382 M.K.G WHATELEY, T. BELL & C.J.

MOON

16
Disseminated Precious
Metals – Trinity Mine,
Nevada
Michael K.G. Whateley, Timothy Bell
and Charles J. Moon

Disseminated precious metal deposits, particu- be little different from the surrounding rocks
larly of gold, have been the most popular target and the margins of the deposit are only distin-
for metallic exploration through the 1980s guishable by assay. An excellent summary is
and 1990s. Their popularity was based on the given in Bagby and Berger (1985), updated in
high price of gold during this period (see sec- Berger and Bagby (1991) and Hofstra and Cline
tion 1.2.3), the low cost of mining near-surface (2000).
deposits in open pits, and improvements in Individual deposits occur mainly within
gold recovery techniques. The key metallur- major linear trends, up to 34 km in length, the
gical innovation was the development of the most important of which is the Carlin Trend
cheap heap leach method to recover precious in Northeast Nevada. This trend and type
metals from low grade oxidized ores. This case of deposit are named after the Carlin Mine,
history deals with the evaluation of the Trinity which was the ftrst signiftcant deposit of this
silver mine. type to be recognized, although others had
previously been mined. The deposits vary from
broadly tabular to highly irregular within
16.1 BACKGROUND favorable beds and adjacent to structures such
as faults that have acted as fluid pathways.
The faulting and fluid flow has often resulted
16.1.1 Overview of deposit types
in brecciation and several deposits are hosted
A variety of deposit types have been mined in in breccias.
this way but the most important in Nevada Siliciftcation is the commonest form of al-
(discussed in detail by Bonham 1989) are: teration and the occurrence of jasperoid, which
1 Sediment hosted deposits of the Carlin type. replaces carbonate, is widespread in many de-
2 Epithermal deposits, such as Round Moun- posits. Gold is thought to occur predominantly
tain. as native metal although its grain size is very
3 Porphyry-related systems rich in gold. small, usually less than 1 m, and the metal’s
exact mineralogical form remains unknown in
some mines. Gold forms ftlms on sulftde and
Carlin-type deposits
amorphous carbon and is particularly associ-
These deposits are responsible for the majority ated with arsenian pyrite. Pyrite is the most
of gold production in Nevada, 261 t in 2001. common sulftde and may be accompanied by
Typically the gold occurs as micronmeter-sized marcasite and arsenic, antimony and mercury
grains which are invisible to the naked eye sulftdes. The arsenic sulftdes realgar and
(“noseeum” gold), within impure limestones or orpiment, which are readily distinguishable by
calcareous silstones. The host rocks appear to their bright color, occur in many deposits as
 16: TRINITY SILVER MINE, NEVADA 383

et al. 1987, Bonham 1989, Henley 1991,

Hedenquist et al. 2000).
Fumeroles and hot springs The high sulftdation or acid sulfate type
SURFACE generally contains more sulftde and is char-
acterized by the occurrence of the copper
sulftde, enargite. Other base metal sulftdes
Fine-grained clastic
sediments are frequently present as is pyrite. Advanced
argillic alteration is common as is alunite
Fluid flow (K2Al6(OH)12(SO4)4) and kaolinite. The low
Jasperoid sulftdation or adularia sercite type generally
Orebody has less sulftde and sercitic to intermediate
argillic alteration. Alunite may be present but
it is only of supergene origin. The structural
Thin bedded, setting of both types may be similar (Fig. 16.2)
carbonate cemented, but the differences between the two types prob-
carbonaceous,
clastic sediments
Limestones
and dolomites
ably reflect their distance from the heat source
(Heald et al. 1987). There are obvious parallels
between the formation of these deposits and
Felsic or modern geothermal systems. Evidence from
intermediate modern day geothermal systems, such as the
intrusive
500 m deposition of gold on pipes used to tap the
systems for geothermal power, suggests that
the gold may have been deposited rapidly.
FIG. 16.1 Diagram for the formation of Carlin style Several major gold deposits occur in alkalic
deposits. (After Sawkins 1984.)
volcanics that may be spatially associated with
alkalic porphyries. The deposits are character-
does stibnite and cinnabar. The general geo- ized by the occurrence of the gold as tellur-
chemical association is of As, Sb, Tl, Ba, W and ides with quartz–carbonate–fluorite–roscoelite
Hg, in addition to Au and Ag. (vanadium mica)–adularia alteration and re-
The deposits are spatially associated with gional propylitic alteration (Bonham 1989).
granites and many authors favor intrusion The Hot Spring deposits of Bonham (1989)
driven hydrothermal systems (Fig. 16.1). Berger appear to be a variant of the epithermal type
and Bagby (1991) favor a mixing model in and are characterized by the occurrence of a
which magmatic fluids have been mixed with recognizable paleosurface that is usually asso-
meteoric water causing precipitation in favor- ciated with a siliceous sinter zone interbedded
able host rocks. with hydrothermal breccia. The sinter passes
downward into a zone of siliciftcation and
stockwork with a zone of acid leaching. The
Epithermal systems precious metals are generally restricted to
Much of the bonanza-style precious metal within 300 m of the paleosurface and probably
mineralisation in the western USA is hosted in reflect rapid deposition. There is an overall
volcanic rocks, e.g. the Comstock Lode which spatial association with major centers of
produced 5890 t Ag and 256 t Au at an average andesitic to rhyolitic volcanism
grade of about 300 g t1 Ag and 13 g t1 Au in the
late nineteenth century (Vikre 1989). More Gold-rich porphyries
recent exploration has resulted in deftnition of
bulk mineable targets such as Round Moun- Gold is of major economic importance in
tain with reserves of 196 Mt of 1.2 g t 1 Au. several porphyry deposits that produce copper
Volcanic hosted deposits can be divided into as their main product, such as Ok Tedi (see
three general types based on mineralogy and p. 273). More recent research (Sillitoe 1991)
wall rock alteration: acid sulfate (or alunite– suggested that there may be related copper-
kaolinite), adularia sericite, and alkalic (Heald deftcient systems that contain economic gold
384 M.K.G WHATELEY, T. BELL & C.J. MOON

