Devils Tower and Volcanic Landforms

"The 3-D thermal model that was constructed for the geometry of model D-19 also revealed that the total cooling time in the center of the hot core at the suture between the cooling fronts above the feeding conduit (Fig. SI-1) corresponds to ∼4000 yr." — Prokop Závada and colleagues, on their own lava-coulée model of Devils Tower, 2015¹

"We found that the column radius and stria size are proportional to each other and inversely proportional to the cooling rate of the lava." — Lucas Goehring and Stephen Morris, on the Columbia River basalts, 2008²

"Devils Tower should have been destroyed quickly, surely in less than 100,000 years." — Michael Oard, 2009³

Summary

Devils Tower is a phonolite porphyry body in northeastern Wyoming.  It rises about 600 ft (183 m) above its own talus and almost 250 m above the Jurassic Sundance Formation at its base.⁴¹  Young-earth geologists treat it as a showpiece of Flood runoff:  hundreds of metres of sediment were stripped from around it by receding floodwater, and the Tower has stood, hardly touched, for the 4,500 years since.³⁵  This article asks how long the body took to cool, what the width of its columns says about that, and whether the Tower could have survived the exposure time mainstream geology gives it.

Nine conclusions, stated first.

1. The young-earth claim is an exposure claim, not a cooling claim. Michael Oard's 2009 paper, his 2026 magazine article and Tim Clarey's 2023 ICR article all argue that Flood runoff removed the cover and that the Tower cannot have stood for millions of years. None states how long the phonolite took to cool; the word "cooled" occurs once in Oard's paper, in a sentence about how the columns formed.³ Andrew Snelling's two-volume Earth's Catastrophic Past never mentions Devils Tower, columnar jointing, Ship Rock or the Giant's Causeway.⁶
2. On the young-earth timeline the cooling has to fit inside the receding phase of the Flood. Clarey places the magma "during the receding phase of the Flood after most of the sediments were deposited" and the exposure "4,500 years ago during the draining of the floodwaters."⁵⁵ The joint pattern records the body's original enclosed shape (Part 4), so the Tower had jointed through most of its volume before the runoff stripped it. That is months.
3. Column width is a cooling-rate gauge, and Devils Tower's columns are among the widest known. Derived by Grossenbacher and McDuffie in 1995, measured by Goehring and Morris in 2008, restated by Li in 2023: "slower cooling develops large columns, faster cooling produces small columns."⁷ Lamur's 2018 experiments show the joints open in solid rock inside a fixed window "regardless of the imposed cooling rate,"⁸ so width is the only free variable. The Tower's lower-tier columns are 2 to 3 m across;¹ the Giant's Causeway's are 40 to 50 cm on Tas Walker's figures.⁹
4. The fastest published mechanism gives decades. Goehring and Morris's Péclet relation, calibrated on a Hawaiian lava lake and on water-cooled Columbia River colonnades, turns a 2.5 m column into a jointing front advancing 3 to 5 m a year. Crossing the 180 m jointed section takes 16 to 45 years. To do it in one year the columns would have to be about 8 cm across.
5. Conduction gives centuries at the floor and millennia for any real shape. Závada and colleagues, who argue the Tower was a lava coulée erupted at the surface and never deeply buried, ran a three-dimensional conductive model and report about 4,000 years.¹ Jaeger's 1957 solution for a 250 m sheet gives about 1,000 years to solidify.¹⁰ This article's own model, with every input printed in Part 4, gives 200 to 470 years for a free-standing cylinder with every face held at 20 °C, which no emplacement allows, and 1,000 to 5,000 years for the shapes anyone proposes.
6. The only young-earth sentence about cooling rate is wrong in the direction that matters. Clarey writes that "Fast cooling makes less orderly fracture patterns" and points to the disorganised base.⁵ Fast cooling makes narrow columns, not disorderly ones,¹¹ and Li's field test found that "cooling processes (slower or faster cooling rate) not affect the ordering behavior."⁷ The massive base is where the body cooled slowest, as Robinson mapped it and conduction predicts.⁴¹¹
7. On exposure the young-earth side is arguing with road signs, and it is entitled to. No peer-reviewed exposure age exists. Robinson says "recent erosion";⁴ the park says exposure began 5 to 10 million years ago and cites nothing;¹² its signs have said 40 million and then 1 to 2 million.³ No cosmogenic exposure age has been measured.
8. But at a measured erosion rate the arithmetic favours the long exposure. Bare-rock outcrops erode at a median of 5.4 m per million years worldwide, and none of 450 exceeds 140.¹³¹³ At the median, 5 to 10 million years costs 27 to 54 m of a 114 to 152 m radius; 40 million costs 216 m and destroys it. Oard's objection defeats the old sign, not the current figure. And "never in recorded history has anyone seen a massive column fall"¹⁴ answers his freeze-thaw argument.
9. The cover was a few hundred metres. Robinson: the Tower "was surrounded and probably covered by several hundred feet of sedimentary rock."⁴ Závada: about 1,850 m came off the Black Hills monocline 10 km west, mostly in the 16 million years before the phonolite was emplaced.¹¹ The park's "one to two miles"¹² and Oard's "more than 2 km"¹⁵ describe a burial that no paper defends.

