Don Anderson, 22nd November, 2010
Let me clarify further. We have been arguing about
data, assumptions and interpretations for 40 years.
Little progress has been made. Plume "theory" has
no theoretical underpinnings and violates well known
scaling relations of fluid dynamics and thermodynamics.
But this has not ended or slowed the debate. I understand
that data mainly controls your thinking, but nevertheless
assumptions and deductions are involved in all of the
arguments for plumes (and different assumptions can
reverse the conclusions). So, let's look at the assumptions.
I suggest that we switch to logic and the identification
of fallacies. All sides can agree that the principles
of deductive logic and mathematical proof are neutral.
So let's analyze the rationale behind plumes as one
would analyze premises and conclusions. Forget the
data, for the time being, and just list why you, or
others, believe in plumes or otherwise.
Very few geochemists challenge the
views that high
3He/4He and rapid eruption
rates signify lower mantle components and plume heads,
respectively. Why do they believe that?
reasoning. There is no theoretical basis for presuming
that an accreting planet will not melt or that
it will retain cold 3He-rich domains.
The idea that that high 3He/4He
was a primordial mantle or plume diagnostic can
be traced back to the observation that Hawaii and
Iceland had higher than MORB ratios and that these
are "known plumes". Likewise, the idea that plume
heads drain their magmas in 1 Ma or less is not
predicted by plume head theory. It can be traced
back to the observation that CFB eruptions are
brief and these are "known to be due to plume heads".
The idea that high 3He/4He means high 3He follows
from the "knowledge" that
the lower mantle is undegassed and therefore a
high 3He content will keep the ratio high.
The plume hypothesis has become a
paradigm in the sense that the same underlying assumptions
are adopted by workers in various disciplines. The
main ones are that mantle plumes exist, depleted MORBs
represent ambient mantle and that they originate in
the upper mantle, magma volumes and seismic velocities
are proxies for mantle temperatures, and high 3He/4He ratios
(higher than MORB) are evidence for a deep undegassed
plume reservoir. It is these assumptions, rather than
data, that are responsible for the longevity of the
hypothesis. If logical underpinnings of the hypothesis
are essentially non-existent, this will become evident
if we look at them in this light. We also need to discuss
proxies (magma volumes, seismic velocities, 3He/4He,
eruption rates, etc.). Are these really reliable proxies
for temperature and plumes?
One can debate the geological, geochemical, seismological
and petrological evidence but thermodynamic and logical
inconsistencies and errors are more fundamental. Among
these are substituting relative plate velocities, seismic
velocities, travel times and temperatures for absolute
values, confusing high ratios with high numerators,
ruling out upper mantle sources because MORBs are from
the upper mantle, and assuming that thermodynamic variables
can vary independently with T, P and depth.
The upper mantle is the reservoir
Hawaiian magmas are hotter than MORB
Large igneous provinces are of short duration
Fluids heated from below develop plumes
High temperatures cause low seismic velocities
Low velocity features exist in the lower mantle
Some OIB differ chemically from MORB
Island chains and intraplate volcanoes exist
Half the seismic arrivals to Hawaii are slower than
the other half
…but to follow up these true statements
with the assertion that therefore mantle plumes exist
is to commit a logical fallacy.
Common Assertions and hidden assumptions:
…the lower mantle is the reservoir
for intraplate magmatism (e.g., O’Nions & Tolstikhin,
1996; Hofmann 1997; Kellogg
et al., 1999; Albarede & van
der Hilst, 1999);
…the potential temperature of the upper part
of the deep boundary layer exceeds any temperatures
in the upper mantle by at least 200°C;
…temperatures in the upper mantle do not exceed
temperatures recorded at midocean ridges;
…if A/B >C/D, then A>C, where A,B etc.
are chemical or isotopic concentrations such as 3He
…if X implies Y, then not-X implies not-Y, where
X are mantle components such as MORB and OIB and Y
are hypothetical reservoirs such as upper mantle;
…homogenous products imply homogeneous sources;
…geochemical ratios that exceed the MORB mean
are from a lower mantle reservoir;
magmatism is of short duration because large igneous
provinces are of short duration (the Texas Sharpshooter
…one should look for anomalous mantle only where
it is available, e.g. oceanic island basalts
and OIB xenoliths; unsampled regions are MORB-like
(the lamppost fallacy);
...let us look at only those
volcanic features where a plume mechanism actually
makes geological sense;
…Evidence from small
seamounts seems completely irrelevant to this debate.
