1 Fitting of ochrons a1
Vugrinovich (1981) Butler (1982), Rockand Duffy (1986),
Ganguly and Ruiz (1986), Rock etal. (1987), Cavazzini
(1988), Castorina and Riccardo (1989),Kostitsy (1989),
Kalsbeek and Hansen (1989), Haack (i990), Kent etal
4990), Provost (1990), and Wendt (1093), Harmer and
Eglington (1990) published a useful review of the statis
tical principles used to fit straight lines to data points.
Ludwig 1992) developed a program for fitting isochrans
‘on IBM PCs based on the equations of York (1966, 1969),
Kullerud (1991) reconsidered the use of analytical er-
rors to obiain weighting factors and emphasized the
importance of determining these ertors accurately in
sets of samples having a wide range of Rb and
centrations. Additional shortcomings of the Rb-St
isochron method of dating have been pointed out by
Schleicher etal, (1983) and Lutz and Srogi (2986).
Brooks etal, (1972) examined the different statistical
procedures proposed by Mclatyee et al, (1966), York
(4900, 1909), Brooks etal. (1968), and by Wendt (1969).
Using simulated testeata,they demonstrated thatall four
models of Mcintyre etal. (1966) yield identical results
in those cases where the observed scatter of the data
points can be accounted for solely on the basis of ran-
dom analytical errors. in ses, the line detined by
the data isa true izockron because the data satisfy the
assumptions for dating. In cases where the seatter ex
ceeds analytical error, the slope and intercept ofthe best
fit ine (errorchron or seatterchron) vary depending on
the statistical model that is used, and therefore may not
be reliable, The dates and initial "'Sr/™Sr ratios derived
from errocchrons may, nevertheless, convey gealogically
useful informationsespecially in cases where additional
facts or judgments support them, such as the results of
dating by other
All of the weighted
by Brooks et al. (1972) make use of a statistical index to
test the goodness of fit of the data points to the best
steaight line. The index used by Metntyre et al.(966) is
the mean square of weighted deviates (MSWD), The nu-
merical value of MSWD for a set of analytical dita can
be used to decide whether the scatter of data points is
consistent with the magnitudes of the
(Wendt and Carl ,g91), If the regression is based! on a
very large number of data sets and if the analytical er
r
duplicates, then a value of unity or less for MSWD signi
fies that the line is an isociran. However, when the
number of data sets and duplicates is less than in the
ideal case the limiting value of MSWD increases, The
magnitudes of the limiting values of MSWD for ditter-
ent numbers of data sets and duplicates are given in an
F distribution tablein Appendix 3 of Brooks et al.(1972)
For example, in case a regression is based on
sets and five duplicates, the limiting value of MSWD is
51 at the 95% conlidence level. Limiting values of
pic methods
gression procedures reviewed
s have been calculated from a very large nursber of
data
MSW for different numbers of duplicates and ata sets
19 MSI, 95% confidence
Number of amples regressed
Fig. 1.4. Limiting vaos of the mean square of wei’
asd duplicates used in the "leasLsquare eubic” regressing of
Melmtytect al 1966) ine York (a6) Ihe SW caletated for
1 cata set equal to «(ess than the theoretial value he scatter
i points is ennsient with the analytical errars 4
tm lie isan iushirone I the MAID is ges 3
p
I vale, the regression line is an errorchra whose sone and
intercept may be inaccurate and require confirmation by other
evidence amrelated tothe regression (Sourer Hrooks eral. 19
npendix
are presented in Fig 14 based on Appendix 9 of Brooks
tal. (
The data in Fig. 1.4 indicate that the distinction be
tweeen sechvons and errorchrons fora givea number of
data sets depends on the number of duplicates used to
evaluate analytical errors. For example, MSWD = 6. f
ten samples and four duplicates indicates that the re
gression line isan isochron, However if the number of
duplicates is greater tham 4, MSWD = 8.0 indicates that
the line is an errorchron and that geological error is
present nother words, as analytic
tolerance for scatter decreases and regression lines be
come errarchrans or scatterchrons
Geological errors may be associated with all of the
samples in a set or with only one or two-of the samples.
