Flood Basalts

"Clearly, the duration of actual eruptions appears to be almost negligible compared to the time between flows, as recorded by alteration, and possibly completely unrecorded in a number of instances." — Anne-Lise Chenet, Vincent Courtillot and colleagues, on 1,200 m of Deccan lava, 2008¹

"even here in the youngest large igneous province, argon dating cannot resolve intervals and durations of eruptions." — Tiffany Barry, Stephen Self and colleagues, on the Columbia River basalts, 2013²

"Many of these red horizons are instead paleosols that record significant hiatuses in basalt flow emplacement." — Nathan Sheldon, on the layers between Columbia River flows, 2003³

Summary

A flood basalt is a pile of lava flows so large that no eruption in recorded history comes close.  The Columbia River Basalt Group in Washington, Oregon and Idaho covers over 210,000 km² and holds about 210,000 km³ of basalt.⁴  The Deccan Traps of India hold about three times that.  The Siberian Traps hold at least 3,000,000 km³, about fourteen times the Columbia River total.⁵  Young-earth geologists have argued since the 1990s that these piles were laid down in months, at the end of Noah's Flood or just after it.  This article asks whether the evidence lets them.

What the young-earth side gets right.  Three things, and each is settled mainstream geology.

1. Single flows are fast. The Roza Member of the Columbia River group, 1,300 km³ of lava over 40,300 km², was built in about 14 years by the slowest published model.⁶ Deccan flow groups of 3,000 to 20,000 km³ went out in a few decades by paleomagnetic evidence.¹ Catastrophic emplacement of a single flow is not a creationist idea. It is the textbook.
2. Global heat is not the objection. This article computes the whole heat content of the Columbia River basalts at about 8.7 × 10²³ J. That is six millionths of the 1.4 × 10²⁹ J that catastrophic plate tectonics must dispose of for the new ocean floor.⁷ Spread across the whole ocean it would warm the water by 0.16 K. The Flood's heat problem is the ocean floor, and the young-earth side's own physicists say so. Regionally the picture is different, and Part 2 prices it.
3. Water touched these flows. Pillow lavas at the margins, palagonite, and the water-quenched entablature texture that fills the middle 60 to 70% of a typical Columbia River flow are all real.⁸ Long and Wood showed in 1986 that ordinary heavy rain cannot make entablature. Standing or flooding water is needed.⁸

What decides the question is the time between the flows, not the flows.  Every clock that has been applied to the gaps says the same thing, and the clocks are of four different kinds.

• Soils. The reddened layers between Columbia River flows carry root traces, soil horizons, ped structure, land snails, fossil plants and peat.³ Hobbs and Parrish count ten such soils in the Grande Ronde and Wanapum formations alone.⁹⁹ Kasbohm and Schoene count eight red boles between two dated ash beds inside one member.¹⁰ Chenet counts 23 in 1,200 m of Deccan. The creationist field party that reported soils as "very uncommon" saw one in 50 outcrops and had no sampling design.¹¹
• A counted lake. A Columbia River flow dammed a river in Idaho and made the Clarkia lake. Its 7.5 m of laminated mud is about 840 annual layers, with deciduous leaves in the autumn layers.¹² One gap, one lake, more than twice the Flood year.
• Deccan lakes without counting anything. A perennial freshwater lake between two Deccan flows left 2.6 m of sediment with diatoms, amoebae and sponges. Its authors refuse to count the laminations as years and still reach "at least several thousand years or more" from thickness alone.¹³
• Animals nesting. Turtle and crocodilian eggshells occur in the beds between Deccan flows.¹⁴ Eggs are laid on dry ground by a breeding population.

The radiometric and paleomagnetic durations agree with the gaps.  U-Pb on zircon puts more than 95% of the Columbia River basalts between 16.7 and 15.9 million years ago, about 800,000 years, and that is the fastest published figure.¹⁰  Two clocks in the same issue of Science put the Deccan at about a million years and disagree only about whether it came in pulses.¹⁵¹⁶  The Siberian Traps took about two million years, with two-thirds of one section in 300,000 ± 126,000 years.⁵  Even Chenet, whose paper gives the young-earth side its fastest numbers, prices the same 1,200 m Deccan section at 15,000 to 115,000 years once the gaps are added.¹

Three weaknesses in the creationist case that the creationist literature has not addressed.