[begging the question].
It is actually fun to decompose an argument into
a logical statement; premisses and conclusions, such
MORB is from upper mantle
Therefore, OIB is from lower mantle
MORB is colder than Hawaii
Therefore Hawaii is from a thermal plume
Hawaiian basalts are not like MORB
Therefore Hawaii is a plume
Heating from below makes plumes
Therefore mantle plumes cause hotspots
Lower than average seismic velocities occur in the
Therefore plumes exist
You can play this game yourself with
your favorite arguments. The above all appear in published,
peer-reviewed papers by eminent scientists. If you
can demonstrate a logical fallacy or inconsistency,
this should end the debate on that isssue. If your
theory violates the second law of thermodynamics, you
should hang your head in shame and admit defeat.
Dean Presnall, 22nd November, 2010
My J. Petrol.
paper (still in the mill) shows that Hawaii samples
normal temperature conditions and that MORBs are the
result of a perturbed geotherm with (cool) melt extraction
within the thermal boundary layer. I think i have a
very solid argument, but it will be interesting to
see what others think.
Yaoling Niu, 22nd November, 2010
Great discussion! May I add some "confusion"?
My assertion and reasoning:
Hotspots must be hot indeed as they are manifested
by volcanoes. Volcanoes must be hot as you can feel
their hotness. I remember that I was wearing "fully
insulated" clothes, but it still took me more
than 10 attempts to get some hot lavas with my geohammer
near Pohoiki (?) on Big Island of Hawaii in 1990.
So, my understanding is:
- Hotspots are hot; they do exist as we can see and
- However, are they surface expressions of deep-rooted
mantle plumes? I don't know because I neither see
nor feel mantle plumes.
- But we can ASSUME these hotspots
are surface expressions of deep-rooted mantle plumes.
Then we can use the geochemistry (make some more
assumptions) and geophysical means (with still more
assumptions) to test if mantle plumes exist.
The result is clear
to some, but not to others, so one has to choose:
plumes DO exist: from some thermal boundary layers
in the Earth, most likely at the CMB or perhaps from
the transition zone, or perhaps some compositional
anomalies (chemical plumes) or in combination.
plumes DO NOT exist because we don't see and seismology
cannot yet reliably resolve them even if they do
are statements in the literature, something like:
evidence for the presence of mantle plumes (large
scale upwelling) - the evidence becomes less evident
if we remove some of the assumptions
- Seismic evidence for
the presence of mantle plumes (large scale mantle
upwelling) - likewise, the evidence becomes less
evident if we are more careful about assumptions
- Nevertheless, "The lack of evidence is not
the evidence of absence"
So, where are we now? I look forward to reading Plates
vs Plumes: A Geological Controversy by Gillian
Gillian R. Foulger, 22nd November, 2010
Don: Thank you for your discussion here below on whether
or not "hot spots" are hot. I hope you will
come along to the session and contribute more of your
thoughts in person there. I look forward to attending
your own invited talk.
The question of whether/which/if any "hot spots" arise
from hot sources is a subject of considerable debate.
Some people assume they are hot without question and
don't bother to test this assumption. Others assume
they are and process data in whatever way is necessary
to get that answer. Other people process data in an
open-minded way, and get a frosty reception from their
colleagues when they fail to find unequivocal evidence
for high temperatures at most places. Still others
follow the scientific method and try to falsify the
conventional hypothesis that "hot spots" are
Several of us have been discussing privately the issue
that the source of "hot spot" lavas may come
from different depths in the surface conduction layer,
and I have illustrated this in Figure 6.3 of my recent
book "Plates vs. Plumes". This hypothesis
predicts variable source temperatures, hotter if the
lavas originate from deeper (and we are talking about
the upper couple of hundred kilometers here, not the
What the temperatures of their sources are is a question
relevant to their origin (e.g., depth, in the above
model) and I think it is a valuable line of enquiry.
It may end up telling us more about the (variable shallow)
origin of the lavas than about whether plumes exist
or not. To me, that might be more important than chasing
the debate about whether plumes exist or not. The plume
hypothesis is a belief system that can't be disproved
and is thus deserving of only limited attention.