(a)

Pervasive acid leaching

and silicification
Intermediate argillic
alteration
Orebody Adv. argillic alteration

Pa
Pa
Pa

Basement

Intrusion

FIG. 16.2 Diagrams showing the

(b) formation of two types of
epithermal precious metal deposit
in volcanic terranes. (a) Acid
sulfate type. Pa, propylitic
Fumaroles and alteration. Note that the
hot springs mineralisation occurs within the
heat source. (b) Adularia–sericite
Silica sinter
Orebody
type. The upwelling plume of
hydrothermal fluid is outlined by
Young volcanics the 200°C isotherm. The
and sediments mushroom-shaped top reflects
fluid flow in the plane of major
Basement fracture systems, a much narrower
thermal anomaly would be present
perpendicular to such structures.
The heat source responsible for the
300

buoyancy of the plume is shown as

an intrusion several kilometers
below the mineralised zone. In
both diagrams the arrows indicate
circulation of meteoric water. (a) is
1 km drawn to the same scale as (b).
Intrusion (Based on Heald et al. 1987 with
modiftcations from Henley 1991.)

contents. The deposits are associated with characteristic concentric zoning from inner
granitic intrusives that vary in composition potassic through phyllic to propylitic zones.
but with quartz monzonite and diorite as the Gold generally correlates with copper grade
most important hosts. Alteration shows a and is associated with K alteration (Sillitoe
 16: TRINITY SILVER MINE, NEVADA 385

1991). In some cases high sulftdation gold elements are As, Sb, Hg, Tl, and W for sedimen-
systems are mined at higher levels than the tary hosted deposits, with the addition of base
porphyry, although a direct connection is often metals for most deposits. Although pathftnder
hard to prove. For further discussion of por- elements can be extremely informative as to
phyry copper deposits see Evans (1993) and the nature of the gold anomaly, they should not
Robb (2004). be used without gold as pathftnder elements
may be displaced from gold. In addition some
gold deposits, particularly adularia–sericite-
16.1.2 Geochemistry
type deposits, have no pathftnder signature
One of the major features in the discovery of and the exploration geologist must rely largely
disseminated gold deposits has been the further on gold geochemistry. An example is the
developments in gold geochemistry. The recog- Ovacik deposit in Turkey (see section 4.2.1)
nition of “noseeum” gold deposits depends on that was discovered by the follow-up of BLEG
the ability to detect ftne-grained gold. As this geochemistry.
is usually in the micron size class it does not In Nevada large stream sediment samples
often form grains large enough to be panned. and extensive soil sampling have proved effec-
Much gold remains encased in ftne-grained tive where overburden is residual. However
silica and therefore sampling and analytical in some areas the overburden has been trans-
techniques must be devised that accurately ported and surface sampling is ineffective.
determine all the gold present. This analytical Chip sampling of apparently mineralized rocks,
problem has largely been overcome, ftrstly by particularly jasperoids, has been successful al-
the improvement of graphite furnace AAS and though jasperoidal anomalies are often slightly
more latterly by the application of neutron displaced from the associated deposits.
activation analysis and ICP–MS (see section There are a large number of case histories
8.2.3). Graphite furnace AAS methods have on gold exploration and the reader is advised to
relied on the separation of gold from the rock read Zeegers and Leduc (1991) and case his-
or soil matrix using an organic solvent. This tories on Nevada in Lovering and McCarthy
allows analysis with detection limits of lower (1977).
than 5 ppb.
As gold is present as discrete grains within
16.1.3 Geophysics
the sample either encased in the matrix or as
free gold, sampling methods have to be devised Disseminated gold deposits are difftcult geo-
that will allow the analytical aliquot to repres- physical targets. Electrical and EM methods
ent accurately the original sample. Investiga- have been used to map structure and identify
tions by Clifton et al. (1969), well summarized high grade veins, e.g. at Hishikari, Japan
by Nichol et al. (1989), have shown that it is (Johnson & Fujita 1985), but are not usually
necessary to have 20 gold grains in each sample effective in detecting mineralisation except
to achieve adequate precision. In real terms indirectly. On occasion unoxidized sulftdes are
this means taking stream sediment samples of present allowing the use of induced polariza-
around 10 kg and carefully subsampling. The tion (IP). One of the more successful uses of
geologist who takes a 1-kg sample can not ex- geophysics in Nevada has been the detection
pect to make an accurate interpretation of ana- of silicifted bedrock under Tertiary cover
lytical data. An alternative strategy, which has (Fig. 16.3). The section shows the results of a
been used where rocks have been oxidized, is to Controlled Source Audio Magneto Telluric
take a large sample and leach it with cyanide; Survey that clearly detected a zone of siliciftca-
this technique is known as Bulk Leach Extract- tion under 50 m of cover. Further case studies
able Gold (BLEG). This technique has found can be found in Paterson and Hallof (1991).
wide acceptance in Australia and the Paciftc
Rim countries.
16.1.4 Mining and metallurgy
Most but not all gold deposits have con-
centrations of pathftnder elements associated The recognition that disseminated precious
with the gold. In Nevada the most widely used metal deposits could be successfully bulk
386 M.K.G WHATELEY, T. BELL & C.J. MOON