Verdict.  The exposure argument is the young-earth side's real contribution, and it lands on a mainstream number that has never been measured.  The cooling argument was never made, and when it is made it goes the other way:  wide columns are slow-cooled columns, and the slowest cooled cannot fit in a Flood year.  Decades by the fastest mechanism, millennia by the one everyone models.  Ship Rock, with a kilometre of cover removed, is the harder case of the same kind, and no young-earth author has taken it.

Part 1: What Devils Tower is

The measurements

Charles Robinson mapped the monument for the US Geological Survey in 1956:⁴  "Devils Tower rises steeply for about 600 feet from a broad talus slope at its base.  The top of the Tower, at an altitude of 5,117 feet, is about 1,270 feet above the Belle Fourche River.  The Tower is about 800 feet in diameter at the base."⁴  The top is nearly flat, about 180 by 300 ft.⁴  Prokop Závada's team, measuring in 2015,¹ put the summit at 1,558 m above sea level, "almost 250 m above the underlying sedimentary rocks of the Sundance Formation," on an elliptical base of 304 by 228 m with a summit plateau of 90 by 55 m.¹  "Steep columnar jointing characterizes the upper 180 m of Devils Tower."¹  Below that is the base, cut by three joint sets into blocks about 3 m across.¹  Robinson's description of it:  "At the base of the tower, below the bench, the rock is massive and jointing, poorly developed.  Here the joints form large irregularly shaped blocks rather than columns."⁴

The columns.  "Most of the columns are 5 sided, but some are 4 and 6 sided.  The larger columns measure 6 to 8 feet in diameter at their base and taper gradually upward to about 4 feet at the top."⁴  Závada sees two tiers that merge about 70 m below the summit.  The lower tier at the shoulder "consists of columns of the lower tier with diameters of 2–3 m.  These columns taper to widths of only 1.5 m at the contact with the upper tier, where they frequently merge into one thick column (> 3 m wide) of the upper tier."¹  He calls the Tower "the world's finest example of columnar jointing in phonolite and possibly the longest columns developed in a volcanic rock."¹  The park's own page¹² says the size of the columns "is unmatched."¹²  Clarey's "column widths (10 feet)"⁵ is the widest merged upper-tier column quoted as typical.  The measured lower-tier width is 2 to 3 m, and that is the number this article uses.

The rock.  "The rock making up Devils Tower is classified as phonolite porphyry (Darton and O'Harra, 1907, p. 6) and is of Tertiary age."⁴  Phonolite is alkali-rich and silica-poor, and it is not basalt.  Závada gives the melt viscosity of "Phonolite magma of 850 °C with <1 wt% H₂O" as about 10⁸ to 10⁹ Pa·s,¹ six to seven orders of magnitude above the 100 Pa·s Delaney and Pollard estimate for the minette at Ship Rock.¹⁶  Oard's "a volcanic rock (originally lava) similar to basalt"¹⁵ is wrong on composition, and it matters:  a magma that stiff piles into a thick dome instead of spreading into a thin flow that chills in weeks.

The age.  Závada reports that "⁴⁰Ar/³⁹Ar dating of sanidines in phonolites from Devils Tower and Missouri Buttes provides indistinguishable plateau ages of 49.04 ± 0.16 and 49.24 ± 0.28 Ma, respectively (Duke et al., 2002)."¹  Duke's 2002 abstract could not be obtained for this article, so the ages are quoted at second hand.  Oard's "33 to 55 Ma"³ comes from a 1995 monument booklet and was seven years out of date when he wrote it.  Nothing in this article's cooling argument depends on the radiometric age.