1. They do not agree on when the basalts erupted. Steven Austin puts flood basalts in a post-Flood decline of volcanic power.¹⁷ John Morris puts them "during the very last stages of the Flood or in the years of readjustment to follow."¹⁸ John Woodmorappe, Michael Oard and Tim Clarey put them inside the Flood's receding phase.¹⁹ Only the last position owes an answer to the 365-day arithmetic. No creationist source found for this article acknowledges the split.
2. Their volume figures are stale and too small. Morris uses 40,000 cubic miles, about 167,000 km³.¹⁸ Woodmorappe and Oard use 174,000 km³.¹¹ The current mapped figure is about 210,000 km³. The province grew with mapping, and the argument understates its own problem by about 20%.
3. One key mainstream paper is cited backwards. Woodmorappe and Oard cite the near-zero cooling of the Ginkgo flow over 500 km as evidence of "extraordinarily rapid emplacement."¹¹ Ho and Cashman, who measured it, say the low cooling rate is "inconsistent with turbulent flow models" and points to an insulating crust, which is the physical basis of the slow inflation model.²⁰

Verdict.  Flood basalts are catastrophic deposits and the heat they carry is small.  On both points the young-earth side is right and its critics are often wrong.  But the record between the flows, read by four independent methods, requires centuries to tens of thousands of years per province, and the fastest mainstream eruption rates would still need over 2,000 years of continuous eruption to build the Columbia River pile alone.  The Flood-year reading fails on duration between flows.  The post-Flood reading, Austin's, survives the arithmetic only by giving up the claim that the basalts are Flood deposits, and it then has to fit soils, lakes and forests into the centuries after the Flood without saying how many centuries that was.

Part 1: What a flood basalt is, and how big

Schoene and colleagues open their 2019 Science paper with the working definition:  "Continental flood basalt provinces are characterized by eruption of >1 million km3 of basalt over a period of <1 million years (1, 2), representing the largest volcanic events on Earth."¹⁶¹⁶  Three provinces carry most of the argument, and it helps to have their sizes side by side.

The Columbia River group is the one creationists write about, because it is the best exposed and the easiest to visit.⁴  It is, in Reidel's words, "the youngest and smallest continental flood basalt province on Earth."⁴  Whatever a young-earth model can do for the Columbia River basalts it must do three times over for the Deccan and fourteen times over for Siberia.

The Columbia River pile has 43 named stratigraphic members¹⁰ and about 300 individual flows, some as large as 2,000 km³, a few of which travelled up to 750 km from their fissures.¹¹  Five formations make it up, and one of them carries almost three-quarters of the lava:  Steens 31,800 km³, Imnaha 11,000 km³, Grande Ronde 150,100 km³, Wanapum 12,175 km³, Saddle Mountains 2,424 km³.¹⁰

For scale, the largest lava eruption in recorded history is Laki, Iceland, in 1783.  Austin gives its figures accurately:  12.3 km³ of lava from a 25 km fissure over eight months, covering 565 km².¹⁷  The Columbia River pile is about 17,000 Lakis.  A single Roza flow field is about 100 Lakis.  Austin's point, that ancient lava eruptions were a thousand times bigger than modern ones, is correct, and nobody in the mainstream disputes it.¹⁷

Two traps for anyone quoting these numbers.  The area in km² and the volume in km³ are both about 210,000, so it is easy to print one for the other;  Clarey's "81,000 square miles" is the area, correctly converted.¹⁹  And the creationist literature uses the 1989 volumes.  Morris gives "some 65,000 square miles and with a volume of about 40,000 cubic miles," about 168,000 km² and 167,000 km³.¹⁸  Woodmorappe and Oard give 163,700 km² and 174,000 km³.¹¹  The Steens Basalt has since been folded in and the mapped total rose to 210,000 km³.

The Siberian Traps deserve one more sentence, because their size hides a second problem.  Burgess and Bowring put the minimum volume at "~3 × 106 km3" and note that "Roughly 50% of this total volume is attributed to intrusive rocks."⁵  A sill is magma injected between layers of rock that already existed and were already hard.  Half of the largest flood basalt on Earth was emplaced inside kilometres of older sedimentary rock, which had to be deposited, buried and lithified before the magma arrived.  Any Flood model that makes those sediments Flood deposits has to bury and harden them before the same Flood's volcanism intrudes them.

Part 2: What the young-earth side gets right

Single flows are fast, and the textbooks say so

Morris frames the Columbia River basalts as a test between two predictions.¹⁸  A Flood model predicts regional catastrophic deposits.  Uniformitarianism predicts slow local ones.  The basalts are regional and catastrophic, so the Flood model wins.¹⁸  The trouble with this test is that mainstream geology stopped predicting slow local emplacement for these flows around 1970.  Shaw and Swanson argued for turbulent flood emplacement in days.  The model that replaced theirs, from Self, Thordarson and colleagues in the 1990s, is slower, and it is still catastrophic by any ordinary meaning of the word.

Thordarson and Self worked out the Roza Member of the Columbia River group in 1998.⁶  Their abstract:  "The Roza Member represents a compound pahoehoe flood basalt lava flow field, with an area of ~40,300 km2 and a volume of ~1300 km3. … The results indicate that the emplacement of individual Roza lobes lasted for months to years and that the lava flow field was constructed over a period of at least 14 years."⁶  Their Table 6 sums four flows to between 6.1 and 13.6 years.⁶  And their conclusion states plainly what that means:  "We arrive at an eruption duration of ~14 years for the Roza eruption, indicating an average mass eruption rate of 4.2 × 106 kg/s. … Although our estimate of Roza eruption rate is ~3 orders of magnitude lower than previous estimates [e.g., Swanson et al., 1975], an eruption like Roza would still have been of catastrophic proportions."⁶