8192 grade material has previously unsuspected high

4096
grade material with different metallurgical
2048
1024 characteristics beneath it. For example, at the
Frequency (Hz)

512 Post deposit on the Carlin trend high grade

256 material in which gold is difftcult to extract, or
128 refractory, underlies the main surface deposit.
64
32 The grade of this material is so much higher
16 that it is worthwhile investing in underground
8 mining and pressure leach extraction to re-
4
cover the gold.
2
1
W800 W400 0 E400 E800
Station location (ft) 16.2 TRINITY MINE, NEVADA
Surface
0
The Trinity Silver Mine was an open pit, heap
Carlin Tertiary Carlin Volcanics
50
Sediments leach, silver mining operation extracting
rhyolite-hosted, disseminated, hydrothermal,
Depth (m)

100

150
Russ Formation Queen Formation silver oxide mineralisation. The mine was
Silicification in the brought into production on September 3, 1987
200 Palaeozoic bedrock
formations, gold
Drillhole and mining ceased on August 29, 1988. The
250 values are present heap leach operation continued for just over
one more year. Our choice of a silver mine
rather than a gold producer has been governed
FIG. 16.3Section showing the application of a
CSAMT survey in detecting areas of siliftcation entirely by the availability of data. Most of
under Tertiary [Link]: contoured data; bottom: our discussion is of course applicable to any
drill section. (After Paterson & Halof 1991.) disseminated precious metal deposit. US
measures are used throughout this chapter as
the mining industry in the USA has yet to
mined and processed, sometimes with specta- metricate. Thus oz t1 are troy ounces per short
cular returns, has been responsible for the allo- ton of 2000 avoirdupois pounds (lb).
cation of the large exploration budgets spent in This case study focuses upon the statistical
Nevada. Oxidized deposits are generally mined assessment of blasthole assay data, for the
in open pits, usually with low stripping ratios. purpose of determining the distribution and
One of the keys to a successful operation is quality of mineralisation. It is essential for reli-
effective grade control (section 16.5) as differ- able ore reserve calculations to derive the best
ent grade ores are treated differently in most grade estimate for an orebody. The actual mined
operations. The higher grade material is usu- reserves and estimated reserves calculated
ally milled and extracted with cyanide in a con- by US Borax and Chemical Corporation (US
ventional manner whereas the lower grade ore Borax) from exploration borehole data are also
is crudely crushed and placed on leach pads. described. The Surpac Mining System Software
These leach pads are built on plastic liners, was used for data management and statistical
sprinkled with cyanide, using a system similar evaluation. This work was undertaken as an
to a garden sprinkler, and the cyanide allowed MSc dissertation at Leicester University (Bell
to percolate through the crushed ore. The 1989) under the auspices of US Borax.
cyanide is then pumped to a central treatment The Trinity Mine is in Pershing County,
plant where the gold is extracted by passing the Northwest Nevada (Fig. 16.4), on the north-
cyanide through activated carbon. Marginal ore west flanks of the Trinity Range, 25 km north-
and waste are stockpiled separately to allow for northwest of Lovelock. The area is covered by
the later treatment of marginal material if the the souhtwest part of the USGS Natchez Spring
gold price increases. 7.5 topographic map and Pershing County
One of the interesting developments in Geological Map. Elevations vary from 1200 m
Nevada is the recognition that some of the low to 2100 m (3900–7000 ft). The mineralisation
 16: TRINITY SILVER MINE, NEVADA 387

Black Rock STUDY AREA

Desert
N + + ●
Lovelock
+ + NEVADA
+
+ +
Las Vegas
●

+
+
+ +
+ Central
+ Seven
Range
+ Trough Range +
T30N + STUDY Humboldt
+ + Range
Selenite AREA +
+ Range Tobin
+ + Range
+ Buena Vista
+ Valley
+ Trinity
+ Willow Buffallo
+ Range Canyon Valley
+
Lovelock
+ + Granite Spring
Valley
+ +
+
+ + 0 20 km
T25N + +
+ +
R30E R40E

Tertiary sediments Triassic sediments

Recent Permian sediments Cambrian sediments
and volcanics (ALSG)

Quaternary Jurassic ++ Cretaceous Granodiorite Ordovician sediments Thrusts

FIG. 16.4 Sketch

map of Pershing County, Nevada showing the outline geology and location of the Trinity
silver deposit. ALSG, Auld Lang Syne Group. (After Ashleman 1988.)

at Trinity principally lies between 1615 and of US Borax, but they were unable to locate
1675 m (5300 and 5500 ft). previously reported gold anomalies. However
in 1982 a signiftcant silver show was found
within altered rhyolites. As a result, agree-
16.3 EXPLORATION ments were reached between ftve existing
claim holders, Southern Paciftc Land Company
(SPLC), holders of the surface rights and US
16.3.1 Previous work
Borax, after which the Seka claims were staked.
Mineralisation was ftrst discovered in the Trin- The Trinity joint venture between Sante Fe
ity Range by G. Lovelock in 1859 and limited, Paciftc Mines Inc., the land holder and a sub-
unrecorded production from Pb-Ag and Ag-Au- sidiary of SPLC, and Paciftc Coast Mines Inc.,
Cu-Zn veins occurred between 1864 and 1942 a subsidiary of US Borax, was operated by US
(Ashleman 1983). Evidence of minor prospect- Borax and mining was contracted to Lost
ing during the 1950s was found within Willow Dutchman Construction.
Canyon (Fig. 16.5) (Ashleman 1988). Geophys-
ical exploration and trenching during the
16.3.2 US Borax exploration
1960s was carried out by Phelps Dodge within
Triassic sediments to the north of Willow Various stages of exploration were completed
Canyon. Later a geochemical exploration pro- by US Borax during the period 1982–86. In 1982
gram aimed at locating gold mineralisation was and 1983 mapping, geochemical and geophys-
undertaken by Knox–Kaufman Inc. on behalf ical techniques were used to located payable
388
388 M.K
K.G WHATELEY,, TT.. BELL
G WHATELEY & C.J. MOON
BELL & MOON
Area> 1.3 oz t–1Ag (45 g t –1)

