All?gre Claude J. Isotope Geology
Подождите немного. Документ загружается.
An analogous procedure of weighting by u ncertainties is followed to calculate variance
and co variance:
V
x
¼
1
N1
P
i
1
2
i
ðx
i
xÞ
2
hi
1
N
P
i
1
2
i
and V
xy
¼
1
N1
P
i
1
2
i
ðx
i
xÞðy
i
yÞ
1
N
P
i
1
2
i
:
In each case,the meansare calculatedby theweighted method:
x ¼
P
i
1
2
i
x
i
P
i
1
2
i
:
As can be seen, all of this gives expressions which are a little heavy to manipulatebut which
arenotrouble forcomputer programs!
5.3.4 Calculating the equation of the straight line of correlation
We shall set out the principle of the least- square s method in a simple case. Let us look
for the equation of the straight line that statisticallydescribes the cloud of points
x
i
;
y
i
ðÞ.
We accept without proving it (it seems acceptable intuitively) that the straight line will
pass through the mean point of coordinates
x;
yðÞ.We are therefore looking for an equ a-
tion of the ty pe y ¼ax þb. The idea in the least-squares method is to look for the best
(least bad) straight line passing through a cloud of points by using a precise criterion.
The usual criterion is to mini mize the distance between the points observed and the
straight line we are looking for, which is done by minimizing the sum of the squares of
the distance between the data points and the ir respective vertical projections. This is
the simple case we shall work through mathematically to make the reasoning clear. If x
and y are connected by a straight lin e, for the var ious values (x
1
, y
1
), (x
2
, y
2
), ...,(x
N
, y
N
)
we have y
1
¼ ax
1
þb, y
2
¼ax
2
þb, ..., y
N
¼ax
N
þb. But this is not quite right because
the straight line does not pass exactly through each point. The square of the distances,
then, iswritten:
ðy
1
b ax
1
Þ
2
þðy
2
b ax
2
Þ
2
þþðy
N
b ax
N
Þ
2
¼ D
2
:
We try to minimize the distances from our two unknown parameters a and b.We there-
fore write: @ðD
2
Þ=@a ¼ 0 and @ðD
2
Þ=@b ¼ 0, which gives two equations with two
unknowns:
@ðD
2
Þ
@a
¼
X
i
2x
i
ðy
i
b ax
i
Þ¼0
@ðD
2
Þ
@b
¼
X
i
2ðy
i
b ax
i
Þ¼0:
177 Sources of uncertainty in radiometric dating
Itcanbeseenthenthattheproblemissoluble:wehavetosolveasystem ofequationswithtwo
unknownsa and b.The solutiontothisproblem canbewrittenverysimply:
a ¼
y r
y
x
x
b ¼ r
y
x
where r is th e correlation coe⁄cient.
The equationofthebeststraight-line ¢tp asses throug h
x and
y:
y ¼
y þ r
y
x
ðx
xÞ:
Theuncertaintyon theslope iswritten:
fpg¼
y
x
ffiffiffiffiffiffiffiffiffiffiffiffi
1 r
2
p
ffiffiffiffi
N
p
:
Theuncertaintyon theordinate atthe origin is:
fIg
xfpg; or fIg¼
x
y
ffiffiffiffiffiffiffiffiffiffiffiffi
1 r
2
p
ffiffiffiffi
N
p
:
Let us turn backtoTable 5.2 and Figure 5.14.We have calculated the coe⁄cientofregression
r ¼0.9375; it is legitimate, then, to try to calculate th e equation of the best straight-line ¢t
whose slope will g ive us the age, and whose i ntercept b will give us the initial isotope ratio.
The calculation gives:
a ¼ r
y
x
¼ 0:013 25 and b ¼ 0:704 84:
By eye, taking a little care, we read a ¼0.0128 and b ¼0.705, which is not so bad in the end!
The uncertai nties are: on the slope, 0.000 347 6, therefore p ¼0.013 25 0.000 34 7 6, which
correspondsi n agetot ¼939 24 Ma.
Theordinate atthe orig in is I ¼0.001567. Hence
ð
87
Sr=
86
SrÞ
0
¼ 0:704 84 0:001 56:
In practice, to be rigorous, we must take account of the experimental unce rtainties on x
and y, which are not necessarily rel ated to the distance from thebeststraight-line ¢t.They
measure the importance attributed to each point in calculating the least squares. The
smaller the uncertainty, the m ore ‘‘important’’ the point is, of course. Introducing these
uncertainties leadsto complications in the mathematics, especially where there are many
data.