The Deccan numbers are faster.  Chenet and colleagues measured the magnetic direction of flow after flow through 1,200 m of the Mahabaleshwar escarpment.¹  Earth's field wanders by a few degrees a century, so a stack of flows that share one direction cannot be separated by much time.  Their abstract:  "Lack of paleosecular variation allows identification of four directional groups, implying very large (40 to 180 m thick) single eruptive events (SEEs) having occurred in a few decades. … The typical volumes of SEEs (corresponding to magnetic directional groups) are estimated at 3000 to 20,000 km3, with flux rates 100 km3 a−1, having lasted for decades."¹  Schoene and colleagues, using U-Pb dates, allow the same thing:  groups of flows in the Poladpur and Mahabaleshwar formations, "each potentially comprising >50,000 km3, lack secular evolution in paleomagnetic poles, suggesting eruption over decades to centuries (29).  Such high eruption rates of >1000 km3/year are permitted by our U-Pb geochronology".¹⁶

So a creationist who says that 20,000 km³ of lava can pour out in a few decades is quoting the Journal of Geophysical Research, and one who says 50,000 km³ can go out at over 1,000 km³ a year is quoting Science.  This is not the point in dispute.

Global heat is not the objection

Skeptics often say that erupting 210,000 km³ of lava in a year would cook the planet.  It would not cook the planet.  It would sterilise the Pacific Northwest.  The inputs below are standard textbook values for basalt, and they are stated so the arithmetic can be checked.  Density 2,800 kg/m³.  Specific heat 1,000 J per kg per kelvin.  Latent heat of crystallisation 4 × 10⁵ J/kg.  Eruption temperature about 1,100 °C, cooling to about 20 °C.

The last row is the one that hurts.  131,000 W/m² is about 96 times the solar constant.  A surface radiating that much sits at about 960 °C, which is another way of saying that the Pacific Northwest would be an open lava field for the whole year.  Nothing lives on it, and nothing grows a soil on it.

Worraker, a young-earth physicist, reaches both halves of this and states them plainly.  He totals the heat of every large igneous province at "not exceeding 4.0 × 10²⁶ J" and calls it "insignificant in comparison with the heat deposition involved in forming the ocean floors (1.4 × 10²⁹ J)".  Then he adds the regional caveat:  "the local impact of the heat from LIPs, which in many cases must have been released within a matter of days or weeks at most, would have been highly dramatic in the absence of some special, hitherto unrecognized, cooling mechanism."⁷  That budget is dealt with in the catastrophic plate tectonics article.⁷  A skeptic who leads with the global heat of flood basalts has chosen an argument the creationist can win.  A skeptic who asks what the Columbia Plateau was like during that year has not.

Water touched the flows

Woodmorappe and Oard report pillow lavas and palagonite at their outcrops and read them as evidence of eruption under water.¹¹  The pillows are real.  Ho and Cashman's temperature study of the Ginkgo flow sampled pillow basalt at Vantage, Molalla and Portland, 120 to 370 km from the feeder dike, and Ginkgo basalt at Yaquina Head 500 km out.²⁰²⁰  At Vantage the petrified logs of the Ginkgo Petrified Forest lie in the basal pillow complex of that flow, stripped of limbs and bark, tilted rather than standing, and Woodmorappe and Oard are right that the word "forest" oversells them.¹¹

The strongest creationist observation is the one Clarey makes about entablature.¹⁹  A thick Columbia River flow has a colonnade of large regular columns at its base and often at its top, and in between a zone of small, irregular, curving columns called entablature.  Long and Wood measured it in 1986⁸:  "The colonnade refers to the well-defined columnar structure of many flows, which appears in the lowermost 10% to 30% and, in places, the uppermost 10% to 20%. … Entablature commonly occupies the central 60% to 70% of a flow."⁸  Under the microscope the entablature has up to 60% glass, which means it cooled faster than the colonnade above and below it.  That is backwards for a flow cooling by conduction, which should cool fastest at its edges.  Long and Wood's answer is water:  "Downward migration of water through cracks would cause convective cooling of underlying, unsolidified lava."⁸  And they tested how much water.  "Despite heavy rainfall on Hawaii (for example, ~250 cm/yr at Kilauea Iki), exposed, prehistoric lava flows on that island exhibit only colonnades.  We deduce from this that heavy rainfall alone is inadequate to quench flow interiors.  Extensive flooding by deranged drainage, or perhaps extremely heavy rainfall (>>250 cm/yr), is apparently required to produce entablatures."⁸

Clarey states the mechanism correctly and then argues that the setting fails:  "Because entablature requires flooding, conventional geologists argue that the lavas flowed across dry land and dammed local rivers, causing water to overflow on top of the lavas.  The problem is that not many lava flows would have encountered a river or water source to flood the lava.  Entablature is too common in the Columbia Plateau for this scenario."¹⁹  His replacement is Flood water still draining off the continent, plus tsunamis from rapid seafloor spreading, washing over each fresh flow.¹⁹

That is a real argument and it deserves a real answer, which is in Part 5.  For now the point stands:  the Columbia River flows met a great deal of water, often, and any account of them has to supply it.