I thus agree with you that determining source temperatures
is not likely to solve the problem of whether plumes
exist or not, if people assume that finding that one "hot
spot" come from a hotter source than another (or
MORs) proves that they do. However, I disagree that
the question is not relevant or important.
I hope this will also be a useful forum for discussion
the validity of some widely used but dubious methods
for deducing temperature, e.g. applying olivine-control
theory to cumulate rocks, or quoting single, isolated,
unrepeated ocean-bottom heat-flow measurements without
taking heed of local context.
Jim Natland, 21st November, 2010
As I have done my best to explain, the olivine-liquid
FeO-MgO backtrack procedure is fraught with difficulties
and almost certainly leads to higher estimated potential
temperatures and parental melt MgO than are either
necessary or demonstrated to occur in nature. If
you wish, you can ignore this and say that potential
temperature doesn't matter. Then I could pack up
my bag and go home. However, I would think that appropriate
estimates of temperature would be useful, even to
your model. Putirka said that differences in temperature
between hot spots and ridges must be on the order
of 200°C if plumes are to exist. Using backtracking
but not taking mixing into account, he concluded
that plumes do indeed exist. By including mixing,
I conclude that the 200°C difference is neither
demonstrated nor likely, even at Hawaii, and that
we consequently have a reason to doubt that plumes
exist. Is this not relevant?
Don Anderson, 21st November, 2010
I notice that there isan AGU Session entitled "Are
hotspots hot?". I maintain that this question
is not relevant to the plume debate and has diverted
attention away from important issues. Of course hotspots
are hot and Hawaii magmas may be hotter than MORB.
The real questions are "Do midplate volcanoes
sample ambient subplate mantle?" and "If
Hawaii is hotter than MORB does this prove the existence
of mantle plumes?".
I realize that, to many, and in particular, the editors
of Science and Nature, if recent papers and commentaries
are examples, that low relative seismic velocities
and mantle hotter than MORB mantle are considered to
be sufficient conditions to declare a plume sighting.
But this goes back to the old cooling plate paper by
McKenzie & Bickle (1988) who simply asserted
than MORB is ambient subplate mantle. This created
more paradoxes; anomalous subsidence and seafloor flattening,
and absence of the predicted low velocity halo (e.g.,
Priestley & Tilmann, 1999). The cooling
plate model itself, as others admit, has no physical
underpinning and is inconsistent with observed lid
thickening. As usual, paradoxes are due to bad assumptions.
Reverse the assumptions and the paradoxes disappear!
The absence of high heatflow around Hawaii and the
absence of low absolute seismic velocities around Hawaii
and absence of slow absolute travel times to Hawaii,
plus the anomalous bathymetry and subsidence throughout
the Pacific suggests that midplate volcanoes sample
ambient mantle in and below the boundary layer. This
turns the debate on its head. Why are midocean ridge
magmas so cold? This is the subject of my invited talk
in V25 at the forthcoming AGU meeting, on the Monday.
Hawaii can be "hot" and this is not inconsistent
with the lack of any thermal or absolute velocity anomaly
around Hawaii. It is the assumption that is wrong or
at least debatable. Recent tomographic studies published
in Science and Nature have just established that half
of the seismic anomalies to Hawaii are later than the
other half. The arrivals are not late in any absolute
The observed anomalies around Hawaii can
be due to heterogeneity and anisotropy above 220 km
depth, which is ignored in the Princeton and Carnegie
studies. There are better ways to study Hawaii than
with the relative travel times of near-vertical teleseismic
- McKenzie, D., and J. Bickle (1988), The volume and
composition of melt generated by extension of the lithosphere,
J. Pet., 29, 625-679.
- Priestley, K., and F. Tilmann (1999), Shear-wave
structure of the lithosphere above the Hawaiian hot
spot from two-station Rayleigh wave phase velocity
measurements, Geophys. Res. Lett., 26,
Dean Presnall, 18th November, 2010
Based on the model system phase
relations (which give a good understanding of P (but
T is too high by a large amount), plus the determination
by Lesher and his colleagues of the natural lherzolite
solidus at 5 GPa, the Puna Ridge olivine-controlled
trend of Clague gives melt-extraction conditions of
~ 4-5 GPa, 1500°C. This is also the
depth of maximum reduction in Vsv (maximum
melt fraction) and is very close to the depth of maximum
shear wave anisotropy throughout the mature Pacific.