All of this is done nowadays by computer programs associated with spreadsheets. But
vigilanceis called for sincetheseleast-squaresprograms avail ableonvarious computers or
inbooks areofvery variabletypesand quality. Generally, theyareprogramsthatdonotpro -
vide for weighting by experimental uncertainties.Where experimental uncertainties are
taken into account, theyareusually uncertainties relativeto a single axisbecause theyoften
178 Uncertainties and results of radiometric dating
a b
c
d
e
y
0 5 10 15
–10
0
10
20
y
0 5 10 15
–10
0
10
20
xx
y
0 5 10 15
–10
0
10
20
x
Regression relative to yCrude data
Regression relative to x
Squared regression
Regression by surface
y
0 5 10 15
–10
0
10
20
x
y
0 5 10 15
–10
0
10
20
x
s = 1.48
y
i
= –2.17
s = 4.05
y
i
= –18.2
d'
f
y
0 5 10 15
–10
0
10
20
x
mean
Squared regression
Regression with variable
and correlated uncertainities
y
x
0 10 20
–5
0
5
10
15
20
s = 2.45
y
i
= –8.22
s = 2.68
y
i
= –10.22
s = slope y
i
= intercept
s = 2.45
y
i
= –8.21
Figure 5.16 The principle of least squares with experimental uncertainties. (a) We consider a cloud of
data points on
x
and
y
. (b) We have calculated least squares relative to
y
, which comes down to
accepting that uncertainty on
x
is negligible compared with that on
y
. (This is the program on good
little pocket calculators, the mathematics of which we have set out.) (c) Here the least squares are
calculated relative to
x
, the opposite case from that in (b): an enormous difference can be seen in the
outcome of these two techniques. (d) By a more elaborate procedure, the least squares are calculated
taking account of uncertainties on
x
and
y
and taking as the criterion the orthogonal projection whose
length is minimized. (d
0
) Right: the true orthogonal projection; left: the representation is distorted to give
a plot with the same look as those before and after. (e) The idea here is to take account of uncertainties
on
x
and
y
by minimizing the areas of the projection triangles. (This is what we did with our simplified
method of calculation.) Notice this method gives acceptable results. (f) We consider each point to be
affected by uncertainties of variable amplitudes on
x
and
y
, which are correlated. So each point is
surrounded by an ellipse of uncertainty. This is the most elaborate program and is the benchmark. Notice
that if we average out the results of (b) and (c), we get values close to (f)
p
¼2.76 and
y
i
¼10.18.
179 Sources of uncertainty in radiometric dating
dominate the others. More often than not, suchuncertainties are considered tobe uniform
forall datapoints.
Programs that do take accountof individual uncertainties on x and y are much rarer and
mustbe search ed out in program libraries. Lastly, there are some even more complete ones
that take account not onlyof individual uncertainties but also of any correlations between
u ncertainties (covarianceagain!).
To gain some idea of these di¡erent progra ms we have illustrated their outcomes for one
example (Figure 5.16).
Again, we must remain clear-sighte d.We can only make rough estimates of geological
u ncertainty. It is not certain, then, that the ‘‘blind’’ least-squares method, which admits of
randomuncertaintieson measurementsandwhichaccordinglymeasuresgeologicaluncer-
taintyimplicitlybypointdispersion, is totallyreliable.However,for wantofanythingbetter,
we choose this method as a ¢rst approximation, while bearing its limits in mind. (Never be
blindedbythe mathematics usedin thenaturalsciences!) Asgeneralreferencesfor this sec-
tion,seeCrumplerandYeo(1940),York(1969),Wendt(1991),BevingtonandRobinson
(2003),andLudwig(2003);seealsotheprogrampackageLudwig(1999).
5.4 Geological interpretations
5.4.1 General remarks
The principle consists in integrating the age (or ages) obtained into an overall geological
scenario in whichagebecomes fully meaningful.This ¢nalstep is n ot without its uncertain-
ties (and risks!).
This maybe an uncertainty in identi¢cation, ifwe makean age correspondto thewrong
phenomenon.Thuswe mayattributean age tothe emplacementofag ranitemassifwhereas
the age is that of late or even subsequent isotopic re-equilibration or conversely the age of
the parentrockremobilizedin thematerialstobe dated,etc.