Part 3: How long each province took

The Columbia River basalts

Barry, Self and colleagues reviewed every argon date on the Columbia River basalts in 2013² and recommended eruption ages by formation:  "Steens Basalt, ca. 16.9 to ca. 16.6 Ma;  Imnaha Basalt, ca. 16.7 to ca. 16 Ma;  Grande Ronde Basalt, ca. 16 Ma to ca. 15.6 Ma;  Wanapum Basalt, ca. 15.6 to ca. 15 Ma;  and Saddle Mountains Basalt from ca. 15 Ma to ca. 6 Ma."²  Then they conceded something the young-earth side should quote:  "Under scrutiny, the data highlight areas of complexity and uncertainty in the timing of eruption phases, and demonstrate that even here in the youngest large igneous province, argon dating cannot resolve intervals and durations of eruptions."²

That concession is honest and it is also limited.  What argon cannot resolve is the fine structure inside a bracket that runs from 16.9 to 6 million years.  Nothing in the chapter offers a shorter total.  And the gap Barry named was closed five years later by a different clock.  Kasbohm and Schoene dated zircon crystals from volcanic ash beds sandwiched inside the lava pile, using U-Pb by the CA-ID-TIMS method, which does not share argon's problems.¹⁰  Their abstract:  "Our data set shows that more than 95% of the CRBG erupted between 16.7 and 15.9 Ma, twice as fast as previous estimates."¹⁰  That is about 800,000 years, and it is the fastest published duration for the province.  The average effusion rate for the main phase comes out at "0.334 ± 0.042 km3/year", rising to "1.18 km3/year" during the Wapshilla Ridge Member.¹⁰  They found no gaps longer than about 200,000 years between any two of their samples, and volcanism is known to have continued through those intervals.¹⁰

The sentence in that paper that matters most for this article is the one where the authors bound the peak rate, because they bound it with the record between flows.  "A potential constraint for the maximum rate can be derived from previous estimates for the minimum duration of the 1300 km3 Roza Member to be 14 years (29), and the ca. 100 years or more needed to develop thin redbole horizons (30).  Because there are at least 18 (and as many as 28) lava flow units in the Wapshilla Ridge Member (31) and 8 more redboles between the samples we dated, we estimate that peak eruption rates could be as high as 20 to 40 km3/year for a duration of thousands of years."¹⁰  Read that carefully.  The fastest the Columbia River basalts ever erupted, on the fastest published chronology, is 20 to 40 km³ a year, and the number that sets it is the century each thin soil between flows needs to form.

The Deccan Traps

Two papers in the same February 2019 issue of Science dated the Deccan Traps with two different clocks and reached different pictures of the tempo.  Sprain and colleagues, using argon:  "we found that the DT lavas erupted quasi-continuously for 991,000 years (see Fig. 2), from ~66.413 Ma ago [the date for Jawhar Formation (Fm.) sample KAS15-3] to ~65.422 Ma ago (the date for upper Mahabaleshwar Fm. sample PAN15-3).  This time interval represents a total estimated volume of ~560,000 km3".¹⁵¹⁵  Their mean extrusion rates are 0.4 ± 0.1 km³ a year before the end-Cretaceous boundary and 0.6 ± 0.2 km³ a year after it.¹⁵  They reject the older idea of three discrete pulses as "an artifact of limited and lower-precision geochronology".¹⁵

Schoene and colleagues, using U-Pb on zircon, see four high-volume pulses instead, each lasting 100,000 years or less, with a hiatus after the asteroid impact.¹⁶  And they allow, as quoted in Part 2, that groups of flows of over 50,000 km³ may have erupted in decades to centuries at over 1,000 km³ a year.  But the same sentence continues:  such rates are permitted only by "requiring hiatuses of hundreds to thousands of years within our resolved pulses so as not to exceed total volume estimates."¹⁶

A creationist reader may be tempted to see two teams disagreeing and conclude that nobody knows.  Be precise about what they disagree on.  Sprain says continuous, Schoene says pulsed.  Both say about a million years for the province.  Both put the fastest bursts at decades to centuries and separate the bursts by hiatuses of centuries to millennia.  Neither position is within a factor of a thousand of a Flood year, and the live dispute is about the shape of the eruption curve inside a million years, not about its length.

The Siberian Traps

Burgess and Bowring dated 17 sills, two tuffs and two lava flows from the Siberian Traps in one laboratory, using U-Pb on zircon and perovskite.⁵  Their result for the main pulse:  "our age model suggests that two-thirds of the composite 6-km-thick Noril'sk-Maymecha-Kotuy section was erupted within the 300 ± 126 ka before ~251.9 Ma."⁵  For the whole province:  "The entirety of the LIP was emplaced/erupted in a maximum of ~2 Ma".⁵  The largest flood basalt on Earth took about two million years, with its main pulse in about 300,000.

