Genesis 1, tested against science and every ancient creation story

The short version

1. We fixed the scorecard before scoring anything.

   Fifteen datable events with ages from science, a fixed menu of reinterpretations, and 13 yes/no content claims, including ones Genesis gets wrong.

   The rules ↓
2. Read plainly, Genesis 1 beats chance but sits mid-table.

   Its order agrees with science (Kendall tau 0.46, p = 0.011), but among 16 ancient creation stories it ranks 7th. Six do better, led by Anaximander, Lucretius and the Zoroastrian Bundahishn.

   The ranking ↓
3. Its misses are day 4.

   17 of its 25 wrong-order pairs come from the sun, moon and stars arriving after the earth, the land and the plants, where its literary plan puts them.

   The misses ↓
4. Give every text the same reinterpretations, and Genesis rises to 3rd.

   With the freedom paid for in the null, it is clearly non-random (best tau 0.92, p = 0.0002), behind Anaximander and the Orphic theogony.

   The fair comparison ↓
5. The real distinctive is theological, not hidden science.

   Its cosmos is not divine and not a god's body, which happens to be true. That fits a Creator who taught in his hearers' picture of the world; it is weak evidence (likelihood ratio about 1.3), not a proof.

   What it shows ↓

The question

Does Genesis 1 (and Genesis 2) describe the origin of the cosmos, the earth and life in a way that agrees with modern science better than an ancient human author could be expected to manage, judged against what ancient human authors actually managed?

A random order is the easy comparison. The fair one is the other creation stories people wrote, from Babylon and Egypt to Greece, Persia, Scandinavia, China, the Maya and New Zealand. We test three hypotheses:

The unified argument needs one number from this page: how much more likely is what we see if God inspired Genesis than if a human wrote it? Theism here is a mix of the first two hypotheses, which is why the answer ends up close to 1. One thing is deliberately left out. The first verse, "In the beginning, God created the heavens and the earth" (Genesis 1:1, WEB),¹ fits the hot Big Bang 13.8 billion years ago. How far that goes is weighed once, on the beginning page: the date is established, but an absolute beginning is not (a first state is a strong inference with named exceptions, and "time began" is not established). It is not counted again here.

Rules of evidence

These were fixed before any text was scored. The full code is in our calculation script.²

1. The scorecard comes from science, not from Genesis. Fifteen event types (the universe, light, stars, the sun, the earth, the moon, the atmosphere, the ocean, dry land, land plants, fruit trees, sea animals, flying animals, land animals, humans), each dated by its first appearance. Each event also has a fixed menu of alternative identifications that apologists actually use: the lights "became visible", birds read as winged insects, vegetation read as microbial mats on land. The menu applies to every text.
2. Coding a text. We list what the text says comes into being, in story order. Things made in the same act share a rank (for Genesis 1, the unit is each "God said"). Things a text does not mention are left out, not scored as misses.
3. The statistic. Kendall tau-b, a rank correlation between story order and scientific order: 1 means the same order, 0 no relation, -1 reversed. It counts pairs: of every two items, does the story put them in the right order (concordant) or the wrong one (discordant)? We list every wrong pair.
4. Four nulls, from weakest to most realistic. A null hypothesis is the "nothing special is going on" baseline. (a) Random order: shuffle the same items 100,000 times and see how often chance does as well. (b) Reading freedom: let every text pick its best reading from the same menu, and compare it with the best that random orders reach under the same freedom. (c) Literary structure: given Genesis' own plan, how many of the possible orders fit science as well? (d) The real one: where Genesis ranks among the creation stories humans actually wrote.
5. A content scorecard of 13 yes/no claims (C1-C13), scored 1, ½, 0 or "silent" for every text. It deliberately includes three items Genesis gets wrong or awkward on a plain reading: the sun before plants, no solid sky with water above it, and deep time.
6. Exclusions. "An early dark Earth" is excluded from the scorecard because it is not established science. The Flood is out of scope. "A beginning" (C1) is scored but not counted (it belongs to the beginning page; a 1 there means "consistent with current cosmology", not "confirmed"), and neither is monotheism itself, which a theological text asserts under every hypothesis.
7. Honesty about the coder. The person who coded the texts knew how Genesis would come out, so nothing here was pre-registered. To compensate, every coding is in the script with its source text, and the sensitivity runs below show which single choices move the result.