It may be an unc er ta in ty of inde termination. An age is determined which, on the face of
it, does not correspond to any phenomenon known atthe time.Thus, say, we ¢nd an age of
420 Ma for a granite from southern France. Is it a mixed age between the age of Hercynian
mountain -building 320 Ma ago and of the Pan-African orogeny 550 Ma ago? Or is it evi-
dence of some previously unreported Caledon ian event in the region and so a new discov-
ery? In th is event, the age determined itself poses a problem for geologists, who must
identify the eventbyrelativestructuraldating.
Uncertainties, then, may well arise. Remember that an age is only th e t ranslatio n of an
isotope ratio.Whenever wearefacedwithan interp retationthatisunclear, wehavetoreturn
to the isotopes, theirhistory, tothebasic mod el ofthe clos ed box, and so on. Ofcourse, ages
mustbe determined by using various methods to con¢rm the results. Here again, the com-
mon expression for making comparisons between di¡erent methods is age, but whenever
we wish to interp ret a particular di¡erence we must com e backto the physical systems and
to geochemical behavior. For example, argon di¡uses more readily than neodymium, the
Rb^Sr methodis moresensitivetohydrothermalphenomenathantheU^Pb systemon zir-
con, which inturn maybeagedbyadjunction of inheritedwhole or fragmented zircon, and
soon.We shallexamine afewcases toillustratethesepointswith examples.
180 Uncertainties and results of radiometric dating
5.4.2 Case studies
The aim here is notto achieve systematic coverage but to give a few examples to illustrate a
method, its di⁄culties, its limits, and its level of reliability. This approach will allow us to
test th e various methods of quantitative discussion of geochronology, to compare them,
and tosee how they canbe integrated into the geological c ontext. After noting how we shift
from th e qualitativeuse ofapparent ages toquantitative models (isochron, concordia, step-
wise degassing), wehavetomeasurehowreliable eachapproach is.
Contact metamorphism of the Eldora stock
We have made a qualitative s tudy of discordances in apparent ages from this example. It
is interesting, then, to see what the result was of the various more elaborate methods in
this geologically well-de¢ned case. As these studies, conducted by Stanl ey Hart when a
student atthe Massachusetts Institute of Technology (see Hartetal.,1968), are ratherold,
we only have results for the U^Pb, Rb^ Sr, and K^Ar systems (see Figure 5.1 7 and
Chapter 3).
TheU^Pbsystem on z ircon
We obtain a di scordia cutting the concordia at 1600 50 Ma and 60 5 Ma, giving
apparently the ages of the Precambrian schist and the Eldora intrusion, respectively.
The points on the discordia are placed regularly with distance from the contact. The
mo st di scordant ones are those closest to the contact.
The
87
Rb^
87
Srsystem in isoch ronconst ruction
Themeasurementsweremade main lyon feldsparsbec ausetheymightprovenotverysignif-
icanton the whole rockof the schists.Theyare aligned on an approximate straight line iso-
chron corresponding to an age of1400 100 Ma, except for the apatites and epidotes close
tothe intrusion (
87
Sracceptors).
The( Rb^ Sr, K^Ar)measurements on m icas discussedwiththegeneralized
concordiadiagram
The discordia is a curve an d not a straight l ine as in the U^Pb systems.We obtain the two
inters ections corresponding to the age of the intrusion of 60 Ma and the ‘‘schist age.’’
However, we cannotsay whether thelatter is1480 or1600 Ma. Itis 1480
þ100
20
Ma.
The
40
Ar ^
39
Armeasurements
Th ese were made later on hornblende, biotite, and feldspar (Berger,1975).W e can de¢ne
plateaus giving increasing ages with distance from the contact. The hornble nde and
large biotite minerals give an age of 1450 20 Ma, that is, about the same as the
87
Rb^
87
Srage.