The paleomagnetic clock, which is not radiometric

Chenet's paper is the one creationists should read most carefully, because it is the strongest published case for fast eruption and it prices the gaps without any radiometric date at all.  The flows in a directional group erupted, on their estimate, in "roughly 500 years, and possibly less" for all 49 units in the section.¹  Then they add the time recorded by the red boles between flows, and the time implied by successive directional groups whose magnetic directions do not correlate, which needs at least a few centuries each for the field to move.  Paragraph 56:  "We finally estimate coarse bounds on the total amount of time recorded in the section by both volcanics and intertrappean levels to be between 85 + 2000 = 2085 yrs and 850 + 100,000 = 100,850 yrs. … The range of estimates then becomes 17,085 to 115,850 years, which we round up as 15 to 115 ka".¹  The lava itself accounts for 85 to 850 years of that.  The gaps account for the rest.

The arithmetic

Set the radiometric dates aside entirely and use only the eruption rates the young-earth side would most like to cite.

The Flood year needs 575 km³ of lava a day for a year.  Laki, the largest historic lava eruption, produced 12.3 km³ in eight months.  The fastest rates that any measurement supports would take two millennia of continuous eruption with no pauses, and every measurement finds pauses.

Part 4: The record between the flows

The flows are fast.  The question is what happened on top of each one before the next arrived.  Four kinds of evidence answer it, and none of them is a radiometric date.

Soils with roots in them

Woodmorappe and Oard visited about fifty Columbia River outcrops in 2001¹¹ and reported:  "Claimed 'weathered horizons' and 'fossil soils' between individual lava flows are very uncommon, suggesting they are the product of chemical reactions between the hot lava and water, not fossil-soil material."¹¹  They saw such a horizon once, and only because a guide took them to it:  "By any standard, they are very uncommon.  It stands to reason that, if long intervals of time had elapsed between the supposedly-episodic lava flows, weathered horizons and fossil soils should be common."¹¹  The one they saw, they suggested, "is a palagonite-like product of a reaction between hot lava and water, not fossil-soil material."¹¹

Sheldon tested exactly that reading a year later on the Picture Gorge basalts of Oregon, apparently without knowing the creationist paper existed.  Picture Gorge is the sixth Columbia River formation in Reidel's count, erupted alongside the Grande Ronde.³  His abstract opens with the hypothesis Woodmorappe and Oard favour and then rejects it on field evidence:  "Reddened, clay-rich horizons between basalt flows have historically been identified as 'baked zones' or as zones of lateral groundwater movement.  The reddening has been attributed to thermal or chemical alteration of the permeable portion of the basalt flow.  Many of these red horizons are instead paleosols that record significant hiatuses in basalt flow emplacement.  Paleosols were identified in the field on the basis of features such as horizonation, ped morphology, and root traces.  Terrestrial gastropods, fossil plants, and peats also occur locally."³  He then checked the chemistry:  "Ca, Na, Mg, Fe2+, and P have been extensively lost from the A and Bt horizons of the paleosols, consistent with modern weathering patterns."³

A hot-lava-and-water reaction does not make root traces, soil horizons, peat or land snails.  Those need a vegetated land surface left alone long enough for roots to grow down into it.  Hobbs and Parrish, working the Grande Ronde and Wanapum formations, found "Ten paleosols" identified "on the basis of pedogenic features such as horizon development, root traces, and saprolitization," and put their formation at "10⁴ to >10⁶ yr".⁹  Kasbohm and Schoene counted eight red boles between two dated ash beds inside a single member.¹⁰  Chenet counted 23 in 1,200 m of Deccan.¹  Three teams went looking with a stratigraphic design and found soils repeatedly.  A field party that visits fifty roadcuts without one has measured how visible soils are from the road, not how common they are.

The buried soils article covers the recognition criteria and Oard's argument that paleosols form faster than claimed.  One concession belongs here.  Hobbs and Parrish themselves note that argon dating puts all the Grande Ronde lavas within 0.42 million years, yet "many paleosols within these units are interpreted to represent hundreds of thousands of years of landscape quiescence".²¹  Oard quotes that accurately.²¹  Two mainstream clocks disagree by a factor of a few.  That is a real problem for one of them, and it is not evidence for a year, because both clocks are hundreds of thousands of years long.

An 840-year lake in one gap

A Columbia River lava flow dammed the ancestral St. Maries River in northern Idaho and made a lake.  The lake filled with fine mud that preserved leaves so well their tissues and biomolecules survive.  Höfig and colleagues dated ash beds in that mud by U-Pb on zircon at 15.78 ± 0.039 million years, then tested whether the laminations are annual, using elemental scans, grain size, colour, fossil content and spectral analysis.¹²  Their abstract:  "The flood basalt dammed a local drainage system to form the paleolake, which preserved a plant fossil Lagerstätte in its deposits. … Consequently, the ~7.5-m-thick varved deposit at the type location P-33 represents ~840 yr of deposition, coincident with the end of the main phase of Columbia River Basalt Group eruptions during the Miocene Climate Optimum."¹²

The creationist objection to varve counting is that a couplet can be a storm rather than a year, and the lake varves article takes that objection seriously.  Höfig's team answered it for this lake in particular:  "Abundant deciduous leaves within the fine-grained autumn sediments are consistent with rhythmic seasonal deposition of varves.  While both storm deposition and dammed lakes may result in high sedimentation rates (Glenn and Kelts, 1991), irregular storms and floods tend to create turbulence that disturbs lake stratification and destroys laminations and fossil material".¹²  The leaves that make the deposit famous are the evidence that the water was still.  The accumulation rate, about 12 mm a year, is one the authors call fast for a varved lake;  a slower rate would make the 7.5 m take longer, not shorter.¹²  And the lake sits at the end of the main eruptive phase, inside the pile, not on top of it.

