α/Fe] population: “was formed in the Milky Way starting roughly ∼ 10 to 11.5 Gyr ago”. and. “a majority of the stars in the high-[α/Fe] population was in place before the merger with the GES progenitor.” This cuts against the earlier story that Gaia-Enceladus created the thick disk by heating it. The article reports the disagreement rather than smoothing it. A research paper that dates the early assembly of the Milky Way from 95 red giant stars whose individual oscillation frequencies were measured, giving each age to about 11 per cent precision. It separates stars born inside the Galaxy from stars that came in from the accreted Gaia-Enceladus dwarf galaxy. The article uses it as the asteroseismic dating of that merger. ↩
77.
Bensby, T. et al. “Chemical evolution of the Galactic bulge as traced by microlensed dwarf and subgiant stars VI.” Astronomy & Astrophysics 605:A89. 2017. Section Abstract: The abstract reports a. “high fraction of intermediate-age to young stars where at [Fe/H] > 0 more than 35 % are younger than 8 Gyr, and for [Fe/H] ≲ −0.5 most stars are 10 Gyr or older”. “several episodes of significant star formation in the bulge has been identified: 3, 6, 8, and 11 Gyr ago”. The authors' own reading of the bulge follows: “our results strengthen the observational evidence that support the idea of a secular origin for the Galactic bulge”. A study of 90 F and G dwarf, turnoff and subgiant stars in the central bulge of the Milky Way. Their spectra were taken during gravitational microlensing events, the only practical route to dwarf-star spectra at that distance. The authors measured the ages and abundances of 11 elements and found several bursts of star formation. The article uses it to show that the bulge is not uniformly old, while the metal-poor stars inside the bulge are still the oldest. ↩
78.
Sestito, Federico et al. “Tracing the formation of the Milky Way through ultra metal-poor stars.” Monthly Notices of the Royal Astronomical Society 484:2166–2180. 2019. The paper covers all 42 known stars with [Fe/H] below −4 dex, with Gaia DR2 orbits. Its key finding, in the abstract's words: “we find evidence that a significant fraction of all known UMP stars (∼ 26%) are on prograde orbits confined within 3 kpc of the Milky Way plane.” A journal paper that uses Gaia satellite distances and orbits to trace the paths of all 42 known ultra metal-poor stars, stars with less than a ten-thousandth of the Sun's iron. The authors assume these stars are old, over 11 billion years, and belong to the halo, the galaxy's outer cloud of stars, and they find that about 26 per cent move the same way the galaxy turns, on orbits that stay within 3 kpc (kiloparsecs, thousands of light years) of its plane. The article uses it to show that the very metal-poor stars are not confined to the halo. Mirror: arXiv ↩
79.
Sestito, Federico et al. “The Pristine survey X: a large population of low-metallicity stars permeates the Galactic disk.” Monthly Notices of the Royal Astronomical Society 497:L7–L12. 2020. The abstract: “The common picture is that these low-metallicity stars are distributed as an isotropic, pressure-supported component … we show that this picture is incomplete. This sample shows strong statistical evidence (at the 5.0σ level) of asymmetry in their kinematics, favouring prograde motion. Moreover, we find that 31% of the stars that currently reside in the disk do not venture outside of the disk plane throughout their orbit.” Their conclusion, which the article calls a change to the story and not a removal of it: “a significant fraction of stars with iron abundances [Fe/H] ≤ −2.5 formed within or concurrently with the Milky Way disk”. A short paper in the Pristine survey series, which hunts for the most metal-poor stars by their colours. It uses Pristine photometry and LAMOST spectra for very metal-poor stars, those with iron at or below [Fe/H] = −2.5, which is under a few thousandths of the Sun's iron. It finds that these stars do not form only a round halo, and the article uses its conclusion that the story has changed rather than been removed. Mirror: arXiv ↩
80.
Silva Aguirre, V. et al. “Confirming chemical clocks: asteroseismic age dissection of the Milky Way disk(s).” Monthly Notices of the Royal Astronomical Society 475:5487–5500. 2018. A 2018 paper that dates red giant stars in the Milky Way's disk from their vibrations (asteroseismology), using Kepler satellite data, and so splits the disk into its older and younger parts by age. It checks the chemical clocks of the disk, the element ratios that are meant to track age, against these measured ages. The article uses it for its finding that stars which look chemically ancient can turn out to be seismically young, and for the test that tells the two explanations apart. Mirror: arXiv
1. 80a.
