to find early Archaean basement rocks in the amphibolite and retrogressed granulite facies or-
thogneisses adjacent to the quartzites. So far, all samples dated by SHRIMP U – Pb geochronol-
ogy, appear to be strongly deformed late Ar- chaean
granitoids Song
and Nutman,
unpublished data. Thus, presently these quartz- ites seem to be an isolated occurrence as early
Archaean orthogneisses associated with them have not yet been detected nearby.
SHRIMP U – Pb zircon results from fuchsitic metaquartzite sample CF8926 were reported by
Liu et al. 1992 and are summarised here Table 1. The youngest detrital zircon in the metaquartz-
ites is ca. 3550 Ma, interpreted to be closed in age to the deposition of the quartzite, estimated at
3500 9 80 and 3470 9 107 Ma from Sm – Nd iso- topic studies of associated amphibolites on nearby
outcrops Huang et al., 1986; Jahn et al., 1987. The detrital zircons obtained from fuchsite
metaquartzite mostly have
207
Pb
206
Pb ages be- tween 3600 and 3850 Ma Liu et al., 1992. The
distribution of ages obtained fall into several ‘peaks’, which suggests that the geological history
of the source region experienced several distinct magmaticthermal events in the 3600 – 3850 Ma
period, rather than a continuum of activity.
6. Ages of zircons in early Archaean sediments
6
.
1
. Data filtering and interpretation of ages Later disturbance to the grains can introduce
common Pb into the zircons and also give rise to discordant ages. As a filter of the data, analyses
that are \ 10 discordant contain unsupported U, \ 5 reverse discordant containing unsup-
ported radiogenic Pb or clearly young in situ metamorphic growths have been rejected. Also
analyses with a high component of common non radiogenic Pb f
206
Pb \ 2, where f
206
Pb is the proportion of non-radiogenic
206
Pb have been rejected. Finally, where duplicate analyses of indi-
vidual grains have been done, only the most con- cordant generally that with the oldest
207
Pb
206
Pb age has be included in this study. All these
measures were undertaken to reduce distortion of age distributions in the detrital populations.
As only close to concordant ages are discussed, the data are not presented in the
206
Pb
238
U versus
207
Pb
235
U or
207
Pb
206
Pb versus
238
U
206
Pb con- cordia plots. Instead, the filtered ‘best’ analyses
are presented as composite histograms and rela- tive probability diagrams Figs. 3 and 4. The
histogram is constructed by binning data into 10 Ma intervals and the relative probability plots
shown in the background are age spectra, where the analytical uncertainty on the age of each
detrital grain is taken into account. Thus, the tails in these diagrams to ages of \ 3900 Ma do not
indicate that detrital grains of these ages have actually measured. Instead, they indicate that the
pooled data indicates that it is possible, but highly unlikely, that grains with these ages could occur.
6
.
2
. Data sets Data on most of the ] 3500 Ma sediments
discussed in this paper have been published in isolation elsewhere PDK fuchsite quartzite and
GGU221122 — Kinny, 1987; Kinny et al., 1988, MR81318 and 83187 — Nutman and Collerson,
1991, G9155 — Nutman et al., 1996, G9325 — Nutman et al., 1997a; CF89 – 26 — Liu et al.,
1992. Data on two additional samples have been presented here VM9010 and G9354, Table 2.
All the data are filtered according to the methods outlined above, and then collated into suites de-
posited at 3500 – 3600 Ma five samples, 117 grains and 3700 – 3800 Ma three samples, 54
Fig. 3. Probability of detecting detrital components of differ- ent magnitude in zircon populations.
Fig. 4.
207
Pb
206
Pb age histograms with relative probability curves in background for detrital grains in sediments de-
posited at 3500 – 3600 and 3700 – 3800 Ma.
majority of its the zircons are markedly discor- dant or are metamorphic in origin Schiøtte and
Compston, 1990.
6
.
3
. Estimation of the maximum contribution of ]
3900
Ma detrital material in ancient sediments The probability P of missing an age component
present at an abundance level x after analysing n grains is P = 1 − x
n
Compston and Pidgeon, 1986; Dodson et al., 1988. For example, for a
population present at the 5 level, the probability of missing it after analysing 20 grains is 0.36, but
after 50 grains has fallen to 0.08 Fig. 3. Thus, for a low probability P = 0.05 is used in this
paper of missing small components in detrital zircon populations, it is necessary to analyse as
many grains as possible.
