Product has been added to the basket

Still Smoking

Smitsmall.jpgJan Smit comes back for more - read the latest instalment in the Great Chicxulub Debate!

This is Smit's second response, made in answer to Gerta Keller's rebuttal of his criticisms of her original piece. Received 24 November 2003.

The riposte by Keller of November 19 contains a handful of old arguments, to which I will add a few comments, correct a few errors, and provide some new information.

I question the existence of so-called 'strong evidence', in as much as this is interpretation rather than fact. Geological data in the field are often hard to interpret, in particular when dealing with a complex, poorly understood, clastic deposit. In my view Gerta Keller consistently chooses to exploit only one explanation, and often fails to consider the alternatives.


Figure a: Panorama of the clastic beds near Bruselas, between el Penon and el Mulato, looking North. The clastic beds are here sheet-like. At the base of the clastic beds here only one layer of spherules has been observed.

The items (I select the major ones) boil down to:

1. Multiple spherule layers

  • Correlations of spherule layers
  • Hemipelagic sedimentation between the spherule layers
  • Burrows
  • Disconformities and Current directions

2. Chicxulub Yaxcopoil-1 core interval 794.11-794.60m.

  • Foraminifera or dolomite crystals?
  • Sandstone of micrite?
  • Glauconite-smectite?
  • Dual layer US western interior.
  • Overall scenario

1) Multiple Spherule layers.

Keller maintains that the alleged occurrence of multiple spherule layers below the the K/T boundary, imply that those are 0.3 myr older than the K/T boundary. Additional spherule layers were already observed during fieldtrips in 1991 with Pemex geologists (localities Cuauhtemoc, Sierrita (1)) when they showed us these outcrops. We tried to follow these 'layers' but they continue only over short distances, and were slumped.

Correlations of spherule layers

I fail to see from Keller's figure 2 that these layers correlate, especially over distances >30 km and I am afraid many others have that problem too. Occurrence in Zone CF1 is not enough to correlate them layer by layer. In the area Mesa Juan Perez to Loma Cerca, (see Keller fig 1) all outcrops are strongly channellized. In the area between the isolated outcrops, the clastic beds probably do not occur, or are very thin. The presence of thicker beds holds up the mesas and hogbacks. In contrast, in the area from el Porvenir to el Mulato (fig a) where more planar beds are exposed over much longer distances, we have not observed any multiple spherule beds.
Fig.b Figure b: Well-bedded hemipelagic marls of the Mendez formation, 5-30m underneath the clastic beds in the Mimbral outcrop.

Hemipelagic sedimentation between spherule layers?

The Upper Cretaceous Mendez shales show a well developed, probably rhythmic bedding (fig b) where the Mendez is better exposed. This bedding, however, disappears several meters below the clastic beds in many outcrops. Pure coincidence? We don't think so. These beds are fluidized probably due to ground shaking as result of the strong earthquakes from the Chicxulub impact. The fluidization depth differs probably from place to place, but is clearly visible at the Rancho Nuevo locality (fig c), but also in the classic Mimbral locality (fig d), and minor in the Mulato area (fige).
Figure c: K/T clastic beds at Rancho Nuevo. The slightly more greenish, fluidized Mendez fm beds are squeezed up, diapir-like, between deeply loaded channels (B). In the 'diapir' the beds are slumped and folded.
Fig.d Figure d: Fluidized Mendez beds (the purple thin bentonite is folded in slumps) underneath a lateral aggradation face of a loaded channel. Mimbral outcrop. You can view a closeup of the slump structures from Jan Smit's website.
Figure e: Mulato, Tamaulipas. Sheet-like K/T clastic beds, above well-bedded Mendez marls. Here no additional spherule layers have been observed, notwithstanding the excellent exposures. It is also unlikely that these sheets have eroded 12-15m into the Mendez to remove these so-called 0.3 Ma older spherule slayers
Fig.f Figure f: Alternations of several spherule-rich and Mendez marl-clast rich layers, filling a channel, underneath the Unit 2 of the clastic beds at Mimbral. Sandro Montanari (175cm) for scale. Due to the excellent exposures the difference between the (fluidized) Mendez and the Mendez-spherule alternations is well visible. When covered with 'gumbo' (Juan Perez, Sierrita, Penon) this distinction is poorly visible.