  On the 'young α-rich' stars, the authors conclude from kinematics that these are: “old stars that went through a mass transfer or stellar merger event, making them appear younger, instead of migration of truly young stars formed close to the Galactic bar.” Miglio et al. 2021 find the same class of stars (abstract, p. 1) at: “of the order of 5% on the RGB, and significantly higher in the RC”. The red-clump excess is the giveaway, since a merger product should preferentially show up after core-helium ignition. This is the right kind of anomaly to report: it looked like a violation of the age ordering, it was chased down, and the resolution was independently testable. ↩
81.
Madau, Piero, and Mark Dickinson. “Cosmic Star-Formation History.” Annual Review of Astronomy and Astrophysics 52:415–486. 2014. Page 1: “the star-formation rate density peaked approximately 3.5 Gyr after the Big Bang, at z ≈ 1.9, and declined exponentially at later times, with an e-folding timescale of 3.9 Gyr. Half of the stellar mass observed today was formed before a redshift z = 1.3. About 25% formed before the peak of the cosmic star-formation rate density, and another 25% formed after z = 0.7. Less than ∼ 1% of today's stars formed during the epoch of reionization.” Page 48: Their fitted form, Equation 15, is ψ(z) = 0.015 (1+z)^2.7 / [1 + ((1+z)/2.9)^5.6] M☉ yr⁻¹ Mpc⁻³. Section abstract: “The rise of the mean metallicity of the Universe to about 0.001 solar by z = 6, one Gyr after the Big Bang.” A review of how star formation, heavy-element production and reionization have changed across cosmic time, built from many kinds of survey. It finds that the rate of star formation per unit volume peaked about 3.5 billion years after the Big Bang and fell exponentially afterwards. The article uses it for the shape of the star-formation history: a rise, a peak and a tenfold decline. ↩
82.
van der Wel, A. et al. “3D-HST+CANDELS: the evolution of the galaxy size-mass distribution since z = 3.” Astrophysical Journal 788:28. 2014. Section Abstract: “we find a significantly different rate of average size evolution at fixed galaxy mass, with fast evolution for the early-type population, Reff ∝ (1 + z)^−1.48, and moderate evolution for the late-type population, Reff ∝ (1 + z)^−0.75”. “Reff ∝ M^0.22 for late-type galaxies … Reff ∝ M^0.75 for early-type galaxies”. The size-mass slope is stable, and the scatter is small: “intrinsic scatter ≲ 0.2 dex for all galaxy types and redshifts.” The sharpest single statement: “The number density of massive (∼ 10¹¹ M⊙), compact (Reff < 2 kpc) early-type galaxies increases from z = 3 to z = 1.5 − 2 and then strongly decreases at later cosmic times.” Conselice 2014 (arXiv:1403.2783) puts the same result in one line: sizes are. “up to a factor of 2-5 smaller at high redshift at a given stellar mass.” A 2014 paper that measures how big galaxies were at each epoch back to redshift 3, using spectroscopic and photometric data from the 3D-HST survey and imaging from CANDELS. It separates galaxies that have stopped forming stars (early types) from those still forming them (late types), and finds that early types grew faster in size over time. The article uses it for the claim that galaxies were smaller at the same mass in the past, and for the rise and fall of compact massive early-type galaxies. Mirror: arXiv ↩
83.
Conselice, Christopher J. “The Evolution of Galaxy Structure Over Cosmic Time.” Annual Review of Astronomy and Astrophysics 52:291–337. 2014. Section Abstract: “from small, compact and peculiar systems in the distant universe to the formation of the Hubble sequence dominated by spirals and ellipticals we find today.” The local merger fraction, in the preprint: Page 17: Preprint page. “the nearby merger fraction gives values of fm = 0.01.” The merger history, in the preprint: Page 31: “The merger history tends to peak at z ∼ 2.5 for massive galaxies with M* > 10¹⁰ M⊙ at values of fm ∼ 0.3 − 0.4 and decline at lower redshifts.” His caution about the fitted exponent, on the same pages: “The values for m found in the literature can vary significantly, but most of this is due to different selections, different redshift ranges used, and the use of various values for the local merger fraction.” A review of how galaxy shapes, sizes and structures have changed with cosmic time, from the first galaxies visible at redshift about 6 to the galaxies of today. It covers the measuring methods, including how mergers are identified, and what the merger history shows. The article uses it for merger rates: about one nearby galaxy in a hundred is caught in a merger, and for massive galaxies the fraction is 0.3 to 0.4 at redshift 2.5. ↩
84.
