X hits on this document





36 / 63

evidence for boiling is absent. The Au grades are essentially identical (0.4 ppm Au) in

samples containing boiling FIAs and those in which they are absent (Fig. 10). The Ag

grade is slightly elevated in samples containing coexisting liquid-rich and vapor-rich

fluid inclusions, compared to those in which fluid inclusion evidence for boiling is absent

(67.1 ppm Ag versus 49.4 ppm Ag; Fig 11).

Data shown on Figures 10 and 11 were evaluated using the t-test to assess

whether the average grades (means)for samples in which a feature is present are

statistically different from those in which the feature is absent. Differences between the

presence or absence of a feature indicative for boiling are statistically different at the 95%

confidence level (p<0.05) for all features except jigsaw texture silica (p<0.10) and fluid

inclusions (p<0.25) for silver, and for fluid inclusions (p<0.10) (Fig. 12). For

comparison, the distribution of grades for samples in which massive quartz (a non-boiling

texture) is present and absent is shown on Figure 12. Note that the sample groups are

statistically different for Au, but that samples containing massive quartz are lower than

those in which it is absent. This is consistent with the observation that Au (and Ag) is

closely associated with textures indicative of boiling and is not related to non-boiling


The textural and fluid inclusion data obtained in this study were analyzed using

the binary classifier within the statistical software package SPSS Clementine. The binary

classifier builds 10 different statistical models and evaluates the performance of each

model (Fig. 13a, b). The Au and Ag grades show a log normal distribution based on

application of the Shapiro-Wilks test for normality (Fig. 14a. b). Our goal was to develop

and test models that could distinguish between samples with high ore grades (=1 ppm Au


Document info
Document views222
Page views222
Page last viewedThu Jan 19 22:26:18 UTC 2017