The four origin models

"The scientific debate on the origin of Devils Tower has lasted for more than 100 years."¹  Závada's summary:  "The majority of previous studies of Devils Tower have concluded that it is a remnant of an intrusive body in the form of a magmatic stock (Russell, 1896;  Robinson, 1956), a laccolith (Pirsson, 1894;  Jaggar, 1901), or a volcanic conduit (Carpenter, 1888;  Halvorson, 1980);  the latter interpretation is presented in many Earth science textbooks.  Although conclusive evidence is lacking, it is generally accepted that it represents an intrusion of phonolite magma that was exposed after erosion of the surrounding weakly consolidated Mesozoic and Cenozoic sedimentary layers."¹

Robinson's conclusion was a shallow stock "which was never much larger in diameter than the present base of the Tower, and which at depth (1,000 feet or more) is connected to a sill or laccolith type body."⁴  Among his seven grounds:  the mineral texture is that of a shallow intrusive rock, and "No evidence of extrusive igneous activity has been found in the surrounding area,"⁴ which is his case against the volcanic neck.  He added that "no conclusive proof of its mode of origin can be presented."⁴

Závada's team tested the shapes against the columns, because columns grow perpendicular to the cooling isotherms and each shape predicts a joint pattern.¹  A conduit "should develop two tiers of horizontal columns in vertical cross section or a chevron pattern of columnar joints," a stock a radial rosette, a laccolith two fans.¹  The Tower's columns are near vertical and converge upward, giving "a rough match only for the laccolith," with the convergence the wrong way,¹ and a close match to a plaster analogue of a lava coulée squeezed into a maar crater:  "Our results indicate that Devils Tower is a remnant of a coulée or low lava dome that was emplaced into a broad phreatomagmatic crater at the top of a maar-diatreme volcano."¹  They could not sample inside the monument and say the hypothesis "requires testing and further analysis" by geophysical surveys.¹  The park lists all four models and calls the origin unsettled.¹²

That is not "even the scientists disagree, so anything goes."  All four models are the same volume of the same rock at the same temperature cooling by conduction into the same country rock.  The shape moves the cooling time by a factor of a few (Part 4), not by the factor of a thousand between the mainstream numbers and a Flood year.

The relatives

Missouri Buttes, 5 km northwest, are "a group of five bodies of similar composition" and the same age.¹

Ship Rock in New Mexico is the volcanic throat the textbooks pair with the Tower.  Delaney and Pollard mapped its radial minette dikes for the USGS in 1981:¹⁶  "emplaced approximately 30 m.y. ago,"¹⁶ and for the depth of emplacement, "Between the extremes of 750 and 2,500 m, we accept 1,000 m as a good estimate."¹⁶  A kilometre of rock has been removed to expose them.

Devils Postpile in California is the clean control.¹⁷  About 100,000 years ago a basalt flow ponded in the valley of the Middle Fork of the San Joaquin River "to an exceptional depth of approximately 122 m (400 ft),"¹⁷ grew columns 12 to 18 m high, and then "glacial erosion removed approximately half of the flow, leaving the thick center exposed,"¹⁷ about 10,000 to 12,000 years ago.¹⁷

The Giant's Causeway in Northern Ireland is the young-earth movement's favourite columnar site, and Tas Walker gives its numbers:⁹  "Columns are 40–50 cm (15–20 in) across,"⁹ in flows "commonly up to 30 m (100 ft) thick."⁹

Part 2: The young-earth position, in its own words

Michael Oard's 2009 paper in Journal of Creation is the primary.³  Its abstract:  "Devils Tower, Wyoming, is likely the conduit of an eroded volcano, but there are three other hypotheses for its origin.  Regardless, more then [sic] 300 m of High Plains sedimentary rock was eroded with the Tower hardly touched.  The uniformitarian story, as formerly stated on a road sign north of the Tower, is that erosion of the High Plains sedimentary rocks took more than 40 Ma.  That sign has been replaced, and it now says it took only 1 to 2 Ma.  However, the erosion of such a vertical tower should be rapid and complete well within 100,000 years.  Although the Tower is actively eroding today, it has not decreased much in size, implying a very recent exposure."³

He picks the geometry:  "The first hypothesis, that of a volcanic neck, is the most reasonable explanation."³  He states the puzzle:  "But, this requires Devil Tower to remain with little change in its diameter or height for 40 Ma!  How could both hard sandstone and soft shale from the High Plains be eroded without any significant erosion of the Tower itself?"³  His destruction mechanism is freeze-thaw along the joints, supported by a 1954 recollection, quoted from the monument booklet by Robinson and Davis, of blocks crashing onto the talus on cold nights, and his conclusion from it:  "Devils Tower should have been destroyed quickly, surely in less than 100,000 years."³  Then the Flood:  "It seems that the only way to explain Devils Tower is to allow for the rapid erosion of the High Plains sedimentary rocks by a wide sheet of flowing water, leaving behind an erosional remnant of the lava conduit.  This is consistent with sheet flow erosion as the floodwater was draining off the continent.  The Tower remained tall after the Flood probably because the rock from the Tower was more resistant and/or the current erosion rates were reduced in the area."³  His analogue is Steamboat Rock, the 275 m basalt remnant left in the Grand Coulee by the Lake Missoula flood.³