North - eastern
Lobe

South - western
Lobe Approximate
outline of
the open pit
(South- western
extension)

0 100 m

Quaternary Surficial Deposits Tertiary Epiclastic + Pyroclastic Tuff

Late Tertiary
Sugary Rhyolite Unconformity
Tertiary Rhyolite Porphyry Early Jurassic-
Auld Lang Syne Group
Triassic
Aphanitic Rhyolite Fault

FIG. 16.5 Sketch map of the geology in the vicinity of the Trinity silver deposit, Nevada. (After Ashleman
1988.)
 16: TRINITY SILVER MINE, NEVADA 389

sulftde mineralisation. In 1983 drilling com- contour maps of the results helped deftne the
menced within an area known to host sulftde extent of surface mineralisation.
mineralisation. Exploration continued from Between 1985 and 1986 rock geochemistry
1985 to 1986 in an attempt to ftnd additional was applied to the oxide zone in the southwest
reserves. Although assessment of the sulftde extension. Rock chips from drilling, trenching
ore continued with metallurgical testing and and surface exposure were analyzed for Pb, Zn,
feasibility studies, the possibility of a high Ag, and Au. High grade samples containing
grade oxidized zone to the southwest of the 30–50 ppm Ag were reported and localized
main target area became the focus of attention. samples containing 40–90 ppm Ag were re-
Efforts were thereafter concentrated upon the corded as the zone of oxidization was traced
evaluation of this southwestern extension southward along strike. Anomalies from the
(Fig. 16.5). rock and soil geochemical surveys were used in
the planning and layout of the drilling program
within the sulftde zone.
16.3.3 Mapping
Geophysical surveys, including gradient
Initial mapping of the claim area at 1:500 was array IP, magnetic and gamma ray spectro-
followed by detailed mapping of the sulftde metry, were carried out concurrently over the
zone at a scale of 1:100. Mapping of structure sulftde mineralisation. The gradient induced
and lithology helped deftne the surface extent polarization (IP) survey required three arrange-
of mineralisation and alteration, and the con- ments of transmitting electrodes with a pole
trols of the distribution of mineralisation. spacing of 12,000 ft (3650 m). Readings were
When the potential of the silver mineralisation taken every 500 ft (150 m) along lines spaced
within the oxidized zone was established in 1000 ft (300 m) apart (Ashleman 1983). A time
mid 1986, detailed mapping of the southwest- domain chargeability anomaly coincided with
ern extension commenced. Mapping showed the area of sulftde mineralisation. A NNW
the host rock, of both oxide and sulftde miner- trending belt of high resistivity was located
alisation, to be fractured rhyolite porphyry adjacent to the eastern margin of an area of
and deftned an area of low grade surface miner- known subsurface mineralisation. The mag-
alisation intruded by barren rhyolitic dykes netic survey proved to be of little value due to
(Fig. 16.5). These barren dykes divide the south- the poor contrast in magnetic values. Gamma
west extension into northeast and southwest ray spectrometry using K, Th, and U channels
lobes. The rhyolite dykes were seen to be was tested, but only subtle differences were
intruded along east and north-northeast faults seen between altered and unaltered rocks
and to disrupt the stratigraphical sequence and the suvey was discontinued (Ashleman
within the southwest lobe. 1983).

16.3.4 Geochemistry and geophysics 16.3.5 Drilling

A geochemical soil survey was completed over Successive stages of drilling on a 100 ft (33 m)
the oxide and sulftde zones and samples were grid orientated 11 degrees west of true north
analyzed for Pb, Zn, and Ag. The mobility of Ag (Fig. 16.6) totalled 29,880 m (98,031 ft) drilled
and Zn in the soil horizons placed constraints (Table 16.1). Most of the drilling was within
upon the signiftcance of anomalies reflected areas of principal sulftde mineralisation, focus-
by these elements. Lead is more stable and ing on the surface exposure of silver mineral-
was therefore used as a pathftnder element for isation, lead, and IP anomalies and areas of
Ag mineralisation. Signiftcant Pb anomalies favorable geology. Drilling deftned a body of
were located over the sulftde zone. Anomalies silver mineralisation dipping to the west.
of greater than 100 ppm deftned potential min- The drilling grid was extended to explore
eralized areas and higher levels (1000 ppm) for additional reserves and high grade silver
coincided with the surface intersection of oxide mineralisation was intercepted to the
Ag mineralisation. Surface rock samples were southwest (Fig. 16.5). In 1985 drilling was
taken as part of a reconnaissance survey, and concentrated within the southwest extension.
390 M.K.G WHATELEY, T. BELL & C.J. MOON

N 10400
Conventional percussion drilling

●
Reverse circulation drilling
Diamond drilling
N 10200

●
Outline of final pit

●
●

●
●
●

N 10000

●
●

●
●

●
●

●
●
●
●
●
●

●
●

●
N 9800

●
●
●

●
●

●
●
●

●
●

●
●
●

●
●

N 9600

●
●
●
●

●
●

●
●

●
●
●
●

●
●

N 9400
●

●
●
●

●
●

●
●

●
●
●

●
●

N 9200

●
●

N 9000 0 50 m

E 8000 E 8200 E 8400 E 8600 E 8800 E 9000 E 9200 E 9400 E 9600 E 9800

FIG. 16.6 Map

showing the location of the exploration boreholes in relation to the ftnal outline of the Trinity
mine open pit from which the oxide mineralisation was extracted.