The evidence

The scientific order

In order of first appearance: the universe and its light, the first stars, the sun, the earth and the moon, the atmosphere and the ocean, emergent dry land, animals in the sea, land plants, land animals, birds, flowering (fruit-bearing) plants, humans. Two facts matter most for Genesis. The sun is older than the earth, all land and all plants. And birds come after the first land animals by about 250 million years, while fruit-bearing trees come after almost everything else.

Sixteen creation stories, coded

Genesis 1 in its own words, with the story order we score:

"The earth was formless and empty. Darkness was on the surface of the deep... God made the expanse, and divided the waters which were under the expanse from the waters which were above the expanse... let the dry land appear... Let the earth yield grass, herbs yielding seeds, and fruit trees bearing fruit... God made the two great lights: the greater light to rule the day, and the lesser light to rule the night. He also made the stars." (Genesis 1:2, 7, 9, 11, 16, WEB)¹

Coded order: heavens, earth and the deep (1:1-2); light (day 1); the expanse (day 2); dry land, then grass, herbs and fruit trees (day 3); sun, moon and stars (day 4); sea creatures and birds (day 5); land animals, then humans (day 6). Fifteen items in all, the most of any text. Fifteen other stories were coded the same way, each from a translation we read or, where marked, from standard scholarly summaries.

What Genesis 1 refuses to say

The parallels with Babylon are real: primeval water, the waters divided, a sky that holds water back, humans made last. But Genesis turns the central moves around.

The content scorecard

On the 13 content claims, Genesis 1 read plainly averages 0.73 on the 11 it addresses (leaving out "a beginning"). Of 17 other texts, only Anaximander's fragments score higher (0.93 on 7 items). Its distinctive hits are C2, the world is not a god's body (other texts average 0.38), and C3, the lights are not gods (others 0.21). Its plain-reading misses are C8 (the sun after the plants), C10 (waters above the expanse) and C11 (six days).

The analysis

Plain reading: Genesis 1 among sixteen

Genesis 1's order agrees with science: 71 pairs of items are in the right order and 25 in the wrong order. Its p-value is among the smallest because it has the most items, 15. Its effect size, tau, is in the middle of the field. Leave out the two naturalist philosophers (Anaximander and Lucretius) and it is still 5th of 14, behind the Bundahishn, Pangu, the Orphic theogony and the Timaeus.²

Against pure chance

So Genesis 1 is clearly not a random list. That is the least interesting finding on this page, because no human author writes at random either. Water before land, plants before animals and humans last are what common sense gives, and the next two sections ask how much is left once common sense and the text's own plan are allowed for.

Where Genesis 1 misses

All 25 wrong-order pairs on the plain reading, grouped. 17 of them involve the sun, moon and stars of day 4, and 2 more involve the light of day 1, dated to the Big Bang.

Two content misses sit beside these: the waters "above the expanse" (1:7), and six days against 13.8 billion years. The Hebrew verb for what God did with the lights in 1:16 is 'asah, "made". Reading it as "made visible" removes most of the day-4 misses, and the next section prices that reading.

Five ways to read Genesis 1, and what each buys

Christians read Genesis 1 in several ways, and each makes a different claim about order. We scored each one that makes a claim we can test.

The day-age score looks spectacular, but each of its identifications was made after the science was known. "Vegetation" becomes microbial land cover 1.2 billion years ago, which is not grass, herbs or fruit trees. "Made" becomes "became visible". The day-5 sea creatures include sea mammals 500 million years after day 5 begins, so the epochs overlap. And the day-6 "livestock, creeping things, and animals of the earth" are narrowed to particular mammals, although creeping things (arthropods, reptiles) are older than the day-5 birds by roughly 160-275 million years.