We have not observed 'normal' hemipelagic sedimentation between the spherule 'layers'. There are often many layers composed of soft hemipelagic Mendez clasts, that are welded together, sometimes with preservation of internal burrows. These create the illusion of extended hemipelagic layers, as seen in figs f and g. (similar to Kellers figure 6b. Fig 6b is not a 'sandy limestone layer', but instead Mendez marl clasts embedded in spherules). Of course these clasts are geochemically similar, and contain the same foraminiferal assemblages as Mendez marls below, they are reworked from slightly lower levels! The foraminiferal assemblages are consistent with CF1, but I fail to see consistent variations within the CF1 zone that would identify a lower or upper part. If such were the case, I would expect that micropaleontologists would create two zones, which has not happened.
Fig.g Figure g: Spherules, squeezed between Mendez clasts. Those clasts are here fused together. Mimbral, width of photo 20 cm.

In the Mimbral outcrop (fig h), which is one of the few outcrops which is sufficiently well exposed to follow the individual layers, lateral aggradational channel-fill layers alternate between spherule-rich and clasts-rich. Elsewhere, e.g. in the Sierrita area, such channels can apparently reach large sizes and depth, in particular if the channels are deeply loaded into the fluidized marls as seen in Rancho Nuevo (fig c). Those outcrops. However, are all covered by surface gumbo and debris, and therefore such details can be missed.
Figure h: Reconstruction of part of the clastic bed channels at Mimbral, with the fluidization and diapirism of Mendez marl between the spherule bearing channels. Near 2, the Mendez tends to squeeze above the spherule bearing channel rim. (same location as fig d)


A major issue in this debate is the interpretation of burrows and burrow-like structures in the clastic beds at the K/T boundary in the Gulf coast outcrops. As we said earlier we believe that Keller got confused in several ways here. In Figure I we show where the different types of burrows and the burrow-like structures occur in the clastic beds.
Fig.i Figure I: Reconstruction of the burrow fabrics observed at the top of the 7.4m thick clastic beds at el Penon. Note that crab-like Ophiomorpha burrows penetrate up till one meter below the top, a heavily burrowed substrate. The J-shaped 'burrow' occurs near level 0.5m at the bottom of the beds.

We strongly question the occurrence of burrows in the lower part of the clastic beds. One swallow does not make a summer, and one burrow-like structure does not produce a colonized, bioturbated surface.

Let us first try to clean up some confusion in Kellers presentation of the 'burrow' evidence. Can it be that the burrows shown by Keller in Fig 4, fig3bB, fig3bC, fig5b, all depict the same 'J-shaped burrow-like structure'?.

I regrouped all these figures here in one figure J, resized to the same size. Figs 4 and 3bC appear counterparts of figs 3bB and 5b. This way it is easy to multiply the 'evidence' of burrows. This J-shaped feature, is first described by Keller from Unit 2(fig4), then from Unit 1 (fig5b), then from Unit 3 (fig 3bB, subscript). If this feature is derived from Unit 3, (unlikely due to the presence of spherules) then I have no problem with its occurrence, since many burrow types occur in this unit. If correct, it would eliminate all the evidence for the presence of burrows in the lower part of the sequence, where, as I said earlier, my students, colleagues and I were not able to find burrow structures in several fieldtrips.
Fig.j Figure j: Four images of a J-shaped burrow taken from the riposte of Keller of Nov.18 from figures, 3b(B), 3b(C), Fig 4 and fig 5b. The burrow(s) were resized, to show them all at the same size. Figs 3bC and Fig 4 are clearly one and the same, as are fig 3bB and 5b (see crack at the right). Left and right are obviously mirror images, left taken from the hand-specimen and right the counter-part in the field. All four are from the same burrow-like J-shaped feature, described by Keller from units 1, 2, and 3 (see text), about 7m apart.

Even world-renown experts can make mistakes, and Tony Ekdales find during the LPI excursion in 1994 of burrows in unit 2 in the Mimbral outcrop was there shown to be tubes of muddy sand, filled by plant-roots. This is a common feature of caliche soils in the area. Ekdale later referred to these as Holocene 'Rhizocretes' (2), p594), and he added that Unit 1 does not contain burrows in any site, and that in Unit 2 only in Penon a few poorly defined burrows occur, 'which could have been excavated very quickly'. Possibly even between arrival of two tsunami waves, estimated by Bourgeois (3) to arrive an hour apart? Subsequently, no burrows were found in Mimbral in the lower levels of units 1 and 2, nor in any other of the outcrops in eastern Mexico, but burrows do occur in unit 3, near the upper part.