TABLE 16.1 Summary of drilling exploration programs.

Drilling methods Number of holes Total meterage

Percussion 199 22,857 (74,990 ft)

Reverse circulation 39 6257 (20,530 ft)
Cored 10 765 (2511 ft)
Total 258 29,879 (98,031 ft)

Ninety-two of the holes were drilled to deftne percussion holes were sampled as 5 ft (1.5 m)
the oxide zone. Although mineralisation composites (Ashleman 1986). Samples were
was located south of the southwest extension sent to the laboratory and analyzed for Ag, Au,
along strike and at depth, it was noted to be Pb, Zn, and As. Analyses on the samples from
discontinous. the exploration holes were carried out using
Diamond drill holes were sampled every 1– atomic absorption spectrometry (AAS), which
3 ft (0.3–0.9 m) for metallurgical testing, while gave the total silver content. Samples recording
 16: TRINITY SILVER MINE, NEVADA 391

Ag in excess of 1000 ppm were reanalyzed a westerly prograding delta ( Johnson 1977).
using ftre assay. Tertiary basaltic and andestic volcanics pass
into a thick, laterally variable, silicic, volcano-
clastic sequence consisting of a complex pile
16.4 THE GEOLOGY OF PERSHING COUNTY of rhyolite domes, plugs, flows, and tuffs. The
AND THE TRINITY DISTRICT Tertiary rhyolitic volcanics are overlain and
interdigitate with lacustrine and shallow lake
sediments, later intruded by silicic dykes, sills,
16.4.1 Regional setting – Pershing County
and stocks. The Tertiary sequence is covered
The Trinity Mining District of Pershing by Quaternary to Recent gravel and alluvial
County is part of the Paciftc Rim Metallogenic deposits. The Miocene to Pliocene deposits
Belt. The mountain ranges of Pershing County form a thick stratifted, well-bedded sequence of
reflect the surface expression of major NE–SW tuffs, shales, sandstone, clay and gravel, locally
structural lineaments sequentially repeated interbedded with basaltic and andestic flows.
(basin and range) across Nevada. Much of the Various structural phases are apparent
present geological setting is the response to within Pershing County (Johnson 1977). Pre-
regional compression subsequently modifted Cenozoic large scale compressional folding and
by extension. The regional geology is outlined thrusting during the Sonoma Orogeny (mid
in Fig. 16.4. Jurassic) resulted in the south and southeast
Cambrian to Permian shallow marine movement of basinal sediments and volcanics
carbonates and clastics are exposed to the over shelf sediments. These folded, imbricate,
northeast and east of Pershing County. The thrust slices were later offset by dextral strike–
Harmony Formation (U. Cambrian) is com- slip faulting that has northwest arcuate and
posed of quartzite, sandstone, conglomerate, discontinuous trends (Stewart 1980). Such
and limestone which pass laterally into clastic thrusting is apparent to the east and southeast
and detrital sediments of the Valmy Forma- within the Central and Tobin Ranges (Fig. 16.4).
tion (Ordovician). Carboniferous to Permian Crustal extension during the Cenozoic resulted
sediments (limestone, sandstone, and chert) in northerly trending basin and range faulting.
and volcanics (andestic flows and pyroclastics) Associated with the basin and range faulting
form most of the Tobin Range to the east. are effusive phases of bimodal volcanism.
During the Mesozoic fault-bounded, deep Stewart (1980) records both pre- and post-basin
water, sedimentary basins dominated the area, and range N–E and E–W fault sets, and implies
with associated widespread volcanic and silicic possible strike–slip components to the normal
intrusive activity. faults.
Triassic cover is widespread in central Tertiary silver mineralisation is reviewed by
Pershing County. The lower units to the east Smith (1988b), who classiftes the silver-bearing
consist of interbedded andesites, rhyolites, and deposits into three categories; disseminated
sediments, locally intruded by leucogranites type, vein type, and carbonate hosted. The
and rhyolite porphyries. To the southeast the former two types are outlined and compared
Middle Trias is exposed as calcareous detrital with the Trinity deposit in Table 16.2. A fur-
and clastic sediments, limestone, and dolo- ther classiftcation into four subtypes (Bonham
mite. The upper unit which forms the major 1988) includes volcanic-hosted epithermal,
sedimentary cover within the Trinity, South silver-base metal limestone replacement de-
Humbolt, and Severn Troughs Range, deftnes posits, sedimentary-hosted stockworks, and
cyclic units of sandstone and mudstone with volcanic-hosted stockworks. Pershing County
limestone and dolomite. Adjacent to Cretace- is dominated by Pb-Zn-Ag-Au veins hosted
ous granodiorite intrusions, the Triassic to within and adjacent to Cretaceous granodi-
Jurassic sediments are locally metamorphosed orites, with minor occurrences in Triassic
to slates, phyllites, hornfels, and quartzite. sediments. Mineralisation tends to be con-
The Auld Lang Syne Group (ALSG) forms the centrated in shear zones and along hornfels
majority of the Triassic sedimentary sequence dykes forming pinch and swell structures along
and reflects the shallow marine conditions of oblique joint sets (Bonham 1988).
392 M.K.G WHATELEY, T. BELL & C.J. MOON

TABLE 16.2 Outline

of silver deposits of Nevada, including a comparison with the Trinity deposit. (Adapted
from Smith 1988.)