Oard's 2026 article in Creation¹⁵ repeats the case, frames the Tower as an inselberg, and enlarges the cover:  "In fact, some geologists think that a thickness of more than 2 km (1¼ mi) has been eroded from the region.  Envision a huge volume of lava moving up this 300-m-diameter (1,000 ft) conduit to erupt on the sandstone surface 2 km above the present Tower."¹⁵

Tim Clarey's 2023 ICR article takes a different body.⁵  He accepts Závada:  "Prokop Závada and his colleagues formulated the latest idea concerning the tower's formation.  They proposed the tower was a maar-diatreme volcano."⁵  He argues the mechanism needs the Flood:  "This process would only work with freshly deposited sediments that were still uncemented.  Similar superheated water at Yellowstone doesn't blow out a big hole since the rocks above are already hardened.  So, if the tower did form as a maar-diatreme, only the conditions of the Flood could have allowed its development."⁵  He dates the magma:  "ICR scientists determined that these magmas formed during the receding phase of the Flood after most of the sediments were deposited."⁵  He names the cover:  "The best explanation is that all of the sediments that once covered the magma intrusion were eroded during the final recession of the Flood.  These missing strata included layers deposited in the Zuni Megasequence."⁵  And he gives the exposure argument a number:  "The base of Devils Tower is piled with broken columns up to 150 feet deep,"⁵ and "if Devils Tower was exposed 5 to 10 million years ago, as conventional scientists claim, it probably would have completely eroded away in this amount of time.  At the very least, it should be entirely buried in its own debris of broken columns. … The relatively thin buildup of broken columns at the tower's base fits its recent exposure 4,500 years ago during the draining of the floodwaters."⁵

Clarey is also the only young-earth author who says anything about cooling rate, in one paragraph:  "Studies suggest all of these columns cool from the top, with contraction forming regularly spaced fractures that propagate downward.  During the Flood, groundwater percolating along the open fractures likely caused the lower part of the magma to cool quickly.  Fast cooling makes less orderly fracture patterns.  We see this at the base of the tower where the columns become chaotic and disorganized."⁵

That is the whole young-earth position, and three things follow.  First, the two authors have chosen different bodies.  Oard needs a volcanic neck left standing by runoff;  Clarey adopts the coulée in a maar.  Robinson argued against the neck in 1956 and Závada's isotherm test rejects it, and the shape sets the cooling geometry.

Second, neither states a cooling time, and no other young-earth source does.  The young-earth Causeway literature, Walker's 2006 article and Angus Kennedy's 2014 open letter,¹⁸ argues about whether the laterite bed between flows is a slow tropical soil or a fast hydrothermal alteration.  Kennedy, a quarry geologist with drill logs, reports "very good evidence for rapid laterite formation."¹⁸  That is real field work, and it never prices a column.

Third, the timeline is tight by the young-earth side's own choice.  The phonolite cuts Jurassic and Cretaceous strata that every young-earth model assigns to the Flood, so it is not pre-Flood.  Clarey's magma formed in the receding phase;  his Tower was exposed by the draining floodwaters.  Oard's conduit was exposed by the same runoff.  On either account the phonolite went from magma to a jointed body that runoff could strip around, all inside the receding phase of the Flood year.

Part 3: How columnar joints form, and what their width says

Robinson gave the physics in the USGS bulletin:  "Columnar joints form as the result of contraction within a rock mass.  In igneous rock the contraction is the result of cooling;  that is, the cold solidified rock requires less volume than the same rock when molten.  As a rock cools it contracts, and the resulting tension is in a plane parallel to the cooling surface.  When rupture takes place, three fractures radiate from numerous centers in the plane parallel to the cooling surface.  Ideally, the fractures are at 120° to each other."⁴

When the joints open

Anthony Lamur and colleagues cooled Eyjafjallajökull basalt under load in 2018⁸ and watched it crack:  "we show that contraction during cooling induces stress build-up below the solidus temperature (980 °C), resulting in localised macroscopic failure between 890 and 840 °C."⁸  The window did not move with the rate:  stress built "from the solidus temperature down to temperatures of 890–840 °C, regardless of the imposed cooling rate and initial stress."⁸  "We demonstrate that columnar jointing takes place well within the solid state of volcanic rocks, and is followed by a nonlinear increase in system permeability of <9 orders of magnitude during cooling."⁸  Columns are not a quench feature;  they form in solid rock below the solidus and keep opening as it cools,⁸ and since the temperature at which they form is fixed, the joint spacing is what records the rate.  The experiments are on basalt;  phonolite's dry solidus is lower, so the window transfers only approximately.

