Milankovitch Cycles and the Age of the Earth


Sean D. Pitman, MD

© July 2006




Milutin Milankovitch was a Serbian engineer and meteorologist – born in 1879 he attended the Vienna institute of  technology graduating in 1904 with a doctorate in technical sciences. He then went on to work in the University of Belgrade where he spent time working on a mathematical theory of climate based on the seasonal and latitudinal variations of solar radiation received by the Earth.

Milankovitch proposed that the changes in the intensity of solar radiation received from the Earth were effected by three fundamental factors.  The first is called eccentricity, a period of about 100,000 years in which the nearly circular orbit of the Earth changes into a more elliptical orbit. The next factor is called obliquity, a period of about 41,000 years where the Earth’s axis tilt varies between 21.5 and 24.5 degrees.  The final factor is called precession, a period of approximately 23,000 years where the Earth’s axis wobbles like a spinning top.



Milankovitch proposed that these regular cycles of the Earth, as they changed the Earth’s relationship to the Sun, had an effect on the Earth’s climate, driving hot and cold cycles, to include the ice ages throughout ancient history.

The orbital calculations are thought be very accurate back several million years. And, they are thought to match markers of temperature variations within layered structures, such as layered ice in Greenland and Antarctica and layered deep sea sediments. Measurements of changes in 18O vs. 16O ratios within these layers are thought to correlate with the Earth’s climate over the course of hundreds of thousands and even millions of years of Earth’s history.




Devils Hole

For many years, Milankovitch Theory (MT) was very popular and generally still is within the mainstream scientific community.  However, fairly recently, several fundamental challenges to the validity of MT have arisen. Perhaps one of the first significant problems was noted by Wallace Broecker in his short paper published in a 1992 issue of Nature.  Broecker wrote:

“One of the fundamental tenets of palaeoclimate modeling, the Milankovitch theory, is called into doubt by isotope analysis of a calcite vein, just reported in Science by Winograd and colleagues. The theory, which is backed up by a compelling bank of evidence, suggests that the ice ages determined, with unprecedented accuracy, in the new record cannot be reconciled with the planetary cyclicity. . .

Winograd and colleagues’ evidence also turns on oxygen isotope data, this time from vein calcite coating the hanging wall of an extensional fault at Devils Hole, an aquifer in southern Nevada. In 1988, the authors published a date, 145,000 years, based on 234U-230Th dating for the end of the penultimate ice age (Termination II), marked by an increase in the 18O to 16O ratio, a change taken to mirror an increase in local precipitation. Although the date was only 17,000 year earlier than the previously accepted date of 128,000 years, if correct, this change is enough to bring Milankovitch mechanism into serious doubt. . .

I remain confused. The geochemist in me says that Devils Hole chronology is the best we have. And the palaeoclimatologist in me says that correlation between accepted marine chronology and Milankovitch cycles is just too convincing to be put aside. . .

One side will have to give, and maybe – just to be safe – climate modellers should start preparing themselves for a world without Milankovitch.” 1


This paper was followed by a rebuttal from Cesare Emiliani, a well known and outspoken supporter of MT entitled, “Milankovitch theory verified”.  Emiliani wrote:


“Broecher compares terminations (the transitions between glacial and interglacial conditions) in deep-sea cores with the recently published delta18O curve from Devils Hole, Nevada to question the validity of the Milankovitch theory.  Terminations, transitional episodes that extend through time, are poor time markers for the correlation of Pleistocene sections. The maxima (hypsithermals) and especially the minima (bathythermals) in the isotope curves are much sharper and thus afford a more precise correlation. . .