Disconformities and Current directions

Erosional disconformities are used by Keller to argue for prolonged deposition. I fail to see the validity of that argument, because successive turbidites or tsunamites can erode easily into the earlier deposited layers. In fact, it would be a miracle if that would not have happened. I would rather turn the argument around. If there were longer time spans involved, I would expect that normal, hemipelagic sediment like the Mendez, would have been deposited during that time between the sandstone layers, and at least be preserved in some of the 40 localities involved. Yet there is none. The levels Keller consistently calls 'normal pelagic sedimentation' between sandstone layers are, according to the grainsize analysis we performed, more silty than the true Mendez layers below the clastic deposit (fig k).
Fig.k Figure k: Histograms of laser grain size analyses of a Mendez marl sample and one of the silt layers between the burrowed top sandstone layers of unit 3 in Mimbral. The samples were dissolved in acid and treated with permanganate prior to analysis, to remove non-lithic particles such as foraminifers and authigenic pyrite grains. The silt layer is coarser grained than Mendez, and therefore unlikely to be of hemipelagic origin.

At Coxquihui, where Keller mentions a Paleocene hemipelagic layer between two spherule layers, a fault intersects the section.

Current directions measured by us (4, 5) (Fig l) often show current reversals (fig, m). However, the layers showing 'upstream' i.e. N300-N30- directions, are thinner, probably because the backwash current was stronger, longer lasting, consistent with an interpretation as turbidites modified by tsunami waves. It is no surprise that Keller finds 94 out of 97 measurements in S90-S180 directions, because it is much easier to encounter measurable current directions in the much thicker layers showing S90-S180 directions.
Figure L (left) Current direction measurements on climbing ripples on one outcrop face of the Lajilla outcrop. Current direction in 4 is different from the adjacent units 3-5 (see also fig m).
Figure m: Current direction measurements on climbing and linguid ripples on two facing outcrop faces in the Lajilla outcrop. The left panel is the same as fig l. In simple turbidites a uniform, downslope current direction is expected, but that is not the case here. Data from Th. B. Roep.

Flute casts at the base invariably show 'downstream' directions. The best-preserved ripple-beds with current reversals occur in the Lajilla (1 and 2) outrops. The sandstone beds there are far from channellized, and are sheet-like over 400m. Therefore, an explanation for the 180 degree different directions by channel-wall refraction/reflection seems unlikely.

2) Chicxulub Yaxcopoil-1 core interval 794.11-794.60m

Keller's interpretation that the interval 794.11-794.60m, just above the suevitic ejecta is Cf1 in age is based on three items: 1) foraminifers, 2) magnetostratigraphy, 3) late Maastrichtian isotope values.

Points (2) and (3) are not contested, but offer no solution: 2) Chron 29R begins about 550kyr before and ends 250kyr after K/T (6). A reversed polarity of the 45 cm interval says therefore little about its age. 3) The limestone dolomite/limestone could be reworked Maastrichtian, no surprise that these have Maastrichtian isotopic values.

Foraminifera or dolomite crystals?

Fig.n Figure n : SEM backscatter graph of corroded dolomite rhombs (sample 319), which might resemble foraminiferal tests.

(1) I strongly contest the interpretation by Keller that shapes shown in Keller's figure 9, (5-11) and on Keller's website ( are foraminifers. In all the figures and plates shown by Keller not a single unequivocal foraminifer is visible, just fortuitous combinations of smaller and larger dolomite rhombs (fig n shows a SEM backscatter graph of such crystals, strongly corroded). Let alone that these 'forms' can be determined down to the species level, as Keller claims to be capable. I have looked, over 30 years, at hundreds of thin sections from the upper Cretaceous of the Scaglia Rossa of Italy, the SubBetic of Spain and I know how tricky it can be to determine species in thin sections. Even with perfect preservation, and with a lucky cross section species designation can be a gamble I rather abstain from. (fig o) But I am curious to know how other foraminiferal specialists think about these. (Figure obis). Jose Antonio Arz has a split of exactly the same samples Keller has studied, and I have studied thin-sections of the same lithologies, i.e., laminated fine-grained dolomitic sandstone. It is not a question of looking at the right lithologies, they are simply not there, in any of the samples. For sample levels used by Arz, Keller and Smit of the core, see fig 4 of And Yet It Smokes, or visit Smit's website.
Fig.o Figure o: Comparison of Maastrichtian planktic foraminifers in true micrites (D from Lajilla and E from Gubbio, Italy) with the forms from the Yaxcopoil-1 core segment claimed by Keller to be forams (C), and two micrographs from the same levels (310 (A) and 316 (B))
Fig.obis Figure obis: Thumbnails of thin sections from samples 310, 313, 314, 315, 316, all from the laminated intervals of yaxcopoil-1 interval 794.11-794.54m, where Keller claims to have found foraminifers.