Muzzin, Adam et al. “The Evolution of the Stellar Mass Functions of Star-Forming and Quiescent Galaxies to z = 4 from the COSMOS/UltraVISTA Survey.” Astrophysical Journal 777:18. 2013. Section abstract: “the stellar mass density of the universe was only 50%, 10% and 1% of its current value at z ∼ 0.75, 2.0, and 3.5, respectively. The quiescent population drives most of the overall growth, with the stellar mass density of these galaxies increasing as ρstar ∝ (1 + z)^−4.7±0.4 since z = 3.5, whereas the mass density of star-forming galaxies increases as ρstar ∝ (1 + z)^−2.3±0.2. At z > 2.5, star-forming galaxies dominate the total SMF at all stellar masses, although a nonzero population of quiescent galaxies persists to z = 4.” To be precise, the survey stops at z = 4, and quiescent galaxies are still present there, though rare. The honest statement is not that there are none beyond z = 4, but that the quiescent mass density falls about twice as steeply as the star-forming density, so by z ≈ 3 the universe is overwhelmingly star-forming. JWST has since found individual quiescent galaxies at z ≈ 7, which is a real anomaly and is reported in the same part of the article. A survey of 95,675 galaxies over 1.62 square degrees of the COSMOS/UltraVISTA field, sorted into star-forming and quiescent (no longer forming stars) galaxies, which measures how many galaxies of each stellar mass there are at each redshift. It finds that the stellar mass density of the universe was only 50, 10 and 1 per cent of today's value at redshifts 0.75, 2.0 and 3.5. The article uses it for the thinning-out of quiescent galaxies with distance. ↩
85.
Labbé, Ivo et al. “A population of red candidate massive galaxies ~600 Myr after the Big Bang.” Nature 616:266–269. 2023. Section abstract: “If verified with spectroscopy…” The arXiv version of 25 July 2022 was titled "A very early onset of massive galaxy formation", and its abstract said "galaxies with masses M* ∼ 5 × 10⁹ M⊙ may be found out to redshifts as high as z ∼ 18." That sentence is absent from the published version. Seven galaxies became six, and 7 < z < 11 became 7.4 < z < 9.1. “We note that these masses and redshifts are not definitive and that all galaxies should be [considered candidates]”. “All objects in this paper should be considered candidate massive [galaxies, to be confirmed with spectroscopy].” Four of the six have spectra: 13050 is CEERS 3210 at spectroscopic z = 5.624, a broad-line AGN; 35300 is CEERS3-1748 at z = 7.769 with a stellar mass of 3.1 × 10⁹ M☉; 38094 is RUBIES-EGS-55604 at z = 6.982 with broad Hβ; 14924 is RUBIES-EGS-966323 at z = 8.353. All four came in below the photometric redshift, and none of the six is robustly confirmed above 10¹⁰ M☉. A research paper that reported six candidate massive galaxies about 600 million years after the Big Bang, and said plainly that they were not yet confirmed. The article uses the gap between its July 2022 arXiv version and the published paper to show how far the claim was walked back. ↩
86.
Curtis-Lake, Emma et al. “Spectroscopic confirmation of four metal-poor galaxies at z = 10.3–13.2.” Nature Astronomy 7:622–632. 2023. Section Abstract: “The spectra reveal that these primeval galaxies are metal poor, have masses between of order ~10⁷-10⁸ solar masses, and young ages.” A JWST spectroscopy paper that confirmed four galaxies at redshifts 10.3 to 13.2, the first galaxies above redshift 10 to get real spectra. They turned out small, metal-poor and young, as the standard picture predicts, and this result is routinely omitted from creationist retellings of the JWST story. The article sets it against the Labbé candidates, which had not been spectroscopically confirmed. ↩
87.