How wide they grow

Grossenbacher and McDuffie built the conductive model in 1995:¹¹  "Much evidence exists that column diameter and stria width of columnar joints in lava flows vary inversely with cooling rate (∂T/∂t) and thermal gradient (∂T/∂x) at the point of fracture as the flow cools," with "faster cooling leading to narrower columns."¹¹  Their model makes a prediction:  "if conductive cooling dominates, then joint spacing (i.e., column diameter) should increase inward from the margins of the body, as should stria width."¹¹

Goehring and Morris measured it in 2008 across the Columbia River Basalt Group:²  "We found that the column radius and stria size are proportional to each other and inversely proportional to the cooling rate of the lava."²  They reduced the physics to the Péclet number Pe = vR/κ, with v the speed of the solidification front, R the column radius and κ the thermal diffusivity, and from the striae found "Pe = 0.3 ± 0.1,"² with R the radius of the cylinder equal in area to a hexagon of side ⟨w⟩, "R ≃ 0.91 ⟨w⟩."²  They checked it against Hardee's borehole record at the Kilauea Iki lava lake, "a steady advance rate of 6.7 × 10⁻⁸ m s⁻¹,"² where the columns are "1–2 m in radius,"² and they claim generality:  "we believe that the conclusions we report will be generally applicable to all igneous columnar joints."²

Yongquan Li's 2023 survey⁷ of columnar joints in intermediate and felsic lavas in China, the rocks nearest phonolite in viscosity, restates the rule, "slower cooling develops large columns, faster cooling produces small columns,"⁷ and gives a recipe for the time one column-width of rock takes to cool, "t = L²/α," with L the side length of the column, about half its width, and α the thermal diffusivity;⁷ a rhyolite column of 20.6 cm side gives 15.7 hours.⁷  Li's paper is a preprint, used here for the recipe and not as an authority.

Two cautions.  Goehring and Morris's Péclet regime is crack-assisted cooling, with water and steam refluxing in the open joints, which is why their calibration site is a lava lake and their field area the water-quenched Columbia River basalts.  Applying it to Devils Tower gives the fastest defensible cooling, not the likeliest.  And Li warns that "an accurate estimation of cooling rate based on column size is usually complex."⁷  The numbers in Part 4 are order-of-magnitude, and the argument needs only that.

What the Tower's columns say

Width 2 to 3 m in the lower tier, narrowing to 1.5 m at the tier contact and widening downward into the massive base.  That is the pattern Grossenbacher and McDuffie predicted for conductive cooling, spacing increasing inward from the margins, and the one Robinson drew, columns "flaring out and merging to form the massive base."⁴  Against the Causeway's 0.45 m the Tower's columns are five to six times wider, and Part 4 turns that into time.

Now Clarey's sentence, "Fast cooling makes less orderly fracture patterns."  Nothing in the measurement literature says so.  Grossenbacher and McDuffie and Goehring and Morris say fast cooling makes narrow columns;  Lamur's experiments say the failure temperature does not depend on the rate;  Li tested regularity directly:  "cooling processes (slower or faster cooling rate) not affect the ordering behavior."⁷  The disorganised base is the massive zone Robinson mapped, where jointing is "poorly developed,"⁴ and in Závada's model it sits over the feeding conduit, the last part of the body to cool.  Clarey has read the slowest-cooled part of the Tower as the fastest.

Part 4: The cooling time, computed

No young-earth source computes this, so it is computed here.  Every input is named so a reader can redo it.

Three routes that already exist

Závada's own model.  Their finite-element runs were "thermal models of conductive cooling" from "An initial temperature of 850 °C … slightly higher temperatures than the dry solidus of phonolite magmas," on the assumption that the magma "was already devolatilized."¹  The best-fit shape is a coulée in a 1 km wide crater, and its three-dimensional run gave "∼4000 yr" for the hot core above the conduit.¹  The thermal constants are in a supplementary file that was not obtained.  This is the friendliest geometry there is for a fast answer:  a body at the surface, cooling into air on top, that its authors argue was never buried.  It is 4,000 years.

Jaeger's solution.  J. C. Jaeger solved the cooling of a sheet in an infinite medium with latent heat in 1957,¹⁰ and for a granite sheet D metres thick, "Solidification is complete after 0.016D² years."¹⁰  The granite-cooling article works through it.  For D = 250 m that is 1,000 years;  for the 180 m jointed section alone, 520 years.  Jaeger's constants give a diffusivity of 6.6 × 10⁻⁷ m²/s;  at Delaney and Pollard's 5 × 10⁻⁷ the times stretch to 1,320 and 690 years.  This is time to the solidus, 500 °C in Jaeger's case, with the country rock still hot around the sheet.