The framework and functional readings are honest about what they are, but they cut both ways. A reading that makes no claim about sequence cannot be refuted by the misses, and it cannot be confirmed by the hits. For sequence, its likelihood ratio is exactly 1.

Choosing the best of the three sequence readings after seeing the data costs almost nothing here (the adjusted p stays at or below 0.00001, the resolution of 100,000 shuffles), because Ross's mapping is nearly perfect on 15 items. The real cost is in the mapping itself, and the only fair way to price it is to offer the same freedom to everyone.

Same freedom for every text

Here every text may pick, item by item, from the same menu of alternative identifications (the lights "visible", birds as insects, vegetation as microbial mats, and so on). Its best score is then compared with the best score random orders reach under the same freedom, so the freedom is paid for in the null.

With the freedom paid for, Genesis 1's order is clearly non-random (p = 0.0002), and with 15 items it has the smallest p-value of any text. But seven other texts also beat random order at p < 0.05, and two of them reach a higher tau. There is also a bias we have not corrected: the menu was built from the moves of Genesis apologists (mostly Ross's), so it suits Genesis better than it suits texts whose own defenders would have invented other moves. That bias favours Genesis.²

The literary plan, and a 1-in-6 coincidence

Most commentators, Kline among them, see two triads in Genesis 1. Days 1-3 form three realms (light; sky and seas; dry land with its plants), and days 4-6 fill them in the same order (lights; birds and fish; land animals and humans). Given that plan, the only free choice left is the order of the three realms. We scored all six:

Genesis uses the best of the six: a 1-in-6 chance if the order were picked at random (p = 0.17). But a human author would not pick at random. Heavens, then waters, then earth is the ordinary top-down order of the ancient three-tier world (the Bundahishn uses it too: sky, water, earth). And the plan itself forces the lights into the fourth slot, after the plants, which explains the biggest miss with no appeal to physics. This is the single most important finding on the page: most of Genesis 1's order is explained by its literary architecture plus common sense, and the choice that remains is modest.

Is there a law that puts the days in this order?

A fair challenge to everything above: rather than matching items one by one, show a principle (information, energy, "stability") that must increase from day to day, so the order had to be this one. Otherwise, the worry goes, this is a story fitted to the data. The challenge can be answered exactly. Any quantity that never decreases over cosmic time, and jumps when a new kind of thing first appears, rises across the days if and only if each day ends at a real moment of Earth history: a time when everything made so far exists and nothing made later exists yet. So the test is on the order itself, and no choice of quantity can do better.²

On this stricter measure the plain reading ranks 10th of 16 texts; Anaximander (5 of 5), the Orphic theogony (2 of 3) and the Bundahishn (3 of 5) lead. It is the same order data seen another way, so it adds no separate weight to the verdict. What it settles is that no law of increase makes Genesis 1's days necessary: one appears only under readings built after the science was known.

Birds before the creeping things

Birds on day 5 before the land animals of day 6 is, with the fruit trees, the miss the day-age reading cannot shed. Two rescues are common: read the flyers as winged insects and the day-6 animals as mammals, or say Genesis orders the conquest of domains (water, then air, then land). We coded day 6 with the three kinds the text names: "livestock, creeping things, and animals of the earth" (1:24, WEB).^{1, 2}

The domain rescue fails on the dates: animals reached land (arthropods about 425 million years ago, four-legged animals by 395) about 100 million years before any animal flew (winged insects about 325). The order of conquest is water, land, air. The miss survives every reading we tried. The only ways out are to drop the creeping things from day 6 or to read the days as not chronological at all, and the second gives up the order as evidence.