Carniani, Stefano et al. “Spectroscopic confirmation of two luminous galaxies at a redshift of 14.” Nature 633:318–322. 2024. In the Methods section, Stellar population properties, the lensing-corrected stellar mass of JADES-GS-z14-0 is log₁₀(M/M☉) = 8.6 (+0.7/−0.2), with a star-formation rate of 19 ± 6 M☉/yr, a gas-phase metallicity of log(Z/Z☉) = −1.5 (+0.7/−0.4) and a radius of 260 pc. The value log(M/M☉) = 8.7 ± 0.1 that appears in the same section belongs to a foreground lensing galaxy, ID 183349, not to JADES-GS-z14-0. “This discovery proves that luminous galaxies were already in place 300 million years after the Big Bang and are more common than what was expected before JWST”. “the excess of luminous galaxies in the early Universe cannot be entirely explained by accretion onto black holes. Galaxy formation models will need to address the existence of such large and luminous galaxies so early in cosmic history.” A JWST spectroscopy paper from the JADES survey team that confirmed the distances of two galaxies from their spectra. JADES-GS-z14-0 is at redshift 14.32 (+0.08/−0.20) and JADES-GS-z14-1 at redshift 13.90 ± 0.17. The article uses it to show that the more massive of the two has a few hundred million solar masses, not 10¹¹. ↩
88.
Wang, Tao et al. “JWST/MIRI reveals the true number density of massive galaxies in the early Universe.” Astrophysical Journal Letters 988:L35. 2025. A 2025 letter by Tao Wang and colleagues that adds JWST's mid-infrared camera, MIRI, to the near-infrared data when measuring the stellar masses of distant galaxies. Without mid-infrared photometry, stellar masses above 10¹⁰ M☉ are overestimated by about 0.4 dex, and contamination of 'massive galaxy' samples by little red dots falls from 32 per cent to 13 per cent once MIRI is included. The article uses it to show that the masses came down once the models were fixed. A naming caution for anyone chasing this citation: this is Tao Wang, not 'Williams et al.', and it must not be confused with Bingjie Wang et al. 2024 (arXiv:2405.01473), the RUBIES spectroscopy paper that re-measured two of the Labbé candidates. Mirror: arXiv ↩
89.
Kocevski, Dale D. et al. “The Rise of Faint, Red AGN at z > 4: A Sample of Little Red Dots in the JWST Extragalactic Legacy Fields.” Astrophysical Journal 986:126. 2025. Of the sources brighter than F444W = 26.5, 71 per cent are confirmed AGN; that 71 per cent applies to the brighter subset of the 341, not to all of them. The class was named by Matthee et al. 2024 (ApJ 963: 129, arXiv:2306.05448), and Greene et al. 2024 (arXiv:2309.05714) found roughly 60 per cent of compact red UNCOVER sources have broad Hα and about 20 per cent are brown dwarfs. An observational paper that identifies 341 "little red dots", compact red objects that JWST sees at redshifts 2 to 11, and tests what they are. It identifies many of them as accreting black holes (active galactic nuclei, or AGN) where it can confirm that. The article uses it because if the light comes from an accreting black hole rather than from stars, the inferred stellar mass, and the "too many stars too early" claim built on it, evaporates. ↩
90.
Weibel, Andrea et al. RUBIES: a complete census of the bright and red distant Universe with JWST/NIRSpec. 2025. A 2025 paper on the massive quiescent galaxy RUBIES-UDS-QG-z7, a galaxy that has stopped forming stars, at redshift 7.3. Its number density is more than a hundred times what any model predicts. The article names it as the cleanest single unresolved anomaly in the current JWST literature rather than burying it. ↩
91.
Boylan-Kolchin, Michael. “Stress testing ΛCDM with high-redshift galaxy candidates.” Nature Astronomy 7:731. 2023. A short paper that tests the standard cosmological model against the early galaxy candidates from JWST. It sets a pre-registered threshold: a star-formation efficiency of at least ε ≈ 0.57 would be required for the candidates to constitute a genuine problem. The measured efficiency, with mid-infrared photometry included, is about 0.3, which is inside the bound. The methodological point matters more than the number, because the bound was published before the measurement. The article uses it as a clean test of the early candidates. ↩
92.