So seismology and experimental petrology are saying
the same thing. They also say the same thing (but a
much lower lower P-T for maximum melting) for ridges.
This resolves –
I hope (and I was already convinced years ago – a
petrological controversy that has lasted almost half
a century. It also resolves (more hope) the plume controversy.
Jim Natland, 17th November, 2010
I don't think that any picrite or meimechite
is a liquid composition, but that they have a lot of
accumulated olivine. Nothing I've seen so far goes
beyond Kilauea mineralogy, and the highest MgO in glass
there has 15% MgO and an eruptive T of about 1350°C (Beattie
geothermometer). The glass accounts almost perfectly
for the spinel with highest Cr# and highest Mg# in
the suite. So who needs more, or hotter?
That's still plenty warm. Spinel from Karoo looks like
Kilauea spinel, and doesn't need more than 1350°C.
Add what you want and however you want to do it to get
it up to a potential T from eruptive T. Samoa is a lot
cooler. MORB tops out at about 1225°C.
Who can tell
with komatiites? Gorgona spinel has slightly higher
Cr# and Mg# than Kilauea spinel, but is more oxidized.
So there's a temperature trade off.
Ever since Langmuir came up with temperature
as the explanation for different parental MORB, there
has been a widespread assumption that everything can
be explained by T variations. But it's heterogeneity,
not difference in T, that is the key. The tendency
is to explain as much as possible with T,
especially invoking plumes. But I'd say there are
two numbers to think about, no more. One is 1225°C for
MORB; the other is 1350°C for Kilauea picrite glass.
Those are the extremal bounds. Plenty of places are
cooler than Kilauea and nothing so far is hotter,
although Baffin-West Greenland and Karoo approach Hawaii.
Dean Presnall, 17th November, 2010
skeptical about assigning temperature to anything
that is not a glass and cannot be compared to experimental
data for similar compositions. If picrites and komatiites
can be shown to have been melts, then they would indicate
very high temperatures. Many (all?) picrites are mixtures
of melt and crystals. Komatiies are more difficult
because they are old, may have once been melts, and
may represent temperatures that existed only in the
past. I think
komatiites would carry more weight as high temperature
magmas than picrites.
Don Anderson, 10th November, 2010
Plume theory predicts that plume basalts
have "high" 3He
contents, "high" temperatures and differ
geochemically from MORB. Most hotspot magmas have extraordinarily
low 3He contents, much lower than MORB.
The depleted character and isotopic similarity between
the highest temperature terrestrial magmas (komatiites,
ferropicrites, meimechites) and MORB are not predicted
by plume theory. The highest 3He magmas
(popping rock) and the highest
3He/4He magmas (Greenland and
Baffin picrites) are indistinguishable from MORB for
Pb, Os, Sr and Nd isotopes. The most common isotopic
component of basalt is C- or FOZO, the component bearing
high 3He/4He ratios. These observations
should give pause to anyone attributing these to mantle
plumes or the lower mantle. All the heavy isotopes
imply ambient upper mantle compositions. These components
all appear to come from the shallow mantle, and often
appear at the onset of rifting or at the early stages
of magmatism. A newly fractured continent will surely
expose mantle that is hotter than under mature spreading
Potential temperatures as high as
~1600°C are considered
to be "excess" because they are higher than
MORB temperatures, at least MORB from mature ridges.
There is no reason other than convention why hotspot
magmas cannot sample ambient shallow mantle and why
they cannot be hotter than MORB. The actual temperatures
of OIB, CFB, komatiites, picrites etc. may be higher
than MORB but this does not imply that they come from
localized hotspots, or plumes, in the mantle. One has
to defend the proposition that MORB temperatures are
the maximum allowable temperatures in the upper mantle,
and that temperatures of ~1600°C cannot exist in
the surface boundary layer.