Disseminated Vein Trinity

Host rock Tertiary felsic volcanics and Tertiary andesites and Teritary porphyry rhyolite,
clastic sediments, Devonian rhyolites minor occurrence in tuff,
limestone and argillite
Type of Discrete grains and vein Banding in quartz veins Disseminations,
mineralisation stockworks microfractures, veinlets,
and breccia inftll (vein
stockwork)
Morphology Tabular, podiform bodies, Veins deftned by fault Lenticular and stacked
deftned by assay cut-off of contacts with assay cut-off tabular, sharp assay
300 g t1 between 200 and 300 g t1. boundaries of 30 g t1
Change in cut-off has little
effect on tonnage
Mean grade 270 g t1 500–800 g t1 160–250 g t1
Geological Structural, rock Structural and fracture Structural, rock
control permeability and proximity density, intersection of permeability, fault
to conduit fault and fractures intersections, percentage
fracture density
Zoning Increased mineral content Sharp vertical zoning, Sulftde-oxide zoning, high
with depth, high Pb–Zn– sulftdes increase with Ag:Au
Mn, low Au, Ag:Au100 depth, Ag:Au50
Mineralogy Acanthite, tetrahedrite, Acanthite, low pyrite and Sulftdes: friebergite–
native Ag, pyrite and Pb– sulftde content pyragyrite, chalcopyrite,
Zn base metals pyrite, stannite. Oxides
Alteration Silica–sericite (carbonate, Extensive and variable, Siliciftcation, adularia–
clay, chlorite, potassic hematite–magnetite, quartz–sericite, minor
feldspar) propylitic, adularia and propylitic, illite and
illite kaolinite
Process of Hypogene Hypogene, hydrofracturing, Epithermal and
formation and brecciation hydrotectonic fracturing
Hydrothermal 165–200°C, 5 wt% NaCl, 200–300°C, meteoric Not available
solutions low salinity groundwater solutions, precipitation
mixing with oilfteld brines induced by mixing, boiling
and wall rock alteration

16.4.2 The geology of the Trinity Mine (Ashleman 1988) have been subjected to both
low grade regional metamorphism and contact
Stratigraphy metamorphism. The latter is associated with
the intrusion of Cretaceous granodiorite dykes
The local stratigraphy of the Trinity Mine and and stocks located to the northeast of the mine
surrounding area is outlined in Fig. 16.5 and (Fig. 16.4). Outcrops of the ALSG to the south
Table 16.3. The ALSG is described by Johnson are dominated by phyllites and slates which
(1977) as a ftne-grained clastic shelf and basin grade into coarse-grained facies (siltstones,
facies with interbedded turbidites. The local sandstones, and quartzites) northwards. To the
silicic and calcareous units of the ALSG east the unit is represented by carbonaceous
 16: TRINITY SILVER MINE, NEVADA 393

TABLE 16.3 Stratigraphy of the Trinity area. (From Bell 1989.)

Age Unit name Lithologies Intrusions

Quaternary Surftcial deposits Gravel, alluvium, and colluvium

Pleistocene Volcanics, fluvial Basalt, fanglomerates and channel Latitic and rhyolitic
deposits, Upper Tuff sands, phreatic-clastic welded and dykes
unwelded, tuff
Tertiary Rhyolite porphyry Rhyolite flows, agglomerates, and Rhyolitic, exogenous
(early Pliocene) breccias domes, porphyritic to
Lower tuff aphanitic
Epiclastic and pyroclastic airfall,
reworked tuff
Cretaceous Argillite breccia Subangular to angular argillite clasts Granodiorite, medium-
(1–10 cm) in ftne-grained matrix. grained, hypidiomorphic
Breccia is gradational, tectonic in (90 Ma)
origin, and fault associated
Triassic–early Auld Lang Syne Group Shelf and basin facies. Shale,
Jurassic siltstone, slate, phyllites, and
argillites. Siliceous and carbonaceous
sandstones and limestones.
Quartzites

siltstones and impure limestones (Ashleman resulted in two phases of metamorphism of

1988). The transition between the Tertiary the Mesozoic sediments: (i) low grade regional
rhyolite volcanics and the ALSG is marked by and (ii) contact metamorphism. Pre-Tertiary
an argillite breccia. The breccia is ftne-grained folding and faulting of the ALSG resulted
and matrix supported, it contains semiangular in isoclinal folds (Nevadian Orogeny). Tertiary
to angular argillite clasts and is closely associ- deformation produced large scale, north-
ated with faulting. trending open folds and low to high angle
Tertiary volcanism resulted in a complex faulting.
sequence of rhyolitic flows superseded by epi- Mid Miocene tectonic activity marked the
clastic and pyroclastic deposits. The Tertiary onset of extensional tectonics, resulting in
rhyolites dominate the geology of the Trinity NNE and NE trending, high angle, basin, and
area, with extensive exposure to the south range faulting. This was preceded by N, NNW,
and flanking the Trinity Range to the north. and WNW fault sets and localized thrust
Latitic and “sugary” rhyolitic dykes disrupt the zones (Ashleman 1988). Faulting appears to
sequence, with effusive phases and breccias postdate the Tertiary rhyolites and predate
spatially associated with rhyolite domes. The alteration and mineralisation; however age
volcanics are capped by Pliocene to Pleistocene relationships may not be so easily deftned.
fluvial sediments and locally by basaltic flows. Fault displacement and reactivation in con-
Quaternary alluvium inftlls valleys and flanks junction with localized NE thrusts and oblique
the Trinity Range forming channel ftll and allu- ENE shear zones (Johnson 1977, Stewart
vial fan deposits. 1980) confuses the structural history of the
area. The structural history of the Trinity
area appears to be one of initial NE trending
Structural geology block faults and associated thrusts and shear
Polyphase deformation involving compres- zones, later offset by NW trending normal
sional tectonics and granodiorite intrusions faults.
394 M.K.G WHATELEY, T. BELL & C.J. MOON