Astronomical parameters recalculated by Berger show that at that time both obliquity [41ka period] and eccentricity [100ka period] were low. If these conditions were responsible for the last ice age, one would expect that similar conditions could be responsible for the preceding ice ages. The table compares the times when these conditions recurred during the past half million years with the ages from Devils Hole. Because the two time scales are independent of each other, their close similarity suggests a common cause which, one would suspect, is the Milankovitch mechanism. Thus, far from invalidating Milankovitch, as maintained by Winograd et al. and by Broecker, the Devils Hole chronology appears to provide support. Support is also suggested by an analysis of the Devils Hole spectrum.” 2


However, Emiliani’s paper was soon rebutted by Landwehr et al. in a 1994 paper entitled, “No verification for Milankovitch” where the authors accuse Emiliani of actually biasing the data by leaving out and even adding key data points in his analysis of the data:


We were puzzled by the table in the Scientific Correspondence by Emiliani. He rejects the conventionally used terminations (glacial-interglacial transitions) as time markers and focuses on bathythermals (the coldest portions of glacial cycles), which he deems to be sharper and therefore more precise time markers. He claims that bathythermals in the Devils Hole delta 18O chronology occur at times when the orbital parameters of obliquity and eccentricity are both “low”, as determined from Berger’s figures, thereby supporting Milankovitch mechanism.

Unfortunately, Emiliani does not specifically define what he means by the critical terms “low” or “when they approach coincidence”, but we assume he takes “low” to mean the times when both obliquity [41ka period] and eccentricity [100ka period] were at a minimum, or obliquity was at a minimum and eccentricity was less than at least the long-term (0-600,000-year) average value. We show in the figure the seven astronomical “low” events that Emiliani gives in the third column in his table, as well as the seven (but not identical) events that satisfy the specific definition of astronomical low conditions using data in reference 4. We were puzzled as to why Emiliani omitted the two-well defined “low” events at 395,000 and 517,000 years and note that they do not correspond to bathythermals in either the Devils Hole or the marine delta 18O chronologies.  Indeed, the 395,000-year “low” event occurs during a peak interglacial time. We also note that Emiliani’s designation of a “low” event at 555,000 and 150,000 years does not fit the earlier stated definitions.

Also shown in the figure are the eight major delta 18O minima, denoting times of full glacial climate, found in the Devils Hole chronology, and the subset of six events that Emiliani gives in the second column in his table. He did not mention the two Devils Hole isotope minima at 223,000 and 173,000 years, which do not correspond to any astronomical “low” event.

In comparing the astronomical “low” events predicted by the specific definition with the minimal isotope events found in the Devils Hole chronology, one sees that although there are four ‘matches’, there are six ‘non-matches’, twice when a bathythermal would be predicted but did not happen, and four times when one did occur but not during an astronomical “low” event. Thus the astronomical conditions that Emiliani specifies is neither sufficient nor necessary for the occurrence of bathythermals.” 3


So, it seems that Emiliani manipulated the data quite extensively in order to make it fit in with MT.  Though this was most certainly done subconsciously, it highlights the pitfalls of bias – of having a strong belief that a particular view or theory is almost certainly “true”.  This does not mean that such a belief isn’t good to have in many cases. It is just that one should be aware of one’s own inescapable biases when approaching and interpreting new or even old data.

But, there are those, such as Imbrie, who argue that the Devil’s Hole data is inappropriately compared to deep-sea core data – that the two data sets should be read independent of each other since they are most likely effected by the weather in very different ways. However, Karner and Muller have responded to this notion with the following comments:

“As long as Devils Hole was unique, it could have been a fluke. It is a land-based site, and perhaps it was only recording a local climate change (although that was unlikely, based on its strong correlation with the Vostok paleotemperature time series). It would have been incautious to abandon the otherwise successful Milankovitch theory based on a single data record. But now there are other records – sea level records from opposite sides of the globe, that show the causality problem is real – and these data are not easily dismissed.”11

A Few More Problems


Despite the problems with contradictions from the Devils Hole data, MT managed to hang on as a generally accepted theory in the mainstream science community. But, not all were so convinced. Consider also the conclusions of Vyacheslav A. Bolshakov from Moscow State University (August, 2005):


“Empirical data reveal considerable inconsistencies of the Milankovitch theory. The main of them are the following:


      1. The climatic cyclicity for the Brunhes chronology is primarily governed by a 100 ka periodicity, attributed to eccentricity variations, whose immediate impact is disregarded in the Milankovitch theory.
      2. According to empirical data, glacial events fall on eccentricity minima, whereas under the Milankovitch theory these are mainly coupled to eccentricity maxima. [known as the causation problem]
      3. About one million years ago, the dominant climatic periodicity switched from 41 ka to 100 ka, which is at odds with the Milankovitch theory, because the variation periods of orbital elements suffered no significant changes at that time.