The result, with its uncertainty

Human author (naturalism):
  an order this good, given the forming/filling plan      about 1/6 to 1/3
  (the realm-order coincidence, raised because top-down is the natural ancient order;
   the rest is common sense, which the best human texts show)
Concordist inspiration:
  a good order                                            near 1
  but the misses are observed, removed only by post-hoc readings
  P(misses like these | concordist)                        about 0.2 to 0.5
Accommodated inspiration:
  predicts what a human author predicts on order and content   ratio about 1
Anti-mythic, non-divine cosmos:
  a little more expected under theism, but expected under
  naturalism once Israel is monotheist                     at most 1.2 to 1.5

Theism taken as about half concordist, half accommodated:
  central  1.3
  low      0.6   misses weighed heavily against concordism; accommodation as ad hoc retreat
  high     3     the 1-in-6 realm order, sea-before-land life and the anti-mythic content
                 credited fully; the literary plan treated as a vehicle God chose
  above 3 requires counting the post-hoc readings as free, which our method forbids

Objections

What this shows, and what it doesn't

Net: weak evidence for inspiration, a likelihood ratio of about 1.3 (range 0.6 to 3). It is consistent with a Creator who teaches theology in his hearers' picture of the world. It is not a demonstration of hidden science.

What we don't claim

• "Genesis 1 perfectly matches the fossil record." On the plain reading, 25 of 96 ordered pairs are wrong. Even on Ross's dates the fruit trees are wrong and the categories must be narrowed.
• "Science confirms the early Earth was dark, as in Genesis 1:2." Not established. "Formless, dark, water-covered" is matched only by a model chosen to match; the young Earth had a sun, and probably a magma ocean before any water ocean.
• "Day 4 is when the atmosphere turned transparent." That is a reading, not a finding. Say "on the day-age reading, day 4 is taken as..." and give its price.
• Day-by-day charts that say Genesis "perfectly matches" the fossil record. One we once relied on contradicts its own dates (it puts land plants, a day-3 event, after the Cambrian sea life of day 5), and birds on day 5 do not match the fossil order, since land animals came first.
• "Genesis is unique in its order." The Bundahishn's six creations (sky, water, earth, plants, animals, mankind) match it closely, and Anaximander and Lucretius beat it.
• "Genesis is unique in creation by word." The Memphite Theology and the Popol Vuh both have it.
• "Tehom proves Genesis rewrote the Tiamat myth", or the reverse. The two words share a common Semitic root.
• "The expanse is just the atmosphere, so Genesis has no ancient cosmology." The waters above (1:7; Psalm 148:4) and the lights set in the expanse are in the text. Concede the cosmology and argue accommodation.
• "Ross's timeline is established science." In Ross's own words, "These dates depend on how one interprets the events."
• Flood geology. A global flood does not explain the geological column or the fossil order; it contradicts the dating used throughout this page, and it is outside its scope.
• "Genesis was written 3,000+ years ago, so it predates Greek science." On the critical dating, Genesis 1 is roughly contemporary with Anaximander. Say "an ancient Israelite text" unless the dating is argued (see the history page).
• "Hindu cosmology has deep time, so..." The Vedic texts scored here give no time scale; the billions-of-years cycles are from the later Puranas. Do not use them either way without a source.
• Quotations of Hugh Ross that circulate without a published source. Quote Ross only from what he published.
• "The chance that an ancient author guessed this sequence is effectively zero." Claims like this come with no calculation. The real numbers: about 1 in 90 against random order on the plain reading, and 7th of 16 against real ancient authors.
• "Genesis 1 follows a proven law: information or 'stability' increases every day." No such quantity has been computed, and any version of it fails on the plain reading at 7 of 8 steps (above).
• "Genesis 1 is a chronologically accurate summary of cosmic history, seen from the Earth." Only on readings chosen to fit, and even then the fruit trees and the flyers before the creeping things remain.
• "The birds problem is a straw man: Genesis orders the conquest of water, air, then land." Animals reached land about 100 million years before the air.
• "Plants oxygenated the air and cleared a cloud canopy, so the sun became visible on day 4." Oxygen first rose with microbes (cyanobacteria) in the Great Oxidation Event, about 2.4-2.3 billion years ago,⁴⁵ long before land plants; a cloud canopy is speculation, not a finding.
• "Science shows the universe began from nothing, as Genesis 1:1 says." The hot Big Bang is established; an absolute beginning is not, and "time began" is not established (see the beginning page).
• "'Erev, evening, means 'mixing', so the evenings are unmanifest potential." Folk etymology: "evening" and "mixture" are different words.⁴⁴
• "'Dominion' means humans make the world real by observing it." It rests on a minority reading of quantum mechanics, and it contradicts Genesis 1 itself, where plants and animals are made and called good before any human exists.
• "Gödel's theorems prove the universe cannot describe its creator." Gödel's theorems are about what consistent formal systems of arithmetic can prove, not about a universe and its cause.