Chworowsky, Katherine et al. “Evidence for a Shallow Evolution in the Volume Densities of Massive Galaxies at z = 4-8 from CEERS.” Astronomical Journal 168:113. 2024. “not in tension with modern cosmology”. Two later papers reach compatible conclusions. Prada et al. 2026 (arXiv:2604.18683) report that. “earlier claims of insufficient stellar masses at z = 8 are inconsistent with our model and are likely driven by systematic uncertainties, including AGN contamination, dust attenuation, and the lack of JWST/MIRI constraints”. and McLeod et al. 2026 (arXiv:2604.16666) find that the ultraviolet luminosity function falls fast above z = 12, with the data. “entirely consistent with a relatively simple galaxy evolution model.” A study of massive galaxies, those with more than 10^10 solar masses, at redshifts 4 to 8 in the JWST CEERS survey. It counted them by fitting their light to stellar-mass models, and compared the counts with a simple model in which dark-matter halos turn gas into stars at a constant efficiency. At higher redshifts the counts exceed that simple model, but modest changes to star formation can explain the excess. The article uses it as the paper that finds these counts consistent with modern cosmology. ↩
93.
Repp, A. “Fifty-Seven Years of Creation Astronomy, Part II: Issues and Advances.” Creation Research Society Quarterly 59(2):81–91. 2022. A survey of fifty-seven years of creation astronomy in the Creation Research Society Quarterly, in its Fall 2022 issue. Its discussion of star formation concludes that the question is unsettled among creationists. The article uses it for that verdict and for its silence on the other stellar-age methods.
1. 93a.
  Page 87: “Rejection of the physical arguments for star formation should logically entail (1) straightforward order-of-magnitude calculations demonstrating that super-Jeans molecular clouds will not collapse and (2) at least the outlines of a self-consistent physical stellar development model (that does not also allow star formation). This bar seems extremely high. On the other hand, acceptance of star formation almost forces consideration of how stellar development processes — with their Gyr.-timescales — fit into a YEC framework. It seems safe to say that there is no consensus on this issue.” On where model-building stands, p. 87: “pointing out difficulties in naturalistic origins scenarios — while necessary and valuable work — falls short of actually synthesizing observations within a Biblical framework.” Page 86: The survey's opening statement of the difficulty reads: “Stellar physics is a third unresolved area. On the one hand, the standard model of stellar development reflects seemingly well-understood physics (noted by Faulkner and DeYoung, 1991). However, this physics also predicts the spontaneous formation of stars when gas clouds reach a critical density. Many creationists reject the latter, and thus they are open to skepticism about the former.” The survey has no entry for main-sequence turnoff ages, globular cluster ages, white dwarf cooling or nucleocosmochronology. The stellar-age question appears only as the star-formation dispute. ↩
2. 93b.
  Page 87: Repp 2022, in the CRSQ: “Other objections seem nothing more than apologetic tactics — for if we disallow star formation, the relatively short lifetime (~ 10⁶ yrs.) of hot blue stars precludes a 13-Gyr. age for the Universe. However, the physical arguments predicting star formation seem as solid as those predicting such lifetimes; if so, it is inconsistent to accept the latter but reject the former. It has also been noted (e.g., Faulkner, 2001a) that star-formation triggers (such as supernova shocks and radiative cooling from dust) require the existence of a previous generation of stars — and thus, even if we accept ongoing star formation, the origin of first-generation stars remains problematic for the naturalist. More to the point, the apologetic potential of an argument is no measure of its validity. If we reject the conclusions of mainstream science, we should do so on the basis of solid exegesis, not dubious inference.” On pp. 86–87 the survey disposes of the two standard creationist objections to star formation, noting that Mulfinger's thermodynamic argument "would disallow any radiative cooling whatsoever and in fact neglects the entropy increase of the environment," and that the scriptural objections come with "Little if any exegesis." On p. 86 it lists approvingly the arguments creationists have already had to give up, including Faulkner's debunking of "the use of supernova remnants as indicators of a young galaxy." ↩
94.
Humphreys, D. Russell. “Accelerated Nuclear Decay: A Viable Hypothesis?” Radioisotopes and the Age of the Earth: A Young-Earth Creationist Research Initiative, 333–380. Institute for Creation Research and Creation Research Society. Edited by Larry Vardiman, Andrew A. Snelling and Eugene F. Chaffin. 2000. Chapter 7 of a young-earth creationist research volume on radioisotopes and the age of the earth. The chapter asks whether accelerated nuclear decay is a viable hypothesis, and it lists reasons God might have changed decay rates. The article uses it for the claim that the change would apply to the heavens as well as the earth.
1. 94a.