It seems logical, that a theory, which contradicts to empirical data, is wrong. Consequently, the Milankovitch theory should be rejected, as was done with regard to it 50 years ago, as well as with regard to Croll’s theory about 100 years ago.”4


This feeling is also shared by several other prominent scientists, such as Richard Muller from Berkeley.  Muller writes:


“We have been studying the cycles of the ice ages using data collected from sea-floor cores, Greenland ice, and other terrestrial sources. We have published a careful spectral analysis that shows that the “standard” Milankovitch theory for the glacial cycles is wrong, and we have proposed an alternative explanation: that the cycles are driven by extraterrestrial accretion.”5


In short, because of the many problems with MT, especially the causality problem where the ice and deep see core data say the Earth should be warm when MT says it should be cold, and visa versa, Muller believes that MT is simply wrong and should be replaced by another theory.  He has even come up with a ready theory to explain away at least one of the major problems with MT – extraterrestrial accretion.  Muller basically believes that as the Earth travels around the Sun it does not always stay in the same plane. Like a slightly wobbly record on a record player, the Earth will sometimes be above the plane and sometimes below the plane.  This happens to occur in a cyclic pattern of about 100,000 years.  Muller believes that as the Earth moves out of plane, it picks up more cosmic dust than usual, which affects the weather of the Earth in 100,000 year cycles.


“So far, Muller and MacDonald have been unable to get their full paper, detailing their work, published, despite their considerable credentials. It’s been rejected by Science. It’s been rejected by Nature three times – the third time as recently as June – though the editors did request, and published, a shorter version summarizing their findings last November. Why? Muller pulls open a long file drawer, crammed with papers. ‘Here it is. Essentially everything that’s been published for the last twenty years assumes the Milankovitch model. I think it’s very hard for people in this field, and all the referees to whom our paper has been submitted are working in this field, to accept our paper. They’d have to say that most of their own work for the past twenty years is fundamentally flawed.'” 6


Isn’t that interesting?  Milankovitch theory is so engrained in the scientific community that even an otherwise well-respected mainstream scientists seems to be having trouble getting anything significant published that fundamentally challenges Milankovitch.

However, some have managed to publish certain problems with MT. Consider the following discussion by Raymo et. al., concerning the “41kyr Problem”:


“While many investigators have attempted to model the 100 kyr world, few have focused their attention on the 41 kyr world. A notable exception is Andre Berger and colleagues who used a two-dimensional ice sheet-climate model to try to simulate the growth and decay of ice sheets over the last 3 million years [e.g., Berger et al., 1999]. While the obliquity period [41ka] is present in the model output, precessional variance [21ka] in ice sheet mass is also strongly present. In other words, although they successfully model the lack of the 100 kyr eccentricity cycle, they were not able to model an ice sheet that varies only at the obliquity frequency. This appears to be because the model is ultimately very sensitive to high-latitude summer insolation.

Secondly, a discussion of the 41 kyr problem can be found in Richard Muller and Gordon MacDonald’s book “Ice Ages and Astronomical Causes” [Muller and MacDonald, 2000]. Following Kukla [1968], they propose that northern latitude winter insolation (e.g., January 65_N) may drive late Pliocene/early Pleistocene climate cycles, even though the total insolation received in January is a factor of 20 less than summer insolation at the same latitude. However, they go on to say this proposition is speculative and that the geologic record is posing a problem that needs to be solved.” 7