Alteration tion is most extensive within the rhyolitic

porphyry and tuffs. Siliciftcation is restricted to
The ALSG was not generally receptive to
discrete W dipping lenticular zones within the
Tertiary hydrothermal alteration, except loc-
upper zone of the deposit, with minor exten-
ally along faults and breccia zones. Tertiary
sions along dykes and veins. Siliciftcation is
volcanics were receptive to hydrothermal
concentrated within the NE lobe extending
fluids which resulted in varying degrees of
from and parallel to NE and ENE faults. Patchy
alteration and hydrofracturing. Rhyolitic tuffs
Fe staining and limonite is spatially associated
and porphyritic flows have been extensively
with high grade Ag mineralisation.
altered, deftning an alteration halo that extends
2.5 km beyond the main mineralized zone
(Fig. 16.7). Sericitization, siliciftcation and Mineralisation
quartz–adularia–sericite (QAS) alteration tends
to be most intense along faults, within per- Silver mineralisation in Pershing County is
meable lithologies and breccia zones. Minor reported to be hosted within breccias periph-
propylitic alteration and kaolinite–illite clays eral to rhyolite domes with mineralisation
are also reported (Ashleman 1988). QAS altera- concentrated in microfractures ( Johnson 1977).

FIG. 16.7 Aerial

photograph of Trinity Mine. The light areas in the center represent waste and low grade ore
stockpiles. The medium tone areas in the foreground show the extent of the alteration in the area. The
rectangular area is the heap on which the ore is placed for cyanide leaching. The ponds in which the pregnant
solutions collect are seen next to the mine buildings. (Photograph supplied by US Borax.)
 16: TRINITY SILVER MINE, NEVADA 395

However, the Trinity silver orebody is described deftned by sharp but irregular boundaries, and
as a hydrothermal, volcanic hosted, silver-base dips steeply (55–70 degrees) to the W and NW.
metal (Cu-Pb-Zn-As) deposit (Ashleman 1988). The SW lobe is disrupted by postmineralisation
Bonham (1988) describes it as occurring in faulting and dykes. The oxide orebody deftnes a
stockwork breccias formed in and adjacent to linear feature 75  610 m (250  2000 ft) ex-
a rhyolite dome of Miocene age. Both sulftde tending to depths of 90–135 m (300–450 ft)
and oxide ore is present, but high grade zones forming lenticular to stacked tabular bodies.
amenable to cyanide leaching are restricted Mineralisation appears to be conftned to the
to the oxide zone. The principal sulftde phases northwest side of a NE trending fault zone with
identifted in pan concentrates from rotary the southern boundary deftned by a steep, pos-
drill cuttings consist of pyrite–marcasite– sibly listric, normal fault which brought tuff
arsenopyrite–sphalerite–galena with lesser into contact with the rhyolitic porphyry. NW
amounts associated with chalcopyrite– trending offshoots cross the footwall contact
pyrrhotite–stannite. They were deposited with into the rhyolitic tuffs and parallel NW faults.
silica along stockwork fractures. The silver- The hanging wall, or northern contact, is not
bearing sulftde phases consist of freibergite marked by any signiftcant lithological change
(silver-rich tetrahedrite) and pyrargyrite. other than a brecciated form of the rhyolitic
Acanthite and native silver are also present porphyry at a high stratigraphical level. This
in trace amounts. Silver in this zone is ftne- boundary has been inferred to reflect a change
grained (0.01–0.5 mm), locked within sulftde in host rock permeability or a high angle re-
phases as inclusions and intergrowths, im- verse fault consistent with basin and range
posing signiftcant metallurgical problems faulting.
(section 16.6). NE trending basin and range faulting offset
Only the southwest oxide zone, on which by a cogenetic or later, minor, NW fault system
this chapter concentrates, has been exploited. probably acted as feeders for the invasion of
The Ag mineralogy is essentially bromine-rich early Pliocene hydrothermal fluids. Hydro-
cerargyrite (Ag(Cl,Br)) with minor amounts fracturing and brittle failure resulted in the
of acanthite, often closely associated with formation of a conjugate fracture zone. Hydro-
limonite and Fe staining. There are some thermal fluids exploited the fracture zone and,
chloride and sulftde minerals present too. in conjunction with alteration, made the host
Payable mineralisation is mainly restricted to rock amenable to mineralisation. Mineralisa-
porphyritic rhyolite flow units (75% of total) tion is concentrated at fault intersections
as disseminations, fracture inftlls, veinlets, and (northwest offshoots) and in a zone parallel but
replacements of potassic feldspars. Weak displaced from the fault planes. The sharp
mineralisation occurs within rhyolitic tuffs, footwall contact is either the function of min-
breccia, and argillites (25% of total) associated eralisation being dissipated by a highly porous
with QAS alteration. High grade zones tuffaceous rhyolite or ponded by an imperme-
(10 oz t1 (343 g t1) ) exhibit a spatial rela- able fault gouge. Alternatively, both the sharp
tionship with areas of strong jointing, sericitic north and south boundaries may reflect the
alteration, and limonite (Ashleman 1988). effects of fault reactivation, i.e. dextral strike
Base metals (Pb-Zn-Cu-As) exhibit a spatial slip displacing mineralisation along strike and
association with the distribution of silver min- producing further uplift and erosion. This how-
eralisation, i.e. a gradual increase to the south. ever is complicated by NW–SE offshoots that
Lead shows the strongest association with show no displacement.
silver in the oxide zone while zinc values tend
to be low.
A sharp redox boundary within the oxide 16.5 DEVELOPMENT AND MINING
zone marks the transition from upper oxide ore
into lower sulftde ore. The oxide orebody is The exploration drilling program deftned the
subdivided into a continuous NE lobe and ore zone on which the block model for the re-
fragmented SW lobe, the former containing the serve estimation and the open pit mine plan
highest grades. The NE lobe trends NE to ENE were based. Drilling continued during mining
396 M.K.G WHATELEY, T. BELL & C.J. MOON