Deccan lakes, counted by nothing

Between Deccan flows lie the intertrappean beds, and some of them are lake sediments.  Samant and colleagues worked up one, the Bagwanya palaeolake in central India:  laminated shales with testate amoebae, five diatom species, sponge spicules and pollen, from "a perennial, warm water, well oxygenated and very low salinity lake, surrounded by tropical to subtropical terrestrial and aquatic plants".¹³¹³  On duration they are unusually careful:  "The time duration of this paleolake is unknown but the stratigraphic record of 2.6 m thickness of these lacustrine sediments suggests that Bagwanya paleolake lasted for at least several thousand years or more.  We do not know if the shales with mm-thick laminations represent annual varves, so we do not attempt to determine the time duration of this 2.6 m section."¹³  They go on to suggest 100,000 years by analogy with another intertrappean lake tuned to orbital cycles.  That figure is second-hand and this article does not use it.  The figure it does use, several thousand years, rests on sediment thickness and on a lake ecology that took time to establish, and a creationist objection to varves does not touch it.

Eggs

Prasad and colleagues screen-washed the intertrappean beds at Kisalpuri, Madhya Pradesh, and recovered "the first definitive Late Cretaceous turtle and crocodilian eggshells from the intertrappean beds".¹⁴¹⁴  The beds also hold freshwater fishes, and the wider Kisalpuri fauna includes mammals.  These are shell fragments, not nests in place.  The famous in-place dinosaur nests of India are in the Lameta beds below the basalts, not between flows, and the two should not be confused.  But eggshell in a lake deposit between two flows still means a breeding population of turtles and crocodilians living on and around that surface.  Reptiles do not nest under a global flood, and they do not nest in the days between two lava flows.

Adding it up

Chenet's own fastest reading gives the thin red boles "less than 10 years" each.¹  Apply that most creationist-friendly number to all 23 boles in the section:  230 years.  Kasbohm's eight red boles at about 100 years each:  800 years.  Clarkia:  840 years.  Bagwanya:  several thousand.  Each is one gap in a province with about 300 flows, and each alone is longer than the Flood.

Part 5: How long one flow takes to cool, and when the creationists think it erupted

The crust-thickness clock

When a pahoehoe flow inflates, its upper crust thickens by conduction, and Hon and colleagues measured the rate on live flows at Kilauea in 1994.  Thordarson and Self state the relation:  "The time it takes to form a crust Hc meter thick can be calculated by the empirical equation t = 164.8 Hc2, where t is time in hours and Hc is the thickness of the upper crust in meters and 164.8 is an empirically determined constant [Hon et al., 1994]."⁶  The relation was checked against a flow whose duration was watched:  "The calculated emplacement time of 323 hours (~2 weeks) for the Kalapana sheet lobe at Kilauea volcano in Hawaii (Table 4) corresponds well with estimated duration of 2-3 weeks from direct field observations".⁶

This article applies that clock to a whole flow.  A flow is solid when the crust growing down from the top meets the crust growing up from the base.  Take the upper crust as half the flow thickness for a lower bound and the whole thickness for an upper bound.  As a check, the second column uses plain conduction, t ≈ (H/2)²/κ, with a thermal diffusivity of 5 × 10⁻⁷ m²/s for basalt.  These are the article's own figures, from textbook inputs, and they are labelled as such.

The two methods agree within a factor of three.  A 30 m flow takes years to solidify.  A 100 m flow takes decades to two centuries.  The Columbia River pile is about 300 flows deep.  If the flows were stacked within a year, then at any moment most of the pile was still liquid inside, which contradicts the "knifesharp" contacts and the solid, jointed, weathered flow tops on which the next flow was laid.

Entablature does not remove the clock

Clarey's answer is that water quenched the flows.  It did.  But the quench has a rate, and the people who modelled it said what that rate is bounded by.  Long and Wood:  "We have found that a water infiltration front can explain the presence of a thick, quenched, flow interior, provided that the infiltration front moves downward just behind the lava solidification front."⁸  Water follows the solidifying lava down.  It does not race ahead of it.  So the flow must already have a solid crust, and be solidifying downward from it, before the water does anything.  Long and Wood also found that a single flow can hold several colonnade-and-entablature pairs, and explained them this way:  "The imposition of boundary conditions that mimic the removal and reoccurrence of flooding yields cooling histories that are consistent with the development of multiple colonnades and entablatures."⁸  The water came, went, and came back while one flow was cooling.  Standing Flood water that never drained cannot do that.  A lava flow that wrecks the local drainage and ponds its own lake, which is what Long and Wood meant by "deranged drainage" and what Höfig documented at Clarkia, can.