We can envisagetheageoftheschistsas1450 Maand theage measuredbytheU^Pb con-
cordia of 1600 Ma corresponds to a few inherited pieces of detrital zircon. But what does
the sch ist age mean? Is it the age of metamorphism or of earlier sedim entation? The
39
Ar^
40
Ar ages on biotite and the (K^Ar, Rb^Sr) concordia age suggest the age of meta-
morphism, becaus e it was then that the minerals crystallized. The age of 1600 Ma is
181 Geological interpretations
206
Pb/
238
U
207
Pb/
235
U
40
Ar/
40
K
87
Sr/
87
Rb
0.3
1.0
1500
50
54
25'
12'
2'
1000
500
Front range (Colorado) biotite
Line A
2.0 3.0 4.0
0.2
0.1
0.2 0.4 0.6 0.8
0.76
0.74
0.72
1600
1200
800
400
10010
intrusive
age
Zircon
238
U/
206
Pb
Zircon
207
Pb/
206
Pb
Concordia
500
1000
1500
1600
Line B
1440
Concordia
Discordia
87
Sr/
86
Sr
87
Sr/
87
Rb
2400
20
248
85
2400
Epidote
2400
Apatite
2500
r = 1400 Ma
0.4 0.6
290
350
0.8 1.00.20
1500
1000
500
Fraction
39
Ar* outgassed
Hornblende
Apparent age (Ma)
a
b c
d
Figure 5.17 Radiometric ages of the Eldora stock examined in Chapter 3 in terms of apparent ages. (a)
The U–Pb system on zircon (Hart
et
al
., 1968). (b) The two
87
Rb–
87
Sr methods on biotite in the
generalized Rb–Sr, K–Ar concordia. (c) The
87
Rb–
87
Sr method on feldspar. (d) Stepwise
40
Ar–
39
Ar ages
(after Berger, 1975). The numbers indicate distances from the granite intrusion (Chapter 3).
182 Uncertainties and results of radiometric dating
probably that of the earlier granites whose zircon was eroded and redeposited in the clays
from which the schists were forme d. But the questioning an d the answers are themselves a
measureofuncer taintyofthe result.Whatdoes the age offormation ofa schist mean?
T he very ancient rocks of Greenland
The rocks of Greenland are the oldest in the world.
7
They have been studied by various
methods: U^Pb on zi rcon,
206
Pb^
207
Pb on whole-rock isochrons,
86
Rb^
87
Sr, and
147
Sm^
143
Nd. Steve Moorb ath of the University of Oxford was both the pioneer and the
prin cipal investigator (Moorbath and Taylor, 1981; Moorbath et al., 1986). These studies,
combinedwithprecise mapping ofth e terrain, led tothe identi¢cation oftwo separate geo-
logical formations (Figure 5.18), the Amitsoq gneiss and the Is ua Group (th e mountain of
0.7
1.0
2.0
3.0 4.0
3.0 40
0.5
0.3
0.7
0.8
0.1
206
Pb/
238
U
207
Pb/
235
U
3.5
4
2
5
8
9
3
7
6
3.824
+12
–9
3.9
3.8
3.7
3.6
3.5
error
3.65
3.824
2.5
1.5
3
Amitsoq gneiss
t
=
3640
±
120 Ma
ε
Nd,0
(t)
=
0.9
±
1.4
t
=
3776
±
52 Ma
ε
Nd,0
(t)
=
2.0
±
0.6
Baadsgard discordia line
2
1
0.5
0.080.06 0.10 0.12 0.14
0.5102
0.5106
0.5110
0.5098
143
Nd/
144
Nd
147
Sm/
144
Nd
147
Sm/
144
Nd
Isua boulders
0.08 0.14 0.18 0.220.06
0.514
0.512
0.510
143
Nd/
144
Nd
Figure 5.18 Results obtained by various methods for the Amitsoq and Isua gneisses. After Michard-
Vitrac
et
al
.(1977); Moorbath
et
al
.(1997).
7
There are older minerals (zircon) but they are detrital and separated from their parent rock in which they
crystallized.
183 Geological interpretations
references includes: Michard-Vitrac et al.,1977; Moorbath et al.,1986,1997; Nutman et al.,
1996; Kamberand Moorbath,1998).
The ¢rst is dated 3.65 0.03 Ga and the second 3.74 and 3.82 Ga. These are the oldest
groups ofrock inthe world. As canbe seenbyexaminingTable 5.4 and Figure 5.18,summar-
izing the results obtai ned by the various methods, the two groups are quite distinct.
However, small di¡erences can be noti ced in the absolute values obtained by the various
methods and whose precise meaning is unclear.We have no explanations other than that
the rocks were subjectedtosubsequent metamorphism andthe ch ronological systemswere
p robablydisruptedsomewhat.
