Thus two types of phenomena are opposed: those producing chemical and isotopic dis-
persion and those which mix, homogenize, and tend to destroy the heterogeneities (see
McKenzie ,1979;Alle
'
greetal.,1980;Alle
'
gre and Turcotte,1986).
Exercise
Suppose that unaltered oceanic crust, altered oceanic crust, and fine sediments plunge
into the mantle in a subduction zone. The extreme compositions for the
87
Rb–
86
Sr
system for unaltered oceanic crust are
Sr
0
¼0.7025 and
Rb/Sr
¼0.1 and for the sediments
Sr
S
¼0.712 and
Rb/Sr
¼0.4. The altered crust has intermediate values:
Sr
0A
¼0.705 and
Rb/Sr
¼0.2.
What will be the dispersion of the isotope ratios if this subducted oceanic crust is in the
mantle without mixing for 1 billion years?
Answer
Dispersion can be evaluated by calculating the two extreme values. We obtain 0.703 92 and
0.717 68 respectively for the pieces of unaltered oceanic crust and for the sediments. The
difference is 0.0137, while the difference in
Sr
values during subduction was 0.0095.
We can also calculate it directly:
D ¼ðDÞ
initial
þ lðDÞ
T
:
Hence indicates the dispersion: ¼0.0095 þ0.004 26 ¼0.0137. (Notice that given the
value of
l
, the term –
l
is negligible in the evolution of .)
Suppose now that these subduction products are subjected to multiple mixing processes in
the mantle. Suppose the result is that
i
, which was 0.3 initially, becomes ¼0.1, and that
becomes 0.003.
The effectiveness of this isotopic mixing can be measured by the ratio 30/137 ¼0.21
whereas chemical mixing, measured by 0.1/0.3 ¼0.33, is not quite as effective. (This reason-
ing probably fails to allow for isotope exchange, but it is a first approximation.)
Letus try togeneralizetheses imple examples.Wearelookingtowriteanequationderived
fromWasserburg’s, but dealing with distributions. So we take as vari ables not the average
values, which has already been done, but the dispersions. Dispe rsion can be measured by
the standard deviation (or by the standard d eviation of two extreme values, which, as we
k now, areaboutthree times the standard deviation) (Alle
'
gre and Lewin,1995).
Thefollowingequation canbeder ivedfromWasserburg’s equationsby notingdispe rsion
hi
.Thus we note the dispersion of isotope ratios
hi
and the dispersion of chemical ratios
hi
withsubscripts
ex
(external),
i
(internal),
ex
(external), and
i
(internal).
d
i
hi
dt
¼
ex
hi
i
hi
½L M
i
hi
where the term ^l
i
hi
is ignored (Figure 8. 17).
The new term thathasbeen intro duced is M
i
hi
, the term ofhomogenization ofthe mix-
ture; M has the dimension oftheinverseofatimewe shall call mixingtime (). Itis the time
it takes to reduc e the dispersion
hi
by a factor (e) exponential. Suppose a steady state is
attained, then:
468 Isotope geology and dynamic systems analysis