Sources

1. World English Bible (public domain): Genesis 1-2 and 7:11; Psalms 74:13-14 and 148:4; Job 37:18. ebible.org/eng-web
2. Our calculation: truth/research/calc/05-sequence_scoring.py (Python) and its output 05-sequence_scoring.out.txt; the snapshot test, the cosmic-waters reading and the birds and creeping-things codings are in 05-lt_checks.py and 05-lt_checks.out.txt, which load the same data. Every event date, every text's coding and every reading is a line in the script, so any cell can be changed and re-run. The permutation histogram on this page uses the script's own tau_test function and random seed (it reproduces p = 0.0109 exactly).
3. Planck Collaboration (2020). "Planck 2018 results. VI. Cosmological parameters." Astronomy & Astrophysics 641: A6 (age 13.787 ± 0.020 Gyr). arXiv:1807.06209
4. Ross, H. (9 October 2020). "What are the scientific dates for the starts and ends of the six creation days?" Reasons to Believe. reasons.org
5. Bromm, V. & Yoshida, N. (2011). "The First Galaxies." Annual Review of Astronomy and Astrophysics 49: 373 (first stars about 100-250 Myr after the Big Bang).
6. Connelly, J. N. et al. (2012). "The absolute chronology and thermal processing of solids in the solar protoplanetary disk." Science 338: 651 (4,567.3 Ma).
7. Wilde, S. A. et al. (2001). Nature 409: 175; Mojzsis, S. J. et al. (2001). Nature 409: 178 (Hadean zircons: liquid water by about 4.4-4.3 Ga).
8. Chowdhury, P. et al. (2021). "Magmatic thickening of crust in non-plate tectonic settings initiated the subaerial rise of Earth's first continents 3.3 to 3.2 billion years ago." PNAS 118(46): e2105746118.
9. Bindeman, I. N. et al. (2018). "Rapid emergence of subaerial landmasses and onset of a modern hydrologic cycle 2.5 billion years ago." Nature 557: 545.
10. Morris, J. L. et al. (2018). "The timescale of early land plant evolution." PNAS 115: E2274.
11. Herendeen, P. S. et al. (2017). "Palaeobotanical redux: revisiting the age of the angiosperms." Nature Plants 3: 17015 (earliest accepted angiosperm pollen about 130 Ma).
12. International Commission on Stratigraphy. International Chronostratigraphic Chart v2023/09 (base of the Cambrian 538.8 Ma).
13. Schachat, S. R. et al. (2023). "Illusion of flight? Absence, evidence and the age of winged insects." Biological Journal of the Linnean Society 138(2): 143-168, on fossil (about 328-324 Ma) versus molecular dates. Smithsonian repository
14. Niedźwiedzki, G. et al. (2010). "Tetrapod trackways from the early Middle Devonian period of Poland." Nature 463: 43.
15. Hublin, J.-J. et al. (2017). "New fossils from Jebel Irhoud, Morocco and the pan-African origin of Homo sapiens." Nature 546: 289 (about 315 ka).
16. King, L. W. (1902). The Seven Tablets of Creation. London: Luzac (Enuma Elish, Tablets I, IV-VI). sacred-texts.com/ane/stc
17. Lambert, W. G. (2013). Babylonian Creation Myths. Winona Lake: Eisenbrauns (Tablet V: the earth from Tiamat's other half).
18. Budge, E. A. W. (1912). Legends of the Gods, "The Legend of the God Neb-er-tcher, and the History of Creation" (Bremner-Rhind papyrus). sacred-texts.com