So, the 41ka cycle isn’t without its own significant problems – and this is well recognized.  This only adds to the problems with the 100ka cycle and the 400ka cycle – none of which really fit MT. Perhaps Vyacheslav Bolshakov isn’t off his rocker after all? – when he notes, along with Muller and others, that causality problems are a significant issue? How can MT be valid when empirical data is interpreted to indicate glacial events to fall on eccentricity minima whereas MT claims they should fall on eccentricity maxima?  Isn’t that a rather fundamental problem all by itself?  How are such problems, combined with dramatically discrepancies between predictions based on U/Pb dating vs. MT, as demonstrated by Hinnov et al. overcome without a multitude of ad hoc Band Aids? Hinnov writes:


“Two principal techniques for high resolution dating of the stratigraphic record, namely, U-Pb dating of single zircons in volcaniclastic interbeds and statistical analysis of orbitally forced sediments, were recently applied to the Anisian-Ladinian Latemar Limestone of northern Italy, a succession of more than 500 meter scale platform cycles, each of which records a low amplitude sealevel oscillation. Unfortunately, the results of the two techniques are in serious conflict. Evidence for strong Milankovitch forcing of the cyclic succession indicates a depositional duration for the Latemar Limestone of 10-12 million years, whereas U/Pb-dated zircons from volcaniclastics in coeval basinal Buchenstein beds indicate only 2-4 million years. This conflict has led to a scientific impasse: either the approach used to determine a Milankovitch origin for the cycles is wrong, or the interpretation of the results from the zircon dating is wrong, or both are wrong.”8



Playing The “Right” Tune

There is yet another potential problem known as “tuning”.  Since deposition of sediments in the bottoms of the ocean and snow in the polar regions is not uniform, the patterns produced by the oxygen isotope ratios has to be “tuned” in order to match up the isotope pattern with the pattern of the Milankovitch cycles.  “The time scale of the data [for 21ka and 41ka cycles] was tuned by adjusting the sedimentation rate to match the expected orbital cycles”.1 In other words, the 41ka obliquity cycle is supposed to be linked to the same cycle in ocean cores, which is “tuned” to match the expected orbital cycle. Doesn’t that sound just a bit like circular reasoning?  If a pattern is tuned to match another pattern, of what independent value is the data behind the tuned pattern?

In a very interesting paper, Peter Huybers, of Harvard University, reports on some experiments he has carried out with the practice of tuning as it relates to Milankovitch theory.


“White noise can be tuned to a ratio of 41:23 in such a way that multiple spectral peaks can be generated with high coherencies as well as precession-like amplitude modulated bands were none previously existed. . .

To demonstrate that tuning probably over-estimates the variance attributable to Milankovitch forcing, I tune white-noise to the orbital parameters. I show that tuning can routinely generate multiple spectral peaks, high coherencies, and precession-like amplitude modulated bands where none previously existed. This indicates that tuning assumes an unverifiable relationship between astronomical forcing and the delta-18O climate proxy and calls into question the accuracy of tuned chronologies. Finally, by tuning the Devils Hole record, I attempt to highlight how tuning can bias our interpretation of climate change toward the Milankovitch hypothesis. . .

It is possible that climate does linearly respond to Milankovitch forcing, but this should be recognized as a largely unsupported assumption. As it stands, tuning adopts a complicated relationship between time and depth to account for its assumption of a phase locked relationship between the orbital parameters and the respective delta 18O frequency bands. This time depth relationship is seen to significantly change the spectral and temporal nature of a record. In the aggregate case of DSPD607, ODP677, ODP849, and ODP980 it doubles the variance concentrated at Milankovitch frequencies. In light of Devils Hole, tuning enforces a potentially false causal relationship between deglaciations and insolation at 65o North. Because the large assumptions inherent in tuning are often not accounted for, the interpretation of tuned deep sea sediment records is generally biased toward the Milankovitch hypothesis.”8


In other words, The original signals found in the deep-sea cores are not used to make these graphs that one sees in published papers.  These graphs are polished quite a bit beforehand.  And, they are polished or “tuned” in a biased way so that they will match the timing of the Earth’s precession (23ka) and obliquity (41ka) phases. For instance, if the known ratio of these phases had been different, like 50:25, this ratio would be the one used to “tune” the signals and this ratio would therefore be the one seen in these papers. Sounds preposterous! – doesn’t it?  But, this is exactly what is going on in many “scientific” papers.