Clarey gives no cooling time.  Woodmorappe and Oard give none either.  Their answer to the inflation model is an appeal to authority:  "Stephen Reidel, who researched the CRBG for 25 years, does not accept these arguments (although we believe he would like to, because rapid emplacement poses a major problem for the uniformitarian viewpoint)."¹¹  That is not an answer to an equation.

A paper cited backwards

Woodmorappe and Oard write that "textural evidence, reported by others, indicates that the lava flows responsible for the Columbia River basalts have traveled some 750 km without significant changes in temperature,6 and this implies 'extraordinarily rapid emplacement'."¹¹  The measurement is real.  Ho and Cashman found the Ginkgo flow lost at most 20 °C over 500 km.  But here is what they concluded from it:  "These cooling rates, substantially lower than rates observed in active and historic flows, are inconsistent with turbulent flow models.  Calculated melt temperatures and viscosities of 240–750 Pa ṁ s allow emplacement either as a fast laminar flow under an insulating crust or as a slower, inflated flow."²⁰  A fast turbulent flow churns hot lava to the surface and loses heat quickly.  A flow that stays hot for 500 km must have been insulated by a solid crust.  That crust is the whole physical basis of the slow inflation model.  The datum Woodmorappe and Oard cite for rapid emplacement is the datum its authors used to rule rapid turbulent emplacement out.

Where the creationists put the basalts

The young-earth literature gives three answers to when the flood basalts erupted, and the difference between them is the difference between a problem and no problem.

Austin's 1998 argument is about a trend.¹⁷  Ancient eruptions were far larger than modern ones, and he reads that as "the declining power with time of post-Flood volcanoes."¹⁷  His figure caption says post-Flood in so many words.  A post-Flood Columbia River pile is not bound by 365 days.  It can have its centuries.

Morris in 2004 left both doors open:  "Flood advocates interpret these mega-eruptions of basalt as probably occurring during the very last stages of the Flood or in the years of readjustment to follow".¹⁸

Woodmorappe and Oard put the group inside the Flood, in the Late Abative Phase of the Recessive Stage of Tas Walker's model.  Clarey agrees:  "The Columbia Plateau lavas erupted near the end of the Flood's receding phase (Miocene)."¹⁹

Only the last position has to fit 210,000 km³, 300 flows, ten root-traced soils and an 840-year lake into the closing months of one year.  Austin's position escapes all of that, but at a price.  If the basalts are post-Flood, they are not evidence for the Flood, and the soils, lakes and forests between the flows have to fit into the centuries after it, alongside the ice age, the dispersion from Babel and everything else the post-Flood window already carries.  No creationist source found for this article states the split, chooses between the positions, or says what the choice costs.

Objections and responses

"The flows have flat tops and bottoms, with no indication of erosion or vast time between them." That is Clarey's wording, and it is the direct denial of Part 4.¹⁹  Three teams that looked with a stratigraphic design found root traces, soil horizons, ped structure, peat and land snails, with a chemistry matching modern weathering.³  Roots grow down from a surface that plants live on.  Snails and peat need land, not a sea floor between waves.  Flood water between two pulses can deposit mud.  It cannot grow a forest in it.

"Varves are not years.  A storm can lay a couplet in an hour." True in general, and the lake varves article grants it.  Two answers here.  First, Höfig's team tested this lake and found the coarse layers barren and the fine layers full of deciduous leaves, which is a seasonal signal;  storms stir a lake and destroy both the laminations and the leaves.¹²  Second, the argument does not rest on varves.  Samant's several thousand years for Bagwanya explicitly refuses to count laminations.¹³  The soils and the eggshells do not involve laminations at all.

"The flows cooled fast because they were under water.  That is what entablature shows." Entablature shows water quenching, and the mainstream literature says so.  But Long and Wood's model has the water following the solidification front, not leading it, and it has the water withdrawing and returning during the cooling of one flow.⁸  Neither creationist source gives a cooling time.  The crust-thickness clock gives years for a 30 m flow even before asking what the water does.  A young-earth account needs to show, with a number, that quenching removes the heat orders of magnitude faster than the infiltration-front model allows.  None has tried.  And flow tops that are vesicular and glassy, which Clarey himself describes, formed by degassing at low pressure, which points to air above the flow, not a deep water column.

"Radiometric durations are circular.  Barry admits argon cannot resolve eruption durations." Barry does admit it, and this article quotes him.  But four clocks appear in this article and only one is argon.  U-Pb on zircon from ash beds is a separate isotopic system with separate assumptions, and it gives 800,000 years.¹⁰  Chenet's paleomagnetic clock uses no isotopes and gives 15,000 to 115,000 years for one section.¹  The Clarkia varves are counted.  The soils are read from roots and chemistry.  A reader who rejects radiometric dating outright still has three clocks to answer.