In both the suites deposited at 3800 – 3700 and 3600 – 3500 Ma, \ 3900 Ma grains were not
found Fig. 5; Appendix A. For the 3800 – 3700 Ma suite where 54 grains are available, there is a
0.05 probability that \ 3900 Ma zircons form 5
5 of the detrital population, but were missed during analysis. Likewise, for the 3600 – 3500 Ma
suite where 117 grains are available, there is a 0.05 probability that \ 3900 Ma zircons form
5 3 of the population were missed Fig. 3.
Older than 3900 Ma detrital zircons have so far only been encountered in some samples of the
Jack Hills and of Mt. Narryer metasediments deposited at ca. 3000 Ma; Kinny et al., 1990 of
the Narryer Gneiss Complex in Western Aus- tralia, and in quartzites of the Wyoming Province,
deposited at 3000 – 3200 Ma Mueller et al., 1992. Although \ 3900 Ma zircons form : 6 of the
population in one Jack Hills conglomerate sample Compston and Pidgeon, 1986, when all of the
analysed sediments from the Narryer Gneiss Complex are pooled, the proportion of \ 3900
Ma detrital grains is 5 1 Nutman et al., 1991. When the 3000 – 3200 Ma global sedimentary
record is taken into account now well over a thousand detrital grains analysed from many lo-
calities the proportion of \ 3900 Ma detrital grains is considerably less. However, because
700 – 900 million years had elapsed between 3900 Ma and the deposition of these sediments, the
grains for examination of detrital grain age spec- tra Table 1, Fig. 4.
Felsic rocks interpreted to be of volcanic origin such as 3776 9 8 Ma Nulliak supracrustal associa-
tion sample DB82-15Z-A , Schiøtte et al., 1989b,
3718 9 4, 3710 9 4
and 3806 9 2
Ma Isua
supracrustal belt samples Compston et al., 1986; Nutman et al., 1997b are excluded from the
database because their zircons are volcanic, with none inherited from older zirconiferous crust. Zir-
con age data from rusty, early Archaean, Green- land gneiss sample GGU242521 Schiøtte and
Compston, 1990 are also excluded, because from field relationships it is likely that this rock is a
granitoid contaminated by adjacent older banded iron formation, rather than a sedimentary rock.
Regardless of the true nature of this sample, the
observed proportion of \ 3900 Ma detrital grains can be accommodated in a wide spectra of crustal
growth and recycling models.
6
.
4
. Early Archaean crustal growth 6ersus recycling
Whereas there is now a substantial global data- base of detrital zircon ages from 3200 – 3000 Ma
sediments from many localities, the database for older sediments is much more restricted. If the
3500 – 3600 Ma detrital sediment suites are repre- sentative of their source terranes, their detrital
zircon age distributions indicate that constant crustal volume recycling = new additions models
require that \ 97 of \ 3900 Ma crust was destroyed by recycling by 3500 Ma Figs. 1 and
3. The lack of detected \ 3900 Ma detrital zir- cons in three 3700 – 3800 Ma detrital sediments
supports this conclusion, and provide even older constraints, requiring that \ 95 of \ 3900 Ma
crust was destroyed by 3700 Ma in constant vol- ume models.
Such rapid recycling rates are hard to envisage, unless the whole lithosphere was being ‘reworked’
by major impactmelting events as late as 3500 Ma. However, such a scenario may be ruled-out,
because since ] 3850 Ma Earth has retained a hydrosphere Nutman et al., 1997a which even
then harboured life Mojzsis et al., 1996. Further- more, such extremely high recycling rates might
be hard to reconcile with the spread of initial Nd and Sr isotopic ratios of well preserved early
Archaean granitoid suites e.g. Fig. 2, Baadsgaard et al., 1986; Bowring and Housh, 1995, which
require significant average crustal residence times of several hundred million years to permit radio-
genic isotopic systems to evolve. Both these con- siderations would seem to rule-out extremely
rapid recycling as the explanation of the low abundance of \ 3900 Ma detrital zircons in the
3500 – 3800 Ma sedimentary record. Thus, from the detrital zircon perspective, it is concluded that
in the crustal segments preserved in the early Archaean gneiss complexes in northern Labrador,
western Greenland and northeastern China, the volume of continental crust at 3900 Ma was small
and that it increased in the period 3900 – 3500 Ma.
7. Discussion