  Page 356: Arguing from 2 Peter 3:7: “Going back to the seventh verse, notice that this change in nuclei applies to the heavens as well as the earth. It was not just a local change in the vicinity of the earth; it affected the whole cosmos.” On the same page, in the list of reasons God might have changed decay rates: “since nuclear fusion rates are related to decay rates, it could increase the activity of stars.” Both halves matter. The scope claim means a critic cannot say RATE confined itself to Earth rocks, because if alpha-decay rates were globally larger, the stellar thorium and uranium chronometers are affected identically. The fusion coupling is where the argument fails. No calculation is supplied of what a factor of 10⁸ to 10⁹ in stellar reaction rates does to a star, and the chapter concedes, citing Noyes 1999 p. 66, that core changes take millions of years to reach the surface, which removes any observable signature and therefore any test. ↩
95.
Chaffin, Eugene F. “Supernova Driven Accelerated Decay Confronts the Evidence.” Creation Research Society Quarterly 57:98–112. 2020. Section Uranium Detection in Stars, pages 103–104: “If this could occur at Earth, other supernovae occurring elsewhere should affect the uranium in the surrounding stars. … one would expect on average less accelerated decay of the uranium in halo stars than on Earth, due to their probable location farther from historic supernovae. … However, uranium has been detected in a handful of halo stars, including Cayrel's star CS31082–001 (Cayrel et al., 2001; Chaffin, 2001), J09544277+5246414 (Holmbeck et al., 2018), and CS29497–004 (Hill et al., 2016).” The errors in that passage are that his log(U/Th) = −0.74 ± 0.15 for CS 31082-001 is Cayrel's 2001 value, superseded by Hill et al. 2002's −0.94 ± 0.11, that "Hill et al. (2016)" for CS 29497-004 is a different paper from the 2002 CS 31082-001 study, and that "BD+17°324" on page 104 is a typo for BD +17° 3248. Page 104: “The detection of such large amounts of uranium in these instances leads one to suspect that not as much accelerated decay occurred at those locations as on Earth, and perhaps is a step toward validation of our logic.” His concession that the chronology is unsolved, from the same paper: “It may be that a time dilation cosmology, a rapid expansion of space, or a larger speed of light prior to the Flood would be necessary to build a consistent history of the chronology of events in our galaxy. However, that is outside the scope of this paper.” A paper proposing that radioactive decay ran much faster on Earth in the past, driven by supernovae, and testing that idea against the uranium found in halo stars. It was accepted for publication on 24 November 2020. The article quotes its conclusion that less accelerated decay happened in those stars than on Earth. ↩
96.
Lisle, Jason. “Blue Stars Confirm Recent Creation.” Acts & Facts 41(9):16. Aug 31, 2012. “Based on their observed luminosity, the most massive blue stars cannot last even one million years. … None of this is a problem for the biblical timescale of about 6,000 years for the age of the universe. But if the universe were 13.7 billion years old, as secularists allege, then it really shouldn't have blue stars. Yet blue stars abound in every known spiral galaxy. It seems that these galaxies cannot be even one million years old.” The load-bearing premise is in the next paragraph, which Lisle states as: “Secular astronomers must assume that new blue stars have formed recently to replace all those that have burned out over deep time. … But it is not. Star formation has never been observed.” Lisle's stellar-lifetime numbers are conventional astrophysics and are correct. The argument stands or falls on the premise that new blue stars are not forming, and Lisle defends that premise explicitly rather than hiding it. The article's star-formation section treats the three physical obstacles Lisle offers, gas pressure, magnetic pressure and angular momentum, and finds that each fails on direct measurement. A short magazine article by Jason Lisle, published by the Institute for Creation Research, arguing that the blue stars in spiral galaxies are too short-lived for the galaxies to be old. The article uses it as the clearest version of the blue-star argument. ↩
97.
Bernitt, Rod. “Globular clusters and the challenge of blue straggler stars.” Journal of Creation 16(2):5. 2002. “Blue stragglers could simply be higher mass stars (e.g. spectral class A, main sequence) and indicators of youth”. “from a creationist perspective very little change would have occurred in the globular clusters since they were formed rapidly some 6,000 years ago”. “For example, energy supplied by gravitational collapse might be a major source of stellar radiation within a creationist model.” “This story sounds convincing until we realize that there is no way of testing it. By adding secondary explanations, the story can accommodate any astronomical observation.” The article also reports two mainstream tests that blue stragglers passed: “it was found that the blue stragglers tended to concentrate in the cores of clusters and have consistent radial velocities”. “blue stragglers are significantly different from horizontal branch stars.” Its figure for cluster ages was drawn from a 1993 textbook, and was already out of date at publication, Chaboyer et al. having printed 11.5 ± 1.3 Gyr in 1998. “they are all 13–15 billion years old”. A young-earth article on globular clusters, the dense star clusters that orbit the galaxy, and their blue straggler stars. It first appeared in August 2002 in the journal, then called TJ, and the web edition was posted 7 March 2007. The article uses it for the young-earth explanation of blue stragglers, and for an internal problem: the framework assumes that the clusters have changed very little since about 6,000 years ago. ↩
98.