Reworked microfossils?

Keller says that the Yucatan platform does not support planktic foraminifers. Yet she mentions that planktic specimens of Cenomanian age are apparently present in Cenomanian rocks of the Yucatan platform. I suppose it is possible that these can rework into the cross-bedded sandstones.

Sandstone or Micrite?

Keller employs a rather strange definition of a sandstone. Apparently she considers a rock a sandstone, when it consist of insoluble lithic grains, such as quartz. Yet sandstones can be composed of any type of coarse grains, including soluble limestone, as is the case here. What is visible in thin sections of the interval are alternations of coarser cross-bedded and finer grained laminated sandstone, composed of dolomite and limestone grains. Those will dissolve if put in acid. At some levels coarser traction carpets occur, loaded with green grains of clay, probably smectite-glauconite. The grains often display internal cavities, which we interpret to be inherited from gas bubbles in impact glass, because similar cavities occur in smectite pseudomorphs of spherules in the Gulf coast K/T outcrops. (fig p) Keller's figure 10 shows the rocks for what they are: sandstone. Fig q shows a picture of the cross-beds of the interval 794.54 794.41m., comparable with the Bouma Tc and Td intervals of turbidites. For more information about the entire interval, see
Figure p: Bubble cavities in glauconite fragments in yax-1 traction carpet (A, 794.41m) and from a glauconitic spherule from Mimbral (B) Inset C whole spherule.

Glauconite or smectite?

Keller argues that the presence of glauconite indicates long-term deposition, but her evidence is far from conclusive.

Impact glass alters to various clay minerals, among which (cheto) smectite, illite, chlorite and glauconite are just a few examples. There are often many types of clay minerals, like Fe-Mg and K-bearing chlorite and illite phases (see Peter Schulte, 8) in the glass alteration phases in northeast Mexico. The green mineral shown by Keller is more typical for poorly crystallized illite or chlorite than the almost opaque dark green color of normal glauconite pellets. In textbooks like Deer, Howie and Zusmann, transformations from smectite to glauconite, a member of the illite group, have been described. The (altered)

Chicxulub ejecta is compositionally very complex and does not only consist of smectite, but also of a large number of different clay-mineral phases derived form a mixture of more acidic and more mafic precursor rocks. Montanari (9) has shown alteration of K/T spherules in the Appenines ranging from smectite (Conero) to glauconite (Contessa), depending on the amount of tectonic stress. There is therefore no reason to interpret glauconite grains as forming at the sediment water interface, they could be alterations of impact glass.

US Terrestrial sections?

Keller argues that until now the sections in NE Mexico and in the Chicxulub crater were considered the most complete. I wonder where that notion comes from, The topmost Maastrichtian is invariably incised and eroded by the base of the clastic beds, and thefore cannot be complete. I argued that the most complete sections are far away from Chicxulub, like those at Caravaca, Agost and Zunaya in Spain and el Kef in Tunisia. Keller also argues that terrestrial sections are condensed and incomplete. Terrestrial sequences in itself are incomplete, as far as the channel and overbank facies are concerned, and those facies do not preserve thin volcanic ash and impact ejecta layers. Coal layers, on the other hand, preserve even the thinnest volcanic ashes. Most coal layers are free from dust and other clastic debris, and in quiet swamp environments the plant material accumulates continuously as long as the swamp remains. It is only in those fossil swamp deposits that we find the dual- K/T impact layers, from Alberta to New Mexico. In the Fort Peck lake area of eastern Montana for instance, the dual impact layer is preserved precisely at a major palynological turnover, in coal swamp deposits in the Hell Creek area, but not in the Bug Creek area, where the same palynological turnover is either eroded by major incised river channels, or occurs just below a coal layer in a overbank deposits. The upper, Ir-rich layer is best correlated with the global K/T anomaly in marine sections, because as far as we presently know (9, fig r) there is no second Ir anomaly in the 10ma straddling the K/T boundary. The lower, spherule rich layer can be correlated with the spherules from the Chicxulub impact. The age and origin of the shocked zircons in the upper layer are well explained by derivation from the panafrican basement of the Chicxulub crater.
Figure q: Core segment 794.41-794.54m of the Yaxcopoil-1 core, showing climbing ripples composed of dolomite sand.