Consider the following report of a conference in which the whole concept of tuning was attacked by Richard Muller.


“Muller scored the most points at the meeting when he attacked a standard technique, called tuning, that oceanographers use for dating layers in sediment cores. The task of dating these strata is difficult because sediments may accumulate more quickly during some eras and more slowly in others. To tell the age of layers between known benchmarks, researchers often use the Milankovitch orbital cycles to tune the sediment record: They assume that ice volume should vary with the orbital cycles, then line up the wiggles in the sediment record with ups and downs in the astronomical record.

“This whole tuning procedure, which is used extensively, has elements of circular reasoning in it,” says Muller. He argues that tuning can artificially make the sediment record support the Milankovitch theory.

Muller’s criticisms hit home with many researchers. “He scared the hell out of them, and they deserved it,” says Broecker.”10



Correlation with Dendrochronology?

The fact of the matter is that matching patterns like this is quite subjective – even given various statistical techniques that are supposed to help. The problem of auto-correlation, as noted by D.K. Yamaguchi, seems to be a problem here just as it is with tree ring dating (i.e., dendrochronology).

In this paper Yamaguchi recognized that tree rings tend to “auto correlate” or actually cross-match with each other in several places within a “master” tree-ring sequence.  What he did to prove this was quite interesting.  He took a 290-ring Douglas-fir log known, by historical methods, to date between AD 1482 and 1668 and demonstrated that it could cross-match in multiple places with the Pacific Northwest Douglas Fir Master Growth-ring Sequence to give very good t-values. A t-value is given to a wiggle-match on the basis of a statistical analysis of the correspondence between two wood samples. This statistical assessment is done by computer which assigns high t-values (3 and above) to good wiggle-matches and low t-values (below 3) to those with poor correspondence between the ring patterns.  Amazingly, using such t-value analysis, Yamaguchi found 113 different matches having a confidence level of greater than 99.9%. For example, Yamaguchi demonstrated that his log could cross-match with different master tree-ring sequences to give t-values of around 5 at AD 1504 (for the low end of the ring age), 7 at AD 1647 and 4.5 at AD 1763. Six of these matches were non-overlapping.11 That means that this particular piece of wood could be dated to be any one of those six vastly different ages to within a 99.9% degree of confidence. This finding calls into serious question the accuracy of building master tree-ring sequences that go back over many thousands of years.12

In this light, it is interesting to note that a number of the crucial dendrochronology sequences, such as the Garry Bog 2 (GB2) and Southwark sequences, which connect the Belfast absolute chronology (i.e. the AD sequence) to the ‘floating’ Belfast long chronology (i.e. the BC sequence), and ultimately used to re-date the South German chronology, have t-values of around 4. These t-values are considerably lower than those obtained for some of the historically incorrect dates produced by Yamaguchi’s experiment. Thus, one would be justified in asking if the crucial cross-links which connect up the floating sequences of the Belfast and German chronologies are based on incorrect wiggle-matches – resulting from the phenomenon of auto-correlation. As noted by several, such as Lasken, this problem prompts a second very basic question.  That is, should one expect tree-ring-growth patterns to produce genuine correspondences at the same historical dates when the climates (and in particular the micro-climates) of Ireland, England and Germany are so different?  Clearly, dendrochronology, although possibly helpful for the dating of certain relative events, is not anywhere near an exact science.13