Faulkner, Danny R. “How Do We Know the Age of the Universe?” Answers in Genesis. Jun 13, 2016. A short popular web article on how astronomers date the universe from its oldest objects. It says that the age of globular star clusters was adjusted to fit the age of the universe, so the two methods are not as independent as they seem. The article uses it because it states that objection in its strongest form.
1. 98a.
  “The age estimates of globular clusters come from comparing observations of their stars (what astronomers call color-magnitude diagrams) to the predicted behavior of stars over time. … The current estimate of the age of globular clusters is about 13.2 billion years, though there is uncertainty of a billion years or so. Since this age is 600 million years less than the big bang age of 13.81 billion years, this is thought to be good confirmation of the universe's age.” “Astronomers realized that a faster expansion rate would shorten the age of the universe, but astronomers also thought that globular clusters were at least 15 billion years old, if not older. If the universe were expanding much faster than generally thought at the time, then the universe would be younger than globular star clusters, which, of course, makes no sense. Once astronomers finally accepted the faster expansion rate and younger age for the universe, they had to rework their understanding of globular star clusters to improve the fit. Clearly, this adjustment would not have been made if it were not necessitated by the new estimated age of the universe. Perhaps these supposedly independent methods of age determination are not so independent after all.” ↩
99.
Chaboyer, Brian et al. “The Age Of Globular Clusters In Light Of Hipparcos: Resolving the Age Problem?” Astrophysical Journal 494:96. 1998. A revision of the age of the oldest globular clusters after new distance measurements. The authors reviewed five independent ways of setting the clusters' distance, including fitting the main sequences of nearby subdwarfs against Hipparcos parallaxes. All five indicated that the clusters are farther away than believed, which lowers the age estimate. The article uses it because the cause of the shift was a distance measurement, not a cosmological assumption.
1. 99a.
  The authors' own statement of the cause of the shift is the quote below. Section Abstract, page 1: Preprint page. “We review five independent techniques which are used to set the distance scale to globular clusters, including subdwarf main sequence fitting utilizing the recent Hipparcos parallax catalogue. These data together all indicate that globular clusters are farther away than previously believed, implying a reduction in age estimates.” “Our best estimate for the mean age of the oldest globular clusters is now 11.5 ± 1.3 Gyr, with a one-sided, 95% confidence level lower limit of 9.5 Gyr. This represents a systematic shift of over 2 σ compared to our earlier estimate, due completely to the new distance scale—which we emphasize is not just due to the Hipparcos data.” Section 1: Their earlier value, quoted in section 1 of the same paper, was. “14.6 ± 1.7 Gyr, with the one-sided 95% C.L. lower bound of 12.1 Gyr”. and they record that. “the uncertainty in Mv (RR) overwhelmingly dominated the uncertainty in the GC age determination.” The RR Lyrae calibrator moved from M_V(RR) = 0.60 at [Fe/H] = −1.9 to 0.39 ± 0.08 (statistical), with an additional systematic of +0.13/−0.18. ↩
100.
Krauss, Lawrence M., and Brian Chaboyer. “Age Estimates of Globular Clusters in the Milky Way: Constraints on Cosmology.” Science 299:65–69. 2003. A paper that reanalyses the ages of the Milky Way's globular clusters, the old star clusters, and what those ages imply for the age of the universe. Its own reanalysis gives a median cluster age of 12.4 Gyr, with a 95 per cent lower bound of 10.2 Gyr, and a best-fit age of the universe of 13.2 Gyr. The article uses it for the authors' own admission that the cluster-age limit did not hold up.
1. 100a.