The column-width floor.  The Péclet relation is the fastest published mechanism, water in the cracks included.  A 2.5 m column has a side of about 1.25 m and R ≈ 0.91 × 1.25 = 1.14 m.  Pe = 0.3 ± 0.1 gives v = Pe κ / R = 0.9 to 1.8 × 10⁻⁷ m/s, about 2.8 to 5.5 m a year.  That sits beside Hardee's Kilauea Iki rate of 6.7 × 10⁻⁸ m/s for columns of 1 to 2 m radius, which is the check that the arithmetic is calibrated.  A front crossing the 180 m jointed section from above and below, 90 m each way, takes 16 to 32 years;  a lower front that must also cross the shoulder and base, 125 m, takes 22 to 45 years.  Li's recipe gives 36 days per column-width, about 10 years per hundred column-widths, the same order.  For the front to cross 125 m in one year at Pe = 0.3, R would have to be 3.8 cm:  columns about 8 cm across, the scale of the finest entablature.  The Tower's are 2 to 3 m.

This article's model

The script research/cooling-model.ts runs one-dimensional conduction with latent heat by explicit finite differences on the inputs above, and reports when the hottest point in the body drops below the solidus, 500 °C, 300 °C and 100 °C.  Three boundary cases bracket every proposed shape, and a product solution for a finite cylinder gives the floor.

The both-faces and free-standing-cylinder rows are not physical:  a body with every face at 20 °C from the moment of emplacement would have to stand free in cold water from day one, which no model on either side describes.  They show the floor, 200 to 470 years.  The infinite-medium rows never reach 100 °C because the rock around the sheet stays warm;  in reality the ground surface a few hundred metres up drains that heat over thousands of years.  The coulée at the surface, Závada's geometry in one dimension, gives 1,600 years to 300 °C and 5,200 to 100 °C, bracketing his 4,000.  Jaeger's sheet gives 1,500 years to 500 °C against his 1,000 to 1,320.  Three routes agree within a factor of two.

The honest statement is a floor and a ceiling.  The fastest mechanism anyone has published, calibrated on water-cooled basalt, gives decades.  Pure conduction gives 200 to 470 years for a geometry no one proposes and 1,000 to 5,000 years for the ones people do.  Nothing gives a year.  On Clarey's timeline the budget is the receding phase, a few months.  The shortfall is a factor of 30 to 100 by the friendliest arithmetic and a factor of thousands by the models both sides cite.

Why the cooling had to come before the stripping

Závada's case for the coulée is that the observed columns match the isotherms of the original enclosed body through the whole jointed section.¹  Clarey accepts that model.  Had the sediments been swept away while the interior was still above the jointing temperature, the inner columns would have grown perpendicular to the new free surface, not to margins that no longer existed, and the match would not exist.  The same holds for Oard's neck, whose columns grow inward from the conduit walls.  On any model the Tower had jointed through most of its volume before the erosion that exposed it, and on the young-earth timeline that erosion is the Flood runoff.

Part 5: The exposure

How much cover

Robinson:  "The exposed portion of the Tower is the result of recent erosion.  At the time of its intrusion it was surrounded and probably covered by several hundred feet of sedimentary rock."⁴  The section exposed around the monument is "in aggregate about 400 feet thick."⁴  Oard:  "more then [sic] 300 m."³  Závada gives the regional figure:  "The combined thickness of eroded layers from the Mowry Shale (youngest Cretaceous member exposed at Missouri Buttes) to the Fox Hills Formation in the Black Hills monocline ∼10 km west of the map area is ∼1850 m."¹  But most of that came off first, in "a sufficient time span (16 m.y.) for erosion between the onset of the Black Hills uplift to the emplacement of the phonolite bodies."¹  The youngest rock the Buttes cut is the Cretaceous Mowry Shale, so the land surface at emplacement stood a few hundred metres above the Sundance, and the post-emplacement stripping around the Tower is a few hundred metres.

The park says "When the Tower formed around 50 million years ago, it was one to two miles below the Earth's surface.  Somewhere between 5 and 10 million years ago, erosive forces began to expose the Tower."¹²  The page cites nothing.  Oard's "more than 2 km" is the same number.  Both Robinson and Závada put the cover at hundreds of metres, and that is what the young-earth author has to explain, not 2 km.

When, and by what

Mainstream geology has no measured exposure age for Devils Tower.  Robinson wrote "recent erosion."  Závada dates the emplacement and the earlier stripping and not the exposure.  The park's 5 to 10 million years is uncited;  the road sign said 40 million, then 1 to 2 million.  No cosmogenic nuclide study of the summit, the talus or the surrounding surface was found.  Oard's complaint that the number keeps changing and rests on no measurement is correct.