19. Allen, J. P. (1988). Genesis in Egypt: The Philosophy of Ancient Egyptian Creation Accounts. Yale; Pinch, G. (2002). Egyptian Mythology (Hermopolis).
20. Hesiod, Theogony 116-138, 371-374, tr. H. G. Evelyn-White (Loeb, 1914). Project Gutenberg #348
21. Aristophanes, Birds 693ff., in The Complete Greek Drama (1938). classics.mit.edu
22. West, M. L. (1983). The Orphic Poems. Oxford: Clarendon (the Rhapsodic theogony).
23. Rig Veda 10.129 and 10.90, tr. R. T. H. Griffith (1896). 10.129, 10.90
24. Bundahishn, tr. E. W. West, Sacred Books of the East 5 (1897), ch. 1-2, 12. avesta.org
25. Snorri Sturluson, Gylfaginning 5-9, tr. A. G. Brodeur (1916). sacred-texts.com
26. Leeming, D. (2005). "Chinese mythology." The Oxford Companion to World Mythology, as quoted at mit.edu; Birrell, A. (1993). Chinese Mythology: An Introduction. Johns Hopkins.
27. Christenson, A. J. (2007). Popol Vuh: The Sacred Book of the Maya, electronic edition. mesoweb.com
28. Grey, G. (1855). Polynesian Mythology, "The Children of Heaven and Earth." sacred-texts.com
29. Plato, Timaeus, tr. B. Jowett. classics.mit.edu
30. Lucretius, De rerum natura 5, tr. W. E. Leonard (1916). classics.mit.edu
31. Kirk, G. S., Raven, J. E. & Schofield, M. (1983). The Presocratic Philosophers, 2nd ed. Cambridge (Anaximander: DK 12A10, A27, A30).
32. Lambert, W. G. & Millard, A. R. (1969). Atra-hasis: The Babylonian Story of the Flood. Oxford: Clarendon.
33. The Memphite Theology (Shabaka Stone), tr. M. Lichtheim, Ancient Egyptian Literature I (1973), 51-57. attalus.org
34. Hasel, G. F. (1972). "The Significance of the Cosmology in Genesis I in Relation to Ancient Near Eastern Parallels." Andrews University Seminary Studies 10: 1-20 (quoting G. von Rad, Genesis, p. 53). biblicalelearning.org
35. Tsumura, D. T. (2005). Creation and Destruction: A Reappraisal of the Chaoskampf Theory in the Old Testament. Winona Lake: Eisenbrauns.
36. Seely, P. H. (1991, 1992). "The Firmament and the Water Above." Westminster Theological Journal 53: 227-240 and 54: 31-46.
37. Poythress, V. S. (2015). "Rain Water versus a Heavenly Sea in Genesis 1:6-8." Westminster Theological Journal 77.
38. Collins, C. J. (2006). Genesis 1-4: A Linguistic, Literary, and Theological Commentary. P&R; (2018). Reading Genesis Well. Zondervan.
39. Kline, M. G. (1996). "Space and Time in the Genesis Cosmogony." Perspectives on Science and Christian Faith 48: 2-15. asa3.org
40. Walton, J. H. (2009). The Lost World of Genesis One: Ancient Cosmology and the Origins Debate. IVP Academic.
41. Young, D. A. (1987). "Scripture in the Hands of Geologists." Westminster Theological Journal 49.
42. Lamoureux, D. O. (2008). Evolutionary Creation: A Christian Approach to Evolution. Wipf & Stock.
43. Origen, De principiis IV.16, tr. F. Crombie, Ante-Nicene Fathers 4 (1885). newadvent.org
44. Strong's Hebrew 6153 'ereb, "evening" (from 6150), and 6154 'ereb, "mixture, woof" (from 6148), with Brown-Driver-Briggs. biblehub.com/hebrew/6153, 6154
45. Lyons, T. W., Reinhard, C. T. & Planavsky, N. J. (2014). "The rise of oxygen in Earth's early ocean and atmosphere." Nature 506: 307-315.