Keller avoids the compelling implications of the dual layers in the K/T clay layer in the US western interior. She draws conclusions from the glauconite in the Yaxcopoil-1 core that are oversimplified. Keller argues for the existence of an abundant Upper Maastrichtian fauna in hemipelagic micrites above the suevite ejecta, where there are only dolomite crystals in sandstone. Keller multiplies artificially 'evidence' to underscore the existence of a bioturbation, colonisation phase in the lower parts of the clastic beds in easterm Mexico, where there is only a single, poorly defined burrow-like structure in el Penon, among the over 40 sections analysed. Multiplets of spherule layers occur at different level only at those places where the clastic beds for deep channels and are deeply loaded into fluidized Mendez oozes.
Fig.5 Figure r: 10 million years of iridium abundance (from 8) across the K/T boundary in the Bottaccione Gorge, near Gubbio Italy. Apart from the iridium anomaly at K/T, no other anomalies are visible. If the large Chicxulub impact would predate the K/T boundary by 0.3Ma, it is unlikely that its iridium anomaly would not have been found in this analysis.

In conclusion, all of Kellers 'evidence', when critically evaluated, evaporates one by one. We still believe that all the features of the K/T boundary, inside and outside the Chicxulub crater, call, with the present knowledge, for just one major impact event at the K/T boundary. And chances are that this is the Chicxulub impact event!


  1. Alvarez, W., Nishimura, J.M.G., Martinez, S.R., Ruiz, L.E., Guzman, R., Zambrano, M.J., Smit, J., Swinburne, N.H.M., and Margolis, S.V., 1992, The Cretaceous-Tertiary boundary impact-tsunami deposit in NE Mexico, Geol. Soc. Am. Ann Meeting: Cincinnati, Geol. Soc. Am, p. A331.
  2. Ekdale, A.A., and Stinnesbeck, W., 1998, Trace fossils in Cretaceous-Tertiary (KT) boundary beds in northeastern Mexico; implications for sedimentation during the KT boundary event: Palaios, v. 13, p. 593-602.
  3. Bourgeois, J., Hansen, T.A., Wiberg, P.L., and Kauffman, E.G., 1988, A tsunami deposit at the Cretaceous-Tertiary boundary in Texas: Science, v. 241, p. 567-570.
  4. Smit, J., 1999, The global stratigraphy of the Cretaceous Tertiary boundary impact ejecta: Annual Review of Earth and Planetary Sciences, v. 27, p. 75-91.
  5. Smit, J., Roep, T.B., Alvarez, W., Montanari, A., Claeys, P., Grajales-Nishimura, J.M., and Bermudez, J., 1996, Coarse grained, clastic sandstone complex at the K/T boundary around the Gulf of Mexico: Deposition by tsunami waves induced by the Chicxulub impact?, in Ryder, G., Fastovski, D., and Gartner, S., eds., The Cretaceous-Tertiary Event and Other Catastrophes in Earth History, Volume Sp. Pap. 307: Special Paper: Boulder, Geol. Soc. of Amer., p. 151-182.
  6. Berggren, W.A., Kent, D.V., III, C.C.S., and Aubry, M.-P., 1995, A Revised Cenozoic Geochronology and Chronostratigraphy, Geochronology, time scales and global stratigraphic correlation, Volume Special Publication 54, SEPM.
  7. Smit, J., and Klaver, G.T., 1989, Multiple impacts at the Cretaceous Tertiary (KT) boundary?: Terra Abstracts, v. 1, p. 421.
  8. P. Schulte, The Cretaceous-paleogene transition and Chicxulub impact ejecta in the northwestern Gulf of Mexico: paleoenvironments, sequence stratigraphic setting and target lithologies. PhD thesis, Karlsruhe 2003
  9. Montanari, A., 1990. Authigenesis of impact spheroids in the K/T boundary clay from Italy: New constraints for high-resolution stratigraphy of terminal cretaceous events. Journal of Sedimentary Petrology, 61(3): 315-339.