Could it survive

Oard's argument needs a rate, and he gives none.  The rate that exists is the global compilation of cosmogenic beryllium-10 erosion rates by Portenga and Bierman:¹³  bare-rock outcrops erode at a median of 5.4 m per million years and a mean of 12,¹³ and "Compiled outcrop erosion rates are slow and do not exceed 140 m Myr⁻¹."¹³  The erosion-rates article discusses that compilation.  Applied to the Tower's radius of 114 to 152 m:

At the median rate the Tower loses a quarter to a half of its radius in 5 to 10 million years and stands, consistent with Robinson's judgement that the body "was never much larger in diameter than the present base."  At 40 million years it is gone.  Oard's objection is right against the old sign and wrong against the current figure, and the mainstream account needs the Tower to erode like a typical outcrop rather than the fastest measured.  Phonolite at 1,500 m in a freeze-thaw climate is not obviously typical, and no one has measured it.

The freeze-thaw anecdote has an answer in the park's own pages:¹⁴  "Small pieces, baseball or basketball size, fall regularly.  Chunks the size of a microwave fall rarely.  Yet never in recorded history has anyone seen a massive column fall.  Many pieces are covered in soil and plants, indicating they have been in that location for a very long time."¹⁴  One boulder near the car park, 6 by 8 ft, is "mostly buried by soil.  It likely fell thousands of years ago (or longer)."¹⁴

The talus, for scale only.  No model claims these numbers.  The boulder field is 13 acres;¹⁴ Clarey puts the broken columns at "up to 150 feet deep";⁵ Robinson describes blocks "as much as 8 feet in diameter and 25 feet long."⁴  Take 13 acres at a mean depth of 10 to 23 m:  0.5 to 1.2 million m³, against about 4 million m³ for the Tower above the talus.  Over 4,500 years that is 120 to 270 m³ a year, one column 2.5 m across and 30 m long every year or two, roughly 75 to 150 column-scale falls since 1875.  Over 5 million years it is 0.1 to 0.24 m³ a year, a few dozen basketballs.  The recorded rate looks like the second.  Two caveats:  talus blocks weather away, so over millions of years the present pile is a lower bound on what fell;  and a fresh face spalls fastest at first, a reply no young-earth source makes but a fair one.  A beryllium-10 exposure age on the summit and the largest blocks would settle it, and none exists.

Steamboat Rock, Ship Rock and the Postpile

Oard's analogue is Steamboat Rock, left when the Lake Missoula flood "eroded through 275 m of basalt lava in a matter of days."³  Steamboat Rock stands in a channel cut by an outburst flood from a known lake, with scabland around it.  Devils Tower stands on the High Plains with no channel, no scabland and no scour reported by anyone.

Ship Rock is the hard case.  Delaney and Pollard's kilometre of removed section is ten times the Tower's, in a region with no outburst-flood channel, and the radial dikes, 2 m thick and 3 km long, stand intact while a kilometre of shale around them is gone.  No young-earth treatment of Ship Rock was found for this article.

Devils Postpile is the control:  a flow of known thickness, 122 m, columns 12 to 18 m tall, cut in half by a glacier 10,000 to 12,000 years ago.  Its columns are far narrower than the Tower's in a body half as thick, as the width scaling predicts, and its exposure took a glacier and thousands of years.

Objections and responses

"Geologists cannot even agree what Devils Tower is, so no cooling time can be trusted." The four models are four shapes for the same volume of the same rock at the same temperature.  Part 4 brackets them, from a few hundred years to 5,000, and the fast end is a case no model proposes.  The disagreement is about shape, not about whether the answer is years or millennia.

"The 49 million year age assumes constant decay rates." The cooling argument does not use it.  It uses column widths measured with a tape, thermal constants measured in a laboratory, and heat conduction.  The exposure argument does not use it either.

"Water in the joints cooled the Tower fast.  Clarey says so." Water in the joints is what Goehring and Morris modelled and calibrated.  Their Péclet number comes from a water-flooded lava lake and from the water-quenched Columbia River colonnades, and it is the decades figure.  Lamur's experiments give the sequence:  joints open at 840 to 890 °C and permeability then rises by up to nine orders of magnitude, so the water cannot enter before the crack front has passed.  The granite-cooling article prices hydrothermal cooling with Cathles' and Lister's numbers at a factor of tens over conduction at best.  A faster mechanism would need a faster front, and a faster front makes narrower columns.  The Tower's columns are among the widest known.

