level and do not include the errors in the 40Ar*/39ArK ratio and age of the monitor. The error in the 40Ar*/39ArK ratio of the monitor is included in the "plateau age" error bar calculation.
Twenty-five mg of plagioclase (160-200 µm) were separated by a magnetic separator and then hand-picked under a microscope to select only the most transparent grains for analysis. The samples were irradiated for 70 hr in the McMaster reactor (Hamilton, Canada) with a total flux of 9 × 1018 n/cm2. We piled up the samples and monitors along the axis of the irradiation canister. Eleven standard Hb3Gr hornblendes (1072 Ma) were included in the 90-mm-long sample pile (Table 2).
In spite of a total flux gradient of 4% from the bottom to the top of the canister, the neutron flux received by each sample was known to within ± 0.2%. The analytical procedure was as described by Féraud et al. (1986). All errors in apparent ages and total ages are quoted at the 1 level and do not include the errors in the 40Ar*/39ArK ratio and age of the monitor. The error in the 40Ar*/39ArK ratio of the monitor is included in the "plateau age" error bar calculation.
From low to high temperature steps, both age spectra (Fig. 3) are characterized by a decrease of apparent ages (from 500° to 760°C), followed by a sharp increase in age, giving a flat section of over 72% (sample 83R77-82) and 44% (sample 85R18-23) of 39Ar released, respectively. Then the high temperature ages increase regularly.
The very disturbed apparent ages displayed at low temperature correspond to very variable and increasing 37ArCa/39ArK ratios probably indicating the degassing of K-rich secondary phases included in the plagioclase (as observed under the microscope). Another phenomenon which may affect these low temperature ages is excess 40Ar, as suggested by both the increasing ages at high temperature (up to 148 and 160 m.y. for samples 83R77-82 and 85R18-23, respectively), and, more clearly, by the high apparent ages (up to 180 m.y.) obtained at low temperature on sample 85R18-23. The classical U-shaped age spectra commonly displayed by plagioclases affected by excess argon are probably disturbed at low temperature by the additional degassing of young alteration phases.
The larger error bars obtained on the apparent ages of sample 85R18-23 are due to a lower K content (but similar Ca content), resulting in a higher relative contribution of Ca-derived interference isotopes and in atmospheric argon contamination.
The determination of an unambiguous, geologically meaningful age from these preliminary data is difficult because of the relative importance of secondary phases in the K-poor plagioclase. The questions we have to resolve are: (1) are the weighted mean ages (given in Fig. 3) calculated on the flat regions at intermediate temperatures geologically significant, and (2) how can we explain the difference between these two "plateau ages." We notice that despite a distinctly different initial grain size, these two rocks were originally similar and suffered the same tectono-metamorphic evolution. Moreover, they are only 16m distant, and therefore the two analyzed plagioclases should have recorded the same geological history.
The apparent ages of the flat region of the sample 83R77-82, ranging from 134.5 ± 0.8 to 138.8 ± 1.2
m.y., do not define a 40Ar/ 39Ar plateau age (which is the usual validity criterion for accepting a geologically significant age) because these extreme apparent ages are not concordant (even at the 2 level) with the weighted mean age of 136.4 ± 0.3 Ma. The apparent ages are clearly correlated with the
37ArCa/39ArK ratios, indicating a probable higher contribution of younger K-rich secondary phases near 1100°C (as shown by Sebai et al., 1991, on altered plagioclases). The 36Ar/40Ar vs. 39Ar/40Ar correlation diagram on steps 820°-1350°C (not given) displays an age of 133.9 ± 0.7 Ma (Mean Square Weighted Deviation = 1.6, initial 40Ar/ 36Ar ratio = 318±6).
Figure 4 reports the measured intensities of the isotopes 40Ar*, 39ArK, and 37ArCa per °C, vs. temperature. We observed one main and wide degassing peak of 40Ar* and 39ArK in Sample 83R77-82, whereas these isotopes were released in two temperature domains (two peaks) for sample 85R18-23. The 37ArCa is degassing in two peaks in both cases. The two "plateau-ages" were defined in the same temperature domain, and correspond to the first degassing peak of 40Ar*, 39ArK, and 37ArCa for sample 85R18-23. The increasing high-temperature apparent ages displayed by sample 85R18-23 correspond to the second degassing peak of the three isotopes. Because the temperature of each step and the weight of analyzed samples were equivalent for the two experiments, we can quantitatively compare the degassing curves. The similarity of 37ArCa and the difference between both 40Ar* and 39ArK degassing curves for the two samples clearly show that the two peaks of 37ArCa (both samples), and of 39ArK and 40Ar* (sample 85R18-23) correspond to the degassing of "pure" plagioclase, whereas the single peaks of 39ArK and 40Ar* (sample 83R77-82) correspond to a mixture of "pure" plagioclase and secondary phases included in the plagioclase.
When we compare the Ca/K ratio deduced from the 37ArCa/39ArK ratio (by the relationship [K/Ca] = 0.539 × [39ArK/37ArCa]) with the microprobe results, we observe (Fig. 3) that the Ca/K ratio of the analyzed plagioclase 85R18-23 (at intermediate and high temperatures) almost corresponds to the composition of the plagioclase measured by the microprobe (the mean values were calculated from a more extensive data set than in Table 1), whereas the Ca/K ratio given by the plagioclase 83R77-82 is much lower. The combination of this observation with the interpretation of the degassing curves previously discussed clearly shows that the "plateau" segment of Sample 83R77-82 corresponds to the degassing of (1) the plagioclase itself and (2) secondary K-rich phases. Therefore, the validity of such a "plateau-age" must be viewed with caution. Nevertheless, we notice that on this "plateau" segment, the significant variation of the 37ArCa/39ArK ratio (by a factor of 3) due to a strong and variable contribution of secondary K-rich phases produces maximum variations of apparent ages of about 3% only. This probably indicates that these secondary phases are not much younger than the closure time of the plagioclase.
The apparent ages of the flat region of sample 85R18-23 range from 119.8 to 127.2 m.y., corresponding to 44% of the total 39Ar released (such a fraction is much too small to consider this flat region as a plateau-age), with error bars on the order of ± 2.3%-3.5%. The apparent age variations at low temperature (from 180 to 88 m.y.) and at high temperature (from 124 to 160 m.y.) are greater than for Sample 83R77-82. Both the 37ArCa/39ArK ratio spectrum and the degassing curves in Fig. 4 show that this "plateau" section corresponds to the degassing of plagioclase less contaminated by K-rich secondary phases. The greater variation in apparent ages of this sample (on the whole age-spectrum) is probably a consequence of a relatively greater effect of alteration processes and excess argon on a K-poor sample. Albitization of the plagioclases in this rock is clearly observed in thin section and detected by the microprobe analyses (Table 1), although sericite was not seen in these plagioclases. If this alteration is significantly younger than the K/Ar closure time of the plagioclase, its effect on the apparent ages is higher for a nearly pure plagioclase (poor in K) (85R18-23) than for a similar plagioclase (83R77-82) contaminated by a K-rich phase (sericite) nearly contemporaneous with the K/Ar closure of the mineral. If this model is correct, we are in a peculiar situation where a K-rich contaminated plagioclase gives a more reliable age than a relatively purer (with regard to K content) plagioclase.