"Chenet says red boles can form in under ten years." For nine thin ones inside directional groups, yes.  The same paragraph gives the two thick ones 1,000 to 50,000 years each, and the next paragraph totals the section at 15,000 to 115,000 years.¹  Quoting the first sentence without the third is the kind of selective citation a skeptic will catch in a minute.  And even the friendly number, ten years times 23 boles, is 230 years.

"Hobbs and Parrish's own clocks contradict each other." They do, and it is fair to say so.  The Grande Ronde erupted in 0.42 million years by argon, yet the soils inside it are read as hundreds of thousands of years each.²¹  One of those readings is too long or the other too short.  Neither is within five orders of magnitude of a year, so the tension helps a young-earth reader only as a reminder that soil-duration estimates carry large uncertainties, which nobody in this article denies.

"The basalts erupted after the Flood, so the year does not apply." This is Austin's position, and it is the strongest one available to the young-earth side.  It should be stated openly as a choice.  Its costs are that the basalts stop being Flood evidence, that Woodmorappe, Oard and Clarey are then wrong about placement, and that an 840-year lake and ten soils of unknown but long duration have to fit into a post-Flood interval whose length the position does not state.

"Erupting that much lava in a year would boil the oceans." Not globally.  The whole heat content of the Columbia River basalts would warm the ocean by 0.16 K, and it is six millionths of the ocean-floor heat that catastrophic plate tectonics already has to remove.  As a global objection this is a mistake and should be dropped.  As a regional objection it is fair, and Part 2 prices it.

"A Flood year is absurd on emplacement alone." Also a mistake.  Individual Roza lobes took months, the whole Roza field about 14 years, and Deccan groups of 20,000 km³ a few decades, by the slowest published models.⁶  The province is a thousand times too big for a year, but any single flow is not.  A skeptic who argues that single flows are slow is arguing against the geology.

"The pillow lavas and palagonite prove nothing." They prove the lava met water, which is also what the mainstream model needs for entablature and what Höfig documented at Clarkia, where a flow dammed a river.  They say nothing about how long the water stayed.  Woodmorappe and Oard are right about the water and silent about the duration.

Issues and unanswered questions

• Three foundational papers were not obtained. Self et al. 1996 and 1997, the founding statements of the inflation model, and Hon et al. 1994, the Kilauea measurements behind the crust-thickness clock, are all paywalled. The clock is quoted from Thordarson and Self 1998, which states it verbatim and credits Hon. The body of Long and Wood 1986, which contains the computed cooling times for entablature, was not obtained either; those numbers would sharpen Part 5 in either direction.
• The cooling and heat computations use textbook inputs. Density, specific heat, latent heat and thermal diffusivity were not traced to a primary. Treat the tables in Parts 2 and 5 as the article's own arithmetic, not as published results. Both were re-derived by a second reader in September 2026; the heat table as first drafted was wrong by a factor of 1,000 and has been corrected.
• The count of interbeds across the whole Columbia River group is not in hand. It is in the closed body of Reidel et al. 2013. With that count the interflow arithmetic could be done for the whole province instead of for single sections.
• Hobbs and Parrish page 1545 was read only through Oard 2017. Their abstract was read directly. The quotation about hundreds of thousands of years of quiescence matches Oard's rendering, and Oard's quotation was verified for the buried soils article, but the page itself has not been opened by this project.
• Sheldon 2003's body is closed. The number of soil profiles he found and any per-profile duration are not in the abstract.
• The Deccan paleosol literature is thin here. Srivastava et al. 2013 on flood-basalt-hosted paleosols is open access but could not be fetched; Dzombak et al. 2020 was not attempted. Either would give the Deccan the soil evidence that Sheldon gives the Columbia River.
• Austin's placement is not as clean as the split makes it look. His title, his figure caption and his lava paragraph all say post-Flood, but one sentence says "during and immediately following" the Flood. He is post-Flood-leaning, not unambiguous. His figure is also drawn from the Yellowstone caldera tuffs, which are younger than the Columbia River basalts, so "declining power" should not be printed without saying what it is measured over.

Verdict

The young-earth side is right that flood-basalt flows are catastrophic, right that water was involved, and right that the global heat is negligible.  On all three points its critics often overstate.  A single flow of 1,300 km³ in 14 years is what the Journal of Geophysical Research prints.

The case fails between the flows.  Root-traced soils with peat and land snails, an 840-year lake with autumn leaves, Deccan lakes priced at thousands of years without counting a lamination, and turtle eggshells between flows all need a land surface that was alive and dry between one eruption and the next.  Four kinds of clock, only one radiometric, agree that each province took tens of thousands to a million years, and the fastest measured eruption rates would still take two millennia to build the Columbia River pile with no pauses at all.

The within-the-Flood reading, held by Woodmorappe, Oard and Clarey, does not survive that.  The post-Flood reading, held by Austin and allowed by Morris, survives the year but gives up the claim that the basalts are Flood deposits, and then owes an account of the soils and lakes that it has not given.  The creationist literature has not noticed that it holds both positions.  The argument is not about how fast lava moves.  It is about how long the land lay still between flows, and every measurement of that says centuries at the least.

Sources