The Australian group from Canberra has used ion probe point analysis metho ds to
measure zircon grainbygrain and determine the U^Pb ages on the Amitsoq gneiss. A his-
togram can be made (Figure 5.19). It shows thetwo peaks at 3.60 Ga and 3.80 Ga, providing
wetake the mo de ofthe twoasymmetricaldistributions as thevalue.T hetwostatistical ages
seem tobe in agreement withthe ages measured byconventionalmethods.
Letus discuss andinterpretthese results.The distributionoftheAmitsoqandIsuaforma-
tions is clear enough. The age of the Amitsoq gneiss is 3.64 0.05 Ga, with t/t ¼1.3%.
The Isua formation is a little olde r but not as well de¢ned. The age of 3.66 Ga given by
87
Rb^
86
Sr on whole ro cks from Isu a must be considered to be re-homogenization at the
timethe Amitsoqgneissformed.
There seem to be rocks dati ng from 3.82 Ga, particularly in the form of conglomerates
whosezircon‘‘survived’’thegeologicalperturbations.
Table 5.4 Results obtained by various methods for the Amitsoq gneiss and the Isua Group
Dating methods Amitsoqgneiss (Ga) Isua Group (Ga)
U^Pb concordia (zircon) 3.45 0.05 3.824 0.05 (conglomerate)
3.77 0.01(massive rock)
87
Rb^
87
Sr (whole rock) 3.64 0.06 3 . 66 0.06
147
Sm^
143
Nd(wholerock) 3.6 40 0. 12 3. 776 0.05
207
Pb^
206
Pb (whole rock) 3.56 0. 10 3.74 0. 1 2
40
2.80
3.20 3.60 4.00
30
20
10
Number of samples
Age (Ga)
Figure 5.19 Histogram of
207
Pb–
206
Pb ages obtained by ion microprobe methods on Amitsoq zircon.
After Nutman
et
al
.(1996).
184 Uncertainties and results of radiometric dating
But we may surmise that the Isua surface formation is rather aged 3.77 0.08 Ga (t/
t ¼2% ).The age of 3.82 Gais the resultof inherited processes. Along thesame lines, we can
infer that some old zircons have been incorporated into the Amitsoq formation, either
through the erosion^sedimentation cycle or by processes of magmati c assimilation or
metamorphism yielding the old values in the statistical distribution. As can be seen not
everyth ing is clear. Butitdidall happen nearly4 billionyearsago!
Archean komatiites
Komatiites are associations of basic and ultrabasic lavas found in Archean rocks alone (see,
e.g.,Hamilton et al.,1979; Zindler,1982;Bre
¤
vartet al.,1986;Dupre
¤
and Arndt,1986).Theyare
theonlyevidenceofwhatthemantlewaslike atthattime.These associations ofrockhavebeen
dated mainly by Sm^Nd and Pb^Pb systems since the other geochronometers, particularly
Rb^Sr,Ar^Ar, andU^Pb, aregenerally verydisturbedsystems. In addition,the U^Pb, con-
cordiamethodis di⁄culttouseasuranium-rich mineralsarevery rare inthesero cks.The old-
est well-identi¢ed komatiitebelt is theBarberton Greenstone Belt in South Af rica. It is dated
3.4 0.12 Ga (almost as old as Amitsoq!). Several datings havebeen obtained on these rocks
by various methods, allofthem moreorlessconcordant.Herearetheresults.
The whole-rock isochron method g ives:
87
Rb^
87
Sr ¼3.35 0.2 Ga;
147
Sm^
143
Nd ¼
3.54 0.07 Ga;
40
Ar^
39
Ar ¼3.49 0.01 Ga;
206
Pb^
207
Pb ¼3.46 0.07 Ga.Thisposes the
question ofthe exact age ofemplacementofthe komatiites. Is it 3.35 Gaor 3.53 Ga? There is
agap ofsome 200 Mabetweenthe two dates, which is as longas thetime separating us from
the Jurassic. Given the datawe currently have, wehave no criterion for deciding one way or
another, and so choose the value of 3.45 0.10 Ga as the most likely age.The resolution of
such problems will answer the question of the duration of the emplacement episode of
komatiites.
An entirely di¡erent situation is found at Kambalda in Western Australia. Both the
147
Sm^
143
Nd and
206
Pb^
207
Pb methods give very handsome alignments on the isochron
diagrams.Unfortunately, these alignments do notyield the sameage.The
147
Sm^
143
Ndage
is 3.26 Gawhile the
206
Pb^
207
Pb is 2.72 Ga. Both methods are reputed tobe robust.Wh ich
shouldwe choosewhen they fail to agree?