The former case may arise in dating voleanic rocks which
‘may have different initial “Sri"Sr ratios because the
‘magma from which they formed may have assimilated
tstal rocks containing varying amounts of radiogenic
Sn. only one or two specimens in a suite of samples
deviate significantly trom the best-fit line, they can be
excluded from the regression. The remaining samph
may then yield an isochron irom which a reliable date
cam be derived. The geological erro
dleviant simples is a clue that they may have been af
fected by geological events not reconded in the other
samples, Therefore,samples baving geological error de
the cause for their
ecognized in the
serve additional study to determin
deviation from the isochron,2 __Cuarten 1 - Chemical Properties and Isotope Systematics
1.5. The Sm-Nd Methed of Dating
‘The sensitivity of Rb-Sr dates to altsration mandates that
such rocks sheuld also be dated by the Sm-Nd method
because i is less Vulnerable toalteration. Samarium (Sin)
and neodymium (Na) ate both rate-earth elements and
feccurin rock-torming minerals in low but easily meas
terable concentrations (Faure 1980).
‘Samarium (Z = 62) hss seven stable and naturally ce
curring isotopes whose abundances are: "Sm = 518s
Sm 15,9855 #8 = 1h.29%5 "Sin = 13.8 Sm = 2.
Sm = 26.7% 81S = 22.896, Samarinm-147 is radi
active and decays by alpha emission ta stable Nd with
a hall-life of 1.06.» 10" years (A =6.54 x10" yr"),
Neodymium (Z.= 60} also has seven stable isotopes
whose abundances ate: '#'Nd = srs “Nd = 12.243,
agiots! Nd = 8.3% Nd = 173%!"
5.0%. The abundances of the Nd isotope
1cks and minerals change with time because ofthe for
imation of radiogenic "Nd by alpha decay of '"”Sm.
‘The 'ONd/"Nd satios of rockssaind minctals are ve
lated to the '"’smy!"'Nd ratios and to time by the equa.
on
a
xd (193
This equation is-analogous to Eq. 1.12 for the Rb-Sr
system and is the basis for the Sm-Nd isochron method,
of dating, Sets of whole-rack samples having the same
age (t) and the same initial "'Ne/"""Né ratio define
the Sm-Nd isachron in coordinates of '*'Nd/'*!Nd and
'Sm/!"'N4. The stope (mn) of the isochron is related to
the age of the rocks that define the isochron by:
mae (0.20)
1
f= ting ty
Insome cases, Sn hnole-rock isochron dates are
older than the Rb-Sr Isochron dates of the same rock
specimens because Sm and Nd tend to be less mobile
during thermal or hydrothermal alteration, For this rea
son, the Sm-Ni method is preferred for dating metavol
canic rocks of Archean and Proterozoic age
1.6 The Epsilon Notation
The epsilon notation was introduced by DePaolo and
Wasserburg (1976a,b) in order to facilitate the interpre
lation of measured '"'Nal'*Nd ratios of basaltic vol:
canic rocks derived from magma sources ip the mantle
‘The epsilon parameter (Na) compares the measured
'8Nd/""Nd ratio of a rock or mineral sample to the
present '"Na/'"Nd ratio of a hypothetical Chondritic
Uniform Reservoir called “CHUR’ and is defined by the
equation:
[UNAH NG
800) =| a
[Pina Nake
108
(a0)
where
( "na
“Nd.
J
‘The epsilon value can be recalculated so that it refersto
a time in the post. The appropriate value ofthe "Ndi! *Nd
ratio of CHUR is obtained from Eg, 1.12
|
= 0.591847 oF 01512638 (clepending on
the way the isotope fractionation cor
rection isapplied) is the present value
of this ratio in CHUR (DePaolo 1988),
Joa |
Na
Nd
tng
(uaa)
where
16m
= value of this ratio at the present time
1967 (DePaolo 19885 Wasserburg
etal1981)
Na)!
4) cuux tio at the present time)
Epsilon values referred to a time in the past are des
ignated by &'(Na) where ris the specified age.
The '*!Ndi*Nd ratio of the chondritic uniform
ervoir is affected by the way in which the measured iso-
Lope ratios of Nd ate corrected for isotope fractiona~
Hon. Most isctope geochemists base the isotope frac
tonation correction of measured ''Nd/"*Ned ratios on
g for the Nd/"Nd ratio, whereas oth:
Na = 0.636151 as the reference ratio,
value of 0.7:
sonra
Wasserburg et al, (1981) demonstrated the relation
among the isotope ratios shown in Table 1.6. Both
schemes yield identical ”(Nd) values provided that the
'ENG/!"Nd ratios of the rock samples were corrected