  Pages 1–2: On the history, the authors write: “By comparing a lower limit on the age of the oldest globular clusters in our galaxy — estimated in the 1980s to be 15-20 Gyr — with the expansion age, determined by measurements of the Hubble constant, an apparent inconsistency arose: globular clusters appeared to be older than the Universe unless one allowed for a possible Cosmological Constant.” Page 2: This is the passage the article quotes. “Unfortunately this constraint did not turn out to be robust. As more careful examinations of uncertainties associated with stellar evolution were performed, and well as refined estimates of the parameters that govern stellar evolution itself, the lower limit on globular cluster ages got progressively lower and lower, so that a wide range of cosmological parameters produced Hubble ages consistent with this lower limit.” This is a working astronomer complaining that the cluster ages had drifted in the direction that weakened Krauss's own argument for a cosmological constant, made before 1998. It is the opposite of what the tuning accusation predicts. ↩
101.
Faulkner, Danny R. “HD 140283: Older than the Universe?” Answers in Genesis. Sep 20, 2017. A short popular web article about HD 140283, a subgiant star, meaning one that has just used up the hydrogen in its core. A research team dated it at 14.46 ± 0.8 billion years, and the author sets out three objections to that age and then says the results depend on the model used. The article uses it because it shows the creationist reply that the age depends on the stellar model.
1. 101a.
  The three objections, in turn. “A plot of the intrinsic brightness versus temperature revealed that HD 140283 is a sub-giant star. Astronomers understand that sub-giant stars have recently exhausted the hydrogen fuel in their cores. Application of the most up-to-date models of how such stars evolve enabled the team to determine the age of HD 140283 to be 14.46 ± 0.8 billion years.” “First, this would require both sets of errors to conspire in a very convenient way. Second, 150 million years probably is not sufficient time for the star to have formed within the current theories of the early evolution of the universe. Third, astronomers generally do not think that population II stars, such as HD 140283, are primordial. Instead, population II stars supposedly were preceded by population III stars.” “Keep in mind that these results are model-dependent. If one changes the model, the problem may not even exist. Certainly, within the creation model, this isn't a problem.” The same article contains a factual slip. Since age is proportional to 1/H₀, a lower Hubble constant gives an older universe. Two sentences earlier the author correctly says the constant 'was revised upward 25 years ago', so the slip reads as a slip of the pen rather than a misunderstanding, but the sentence as printed is wrong. “was revised upward 25 years ago”. “when the Hubble constant was revised downward, the age of the universe decreased.” ↩
102.
Samec, Ronald G. “The apparent age of the time dilated universe: explaining the missing intracluster media in globular clusters.” Journal of Creation 27(2):5–6. Aug 2013. Samec's framing, in his own words: “In creation time dilation cosmologies (e.g. Humphreys and Hartnett), while the earth experiences less than 10,000 years of recorded history (God's time clock), millions, and possibly billions, of years pass in the distant universe.” His method: “Our quest is to find natural reference clocks (NRCs) like Newtonian orbital periods, the speed of light, and other well-known, well-observed physical rates or frequencies accepted by scientists, worldwide, to determine the apparent age of the time dilated universe.” The dependency on the age of the Sun, in his own words: “astronomical dating schemes are corrupted by the assumption that the age of the sun is 4.57 x 10⁹ years. I call this the Solar Age Condition (SAC). … We now know that RATE results completely discredit this age.” The prediction: “If the ICM is 1/10 to 1/100 the amount expected, as determined by the observations, and plane crossings are 10⁸-10⁹ years apart, our NRC gives a range of age limits of the globular clusters of only 10⁶-10⁸ years … This is much less than the 13.82 x 10⁹ years given by cosmologists for the age of the cosmos (100th to 10,000th of this age!)” The paper he cites for the gas deficit, Moore & Bildsten 2011 (ApJ 728(2): 81), proposes outflows from classical novae as the clearing mechanism. Samec's objection is that the nova rate is uncertain, not that the mechanism is wrong. “The mechanism is uncertain”. That does not license reading a gas deficit as an age. His own footnote 9 presents the age as an open programme: “Is the apparent age of the universe ~14 x 10⁹ years as calculated by current astronomical thought or some other value? Here we have estimated the quantity to be 100th to 1,000th of this age … But it is too early to make the call.” A creationist article by Ronald Samec, first printed in August 2013 and put on the web on 29 March 2015, updated 24 April 2015. It argues that globular clusters, tight balls of stars that orbit a galaxy, hold far less gas and dust than their stars should have shed, and that a clock built on that gap gives the clusters an age of only 10⁶-10⁸ years. The article quotes that figure against the 13.82 x 10⁹ years that cosmologists give for the universe. ↩