Dupre
¤
and Arndt (1987), then working together at the Max-Planck Institute in Mainz,
showed thatthe
147
Sm^
143
Nd straight lineswere in factstraightlines ofmixing, as shown in
the ("
Nd
,1/Nd) plot (Figure 5.20).The most likely age is therefore 2.72 Ga, which is consis-
tent with the loc al geological context and datings of other associated terrains. Dupre
¤
and
Arndt(1987) generalizedthe discussionofcomparativeSm^NdandPb^Pbageson koma-
tiites and madeasystematic compilation (seeTable 5.5).
Therearethreecaseswheretheageis¢xedtowithin20 Ma:Barbe r ton in SouthA frica,
theAbitibi komatiitebeltof Canada, and Zimbabwe.This asse rtion isbasedonthe concor-
dance of ages determined by both methods and on the geological context and dating of
neighbor inggranitic rocks. Noticethat t/t ¼0.7%.
Cape Smith isaspecial casebecausethe
147
Sm^
143
Nd and
207
Pb^
206
Pb ages are not verydif -
ferentand haveoverlapping marginsofuncertainty.Forwantofanyotherinformation,wemust
putdownanageof1.73 0.1 Ga with t/t ¼5%,whichisnotbadcomparedwiththeothers.
The case of We s t P i l b a ra in Australia is rather similar.The two
207
Pb^
206
Pb measure-
ments seemweakerandincorrectbecausethe geological context argues rather for an age of
185 Geological interpretations
3.56 Ga. It is the context alonethat allows any conclusion, butthe
207
Pb^
206
Pb results indi-
catetherewas a secondary, disruptivephenomenon.
As said, Kambalda is the opposite case. The geological context indicates that it is the
207
Pb^
206
Pb system thatyields the more reliable age.
The results ofthese last two cases show thatthere is nothing automatic aboutthe process
andotherdatingsprobablyneedtobemadebyathirdmethodorconstrainedbyothergeolo -
gicaldata.
T he rhyolite of LongValley,California
It isprobablyon thisveryamenable rockthatthe Universityof California^Los Angeles team
aroundMaryReid ( Reidetal.,1997)madethebestuseofthevariousmethodsfordetermining
theagesofgeologicalunits ofvery recentmagmatic or igin.The ageoferuption andofempla-
cement are estimated, allowing for erosion, as 140 ka. The results obtained by the di¡erent
methods are: 257 ka (
87
Rb^
87
Sr) and 200^235 ka (isochronous
230
Th^
238
U), and the
39
Ar^
40
Arages are muchyounger, rangingfrom100 to150 ka (Reid etal.,1997).
The discordances here (several thousandyears) are construed as not arising from subse-
quent disruptive phenomena but rather from complex magmatic sequences (Figure 5.21).
Table 5.5 Comparative Sm–Nd and Pb–Pb ages on komatiites (these are solely ages for which
alignments are statistically reliable)
147
Sm^
143
Ndages
206
Pb^
207
Pb ages Mostlikelyage
West Pilbara 3.5 Ga 3.56 0 Ga 30 Ma 3.15 Ga 3.5 Ga?
Barberton3.5 Ga 3.540 Ga 30 Ma 3.46 0 Ga 70 Ma 3.45 Ga
Abitibi Munroe
Alexo(2.7 Ga)
Newton
2. 662 Ga 120 Ma
2.7 50 Ga 90 Ma
2.826 Ga 64Ma
2.724 Ga 20 Ma
2.690 Ga 15 Ma
2. 65 Ga 15 Ma
2.7 Ga
Kambalda 2.7 Ga 3.260 Ga 40 Ma
3.230 Ga 120 Ma
2.720 Ga 105 Ma
2.730 Ga 30 Ma
2.7 Ga
Zimbabwe2.7 Ga 2.640Ga 140 Ma
2.933 Ga 147 Ma
2.690 Ga 1Ma 2.7 Ga
CapeSmith 1.8 Ga 1.871 Ga 100Ma 1600Ga150Ma 1.73Ga
3
4
5
2
1
0.5 1
1/Nd
ε
Nd
0
Figure 5.20 Diagram of (2
Nd
, 1/Nd) plot on komatiites of Kambalda (Australia) showing a correlation in
agreement with the idea that the rocks are a mixture. After Dupre
´
and Arndt (1987).
186 Uncertainties and results of radiometric dating