"The Tower cooled after the Flood, standing in the open for centuries, and the columns grew then." Then the columns would be perpendicular to the exposed surface, not to the margins of an enclosed body, and Závada's pattern match, which Clarey adopts, would not exist.  The joints record cooling inside the original enclosure.  This escape route also gives up Clarey's own reading, that the base cooled quickly by Flood groundwater, and Oard's, that the conduit was stripped by runoff as a finished rock.

"The mainstream exposure age is unsupported, so 4,500 years is as good a guess as 5 million." Half right.  There is no measured exposure age on either side, and this article says so in the summary.  But the two guesses make different predictions.  At measured outcrop erosion rates, 5 million years costs 27 to 60 m of radius, which the Tower can afford.  A 4,500 year exposure requires the talus to have been shed at roughly a column every year or two, and the recorded fall rate is basketballs.  The test that decides it, a cosmogenic exposure age on the summit and the largest blocks, has not been run by anyone.

"Jaeger's constants are for granite, and Závada's thermal constants were never read." True, and the Part 4 table shows what they are worth.  Doubling the diffusivity halves the time;  dropping latent heat cuts it by a third;  Jaeger's 0.016 D² and the finite-difference model differ by less than two.  None of these moves 1,000 years to one.

"The Péclet relation is for basalt, not phonolite." The relation is dimensionless and the diffusivities of basalt and phonolite are within a factor of two.  Li's survey of intermediate and felsic lavas finds the same inverse rule.  Even if phonolite were ten times off, decades become years, not months, and conduction still says millennia.

"Oard's paper is about erosion.  Criticising it for not computing cooling is unfair." His paper is fairly summarised in Part 2 and his erosion argument is granted what it earns in Part 5.  But the catalogue question for this article is whether the Tower fits inside the Flood year, and the two halves cannot be separated:  on his own account the runoff stripped the sediment from around a finished, jointed conduit.  A jointed conduit is a cooled one.

Issues and unanswered questions

• Duke et al. 2002, the ⁴⁰Ar/³⁹Ar ages, is a GSA abstract with no online copy; the ages are quoted through Závada.
• Hetényi et al. 2012, the field compilation of column widths by rock type, could not be opened by any open route (four repository links dead or blocked). It would sharpen the Causeway comparison.
• Grossenbacher and McDuffie 1995 was read as abstract only.
• Robinson and Davis 1995, the monument booklet Oard quotes for the 1954 freeze-thaw recollection and the "appreciably larger" sentence, was not obtained. Those two passages are quoted here as Oard gives them and have not been checked.
• Závada's supplementary file holds the thermal constants of the 4,000 year model. The main text gives only the initial temperature and the mechanism.
• No cosmogenic exposure age exists for the Tower, its talus or the plain around it. This is the single measurement that would settle the exposure dispute, and it is cheap.
• The talus arithmetic rests on a guessed mean depth. It is labelled for scale and should be replaced by a survey.
• Ship Rock has no young-earth treatment that could be found, and its kilometre of removed cover is the harder version of the Devils Tower problem.
• Závada's proposed gravity and magnetic survey would fix the shape of the body and with it the cooling geometry.

Verdict

Devils Tower is a 250 m body of phonolite that reached 850 °C and stands in columns 2 to 3 m wide, among the widest known.  Column width is a cooling-rate gauge that three groups have derived, measured and restated, and it reads slow.  The fastest published mechanism, water refluxing in the joints, gives 16 to 45 years for the jointing front to pass through the body.  Conduction gives 200 to 470 years for a free-standing cylinder that no one proposes, 1,000 to 1,300 years for Jaeger's sheet, and 4,000 years for the coulée model of the authors who argue the Tower was never buried.  On the young-earth timeline the body had to be jointed before the receding Flood stripped it, inside a few months.  No young-earth author has published a cooling time, and the one sentence about cooling rate in that literature reads the slowest-cooled part of the Tower as the fastest.

The exposure argument is different.  Oard is right that the mainstream exposure age has never been measured and has been quoted at 40, then 1 to 2, then 5 to 10 million years.  But at the only measured erosion rates that exist the Tower survives 5 to 10 million years and not 40, so the objection defeats the old sign and not the current figure;  the recorded rockfall, basketballs and never a column, fits a long exposure better than a short one;  and the cover was a few hundred metres on both Robinson's and Závada's reading, not the 2 km Oard attacks.  Ship Rock, with a kilometre removed, has not been addressed by the young-earth side at all.

Verdict:  the Tower's columns are evidence against a Flood-year cooling by a factor of thirty at the friendliest arithmetic and thousands at the standard one.  The exposure question is open on the evidence, favours the long answer on the arithmetic, and could be closed by one beryllium-10 study.

Sources