The work of Douglas Keenan is also quite interesting in this regard. “There is currently only one (substantial) master dendrochronology from anywhere in the Ancient Near East. Hence this master dendrochronology has great importance. This master is from Anatolia. “Anatolia” is a geographical term, roughly designating modern Turkey. A master dendrochronology for Gordion (39.7 °N, 32.0 °E), in central Anatolia, was first developed in the 1970s. This master dendrochronology, however, does not extend continuously from the present to the past. The master has been anchored in time-i.e. dated-largely via radiocarbon (originally, the master was dated via archaeo-history). In what follows, much of the work that has been done in Anatolian tree-ring matching is reviewed. The conclusions are disturbing, and have implications for tree ring studies generally.” 14,15

Turkish dendrochronology is quite interesting and relevant to this discussion and the understanding of the science of dendrochronology as a whole. One especially interesting example has to do with the work of Kuniholm and his dating of wood from the city gateway at Tille Hoyuk – an ancient city. What happened was that the t-value wiggle-match produced by computer analysis came up with not one, but three matches of 1258, 1140, and 981 B.C. – each with a greater than 99.9% certainty. 14,15,16

Consider that perhaps desire plays more of a part than actual detached science when it comes to dendrochronology. In this light, note that dendrochronologists working in Hohenheim, Germany, were proven wrong three times in the mid 1990s, each time after very strong assertions of reliability. 14

As with dendrochronology master sequences going back beyond a few thousand years, all kinds of ad hoc hypotheses are required to support MT so that it doesn’t collapse completely.  The use of many such ad hoc fixes is usually a good sign that the theory really isn’t that solid after all.




  1. Wallace S. Broecker, Upset for Milankovitch theory, Nature 359, October 29, 1992
  2. Cesare Emiliani, Milankovitch theory verified, Nature 364, August 1993
  3. J.M. Landwehr, Isaac J. Winograd and T.B. Coplen, “No verification of Milankovitch”, Nature 368, April 14, 1994: 594
  4. Vyacheslav A. Bolshakov, The problems of the orbital theory of paleoclimate: new way for their solution.119899, Moscow State University, Geographical faculty, Moscow, Russia The poster presentation at the Second Open Science Meeting: Paleoclimate, Environmental Sustainability and our Future. August 2005. (’shakov.pdf)
  5. Richard A. Muller and Gordon J. MacDonald, Specturm of 100-kyr glacial cycle: Orbital inclination, not eccentricity, Proc. Natl. Acad. Sci., Vol. 94, pp. 8329-8334, August 1997
  6. Richard A. Muller, An Astrophysics Experiment (
  7. Maureen Raymo, The 41kyr World: Milankovitch’s other unsolved mystery, Paleoceanography 18, No. 1, 2003
  8. Linda A. Hinnov, Nereo Preto, Lawrence A. Hardie, Johns Hopkins University, Baltimore, MD, University of Padua, Padua, Italy, “Triassic Geochronology Controversy: Milankovitch Versus Zircon Radioisotope Time Calibration of the Latemar Platform Cycles”, AAPG Annual Convention, May 11-14, 2003, Salt Lake City , Utah (…)
  9. Peter Huybers, Milankovitch and Tuning, Harvard University, 2001 (
  10. Richard Monastersky, “The Big Chill – Does dust drive Earth’s ice ages?”, Science News, vol 152, October 4, 1997, pages 220-221. (
  11. Daniel B. Karner and Richard A. Muller, “A Causality Problem for Milankovitch” (
  12. Yamaguchi DK.1986. Interpretation of cross correlation between tree-ring series. Tree-Ring Bulletin 46:47-54.
  13. Allen Roy, C14-Dendrochronology ( ) Sun, 2 May 1999 21:09:50 -0700 (
  14. Douglas J. Keenan, Anatolian tree-ring studies are untrustworthy, The Limehouse Cut, London E14 6N, United Kingdom, 16 March 2004  (
  15. Douglas J. Keenan, Why Radiocarbon Dates Downwind from the Mediterranean are too Early, Radiocarbon, Vol 44, Nr 1, 2002, p 225-237 (
  16. Kuniholm, P. — 1993: Appendix in G. Summers: Tille Huyuk 4, pp. 179-90






Leave a Reply