If the thermal boundary
layer is ~220 km thick, as implied by seismology, and
conduction gradients are 6-10 °C/km, then the base
of the boundary layer can be 1320-2200°C, and much
hotter when the T dependence of conductivity is taken
into account. If one accepts the 1300°C at 100 km of
the Cambridge cooling plate model, then the potential
temperature at 220 km depth can be >2000°C!
A more rigorous derivation gives Tp of 1600±200 C as
a plausible ambient temperature range at the base of
the upper mantle boundary layer. MORB come from shallower
depths or colder mantle or both. If there is radioactive
heating in the mantle, the Tp at the top of the lower
mantle thermal boundary layer will be less than at
the base of the upper mantle boundary layer.
Seismic waves that bounce off the
surface (SS) in the Pacific show that Hawaii is not
anomalous; it is just like other places in the Pacific
and is faster than many.
10th November, 2010
It is great that our recent
webpage has fueled discussion on Karoo origins.
As Dr. Natland mentions in his comment below, we are
looking at mixed magmas. Importantly, however, all
the evidence points to these magmas being very primitive
(i.e. olivine-controlled). Thus we can deduce the parental
magma compositions by adding olivine until Fe-Mg equilibrium
is reached. There is no geochemical, mineralogical
or petrographical evidence for mixing with cpx- or
plag-saturated (and thus Fe-enriched) magmas. In any
case, the predominance of forsteritic olivine phenocrysts
(up to Fo92) in the meimechites indicates
that (a) mixing was minor and (b) that the olivines
had to have crystallized from highly magnesian parental
magmas. Importantly, the calculated parental melt compositions
are in equilibrium with depleted upper mantle peridotite – a
source also indicated by the isotope compositions.
Concerning the calculated temperatures, we took the
presence of water into account in the calculations.
The composition of amphibole and its abundance indicates
that the water contents in the melt were not likely
to be higher than 2 wt. %. Equivalent dry magmas would
have temperatures around 1700°C. One additional
factor that was mentioned by Dr. Ivanov and could lower
the calculated temperatures is the CO2 content
of the parental magma. The overall subalkaline character
of the meimechites (in fact, the major element compositions
are more reminiscent of komatiites than Siberian meimechites)
and the absence of associated carbonatites do not indicate
marked mantle CO2 influence so far in the
case of Vestfjella, however.
I would also like to add that although the ferropicritic
and meimechitic magmas were not parental to Karoo CFBs in
sensu stricto, their source is likely to represent
a significant sublithospheric end-member for Karoo
magmatism. Parental magmas of some Karoo CFBs formed
from this same source at lower pressures and at higher
degrees of melting and are thus not compositionally
equivalent to ferropicrites/meimechites.
All of these issues are more thoroughly discussed
& Luttinen (2010), Heinonen
et al. (2010), and in my nearly complete
PhD Thesis (Heinonen, in prep).
James Natland, 5th November, 2010
to the new webpage
by Heinonen & Luttinen, I have been pondering
these very same meimichites and ferropicrites in connection
with temperatures. Some of this will be in my forthcoming
AGU poster. First of all, they are manifestly mixed
rocks, based on both geochemical and mineralogical
grounds. So they are not all that hot. Second, they
are not parental to the bulk of Karoo tholeiites, therefore
they are irrelevant to consideration of temperatures
of those rocks. The rest of the web page is a measured
consideration of geochemistry. Some of it reminds me
of Iceland, but at the core of it, everything is depleted.
if it were not for the (erroneous, in my opinion) interpretation
of high T, there would be nothing at all for plumes.
Alexei Ivanov, 5th November, 2010
The finding of meimechites shows
that Karoo is very similar to the Siberian traps -
(1) located in a back-arc
subduction system, (2) the dominant type of magma
trace element signatures, which are usually interpreted
as evidence of lithospheric involvement, and (3) high-Mg
magma. Probably findings of carbonatites will follow.
What is high-Mg magma? It is usally interpreted as
evidence of high-T. However, both Siberian and Karoo
meimechites show mineralogic evidence of water in the
source (magmatic amphibole and mica). It was shown
by Elkins-Tanton et al. (2007) [Cont.
Min. Pet., 153, 191-209] that
both water and CO2 lower the melting point
for meimechite sources. So the temperature may not
be so high. Meimechites = depleted MORB-type source
in terms of radiogenic isotopes, depth and water.
last updated 10th November, 2010