FIG. 16.8 Photographshowing the

color-coded flags used to delineate
ore from waste in the pit. The dark
flags denote ore blocks and the pale
flags denote waste blocks.

to conftrm ore reserves. Pit modiftcation was (CN) leach method, giving a more represent-
required during mining to improve ore control ative extractable silver content. Statistical
(section 16.8.4), slope stability, ore shoot ex- and ftnancial analyses had shown that the
cursion, and operational access. The ftnal pit silver grades could be subdivided into four
dimensions were 1400  500  246 ft (427  152 groups: (i) ore at 1.3 oz t1 (45 g t1); (ii) low
 75 m) (Fig. 16.6), with 15-ft (4.5-m) benches grade ore at 0.9–1.3 oz t1 (31–45 g t1); (iii) lean
attaining slopes of 60–72 degrees. Pit slopes ore at 0.5–0.9 oz t1 (17–31 g t1); and (iv) waste
were 52 degrees to the west and 43 degrees to at 0.5 oz t1 (17 g t1). All conversions of
the east, and safety benches were cut every ounces per short ton to grammes per tonne
45 ft (14 m). Production haulage ramps were have been made using a factor of 34.285
55 ft (16.5 m) wide with 4 ft (1.2 m) berms and (Berkman 1989). Each bench was divided into
had a maximum gradient of 10%. blocks, each block centered on a blasthole. The
Standard mining methods of drilling and assayed grade for each blasthole was assigned
blasting, front end loading, and haulage by to the block. The grades were color coded and
truck were carried out by the contractor. A total each block was marked on the bench in the
of 16,500 t of ore and waste were removed daily pit by the surveyors with the appropriate color-
in a 10-hour shift, 5 days a week. Drilling for coded flags (Fig. 16.8). This ensured that waste
blasting, using a 55/8 in (143 mm) down-the- was correctly identifted and sent to the waste
hole hammer, was carried out at 15 ft (4.5 m) pile, while the different grade materials were
centers, and at 12 ft (3.7 m) centers in areas of sent either to the cyanide leach heap or to one
blocky siliciftcation. of the two low grade stockpiles.
Grade control was based upon the blasthole Sulftde ore was not amenable to heap leach-
samples and pit geology. The orebody was ing and where encountered was selectively
mined principally as a vein type deposit (Perry mined and hauled to low and high grade sulftde
1989). The eastern boundary (hanging wall ore stockpiles (Fig. 16.8) (Perry 1989). Sulftde
contact) was taken to be a steeply dipping ore was differentiated from oxide ore on the
range front fault and the western boundary was basis of color (yellow oxide ore and grey sulftde
an assay boundary. Blastholes were sampled, ore) and a percentage comparison of total AAS
by sample pan cut or manual cone cuts, each assays (or ftre assay) and cyanide leachable
representing a 15 ft (4.5 m) composite. In assays i.e.:
contrast to the AAS analytical method used
on the exploration borehole samples, the AAS  CN  100
blasthole samples were analyzed by a cyanide AAS
 16: TRINITY SILVER MINE, NEVADA 397

a percentage difference of 6–17% deftned the After initial ftltering the precipitate was dried,
oxide ore and the sulftde ore was marked by a analyzed, fluxed, and smelted (Fig. 16.9), to
24–40% difference. produce silver doré bullion (99% silver). It was
estimated that 75% of the silver in the oxide
ore was recovered by cyanide leaching.
16.6 MINERAL PROCESSING

Metallurgical testing of core samples deter- 16.7 ENVIRONMENTAL CONSIDERATIONS

mined that cyanide leaching was the most cost-
effective method of treating the oxide ore. An Any mining project which will result in the
excellent overview of heap leach technology is alteration of the surface of a property, either
given by Dorey et al. (1988). Sulftde ore resulted through the addition of mine buildings or
in high cyanide consumption and required ftne because of mining, is required to submit a
grinding to give a 78–84% recovery. The oxide Planning Application (in the UK) or a Plan of
ore was amenable to direct leaching resulting Operation (in the USA) to the authority respon-
in a 94–97% recovery. Flotation tests on the sible for administering the area. This applica-
sulftde ore liberated 90–95% of the silver and tion should describe the proposed mining
90% of the Pb and Zn providing the pH values operation and the infrastructure requirements
of the collectors, which suppressed Fe and As, and include a schedule of the proposed activ-
were high. Oxide ore recoveries by flotation ities (Thatcher et al. 1988). Often the operator
were low (50–60%). Laboratory tests indicated will include a plan of the procedures he pro-
that it would be possible to heap leach the ox- poses to implement to protect the environment
ide ore, but the sulftde ore had to be stockpiled during and after mining (reclamation).
with a view to possible flotation extraction at a In the case of a precious metal heap leach
later date. A flow sheet of the process is shown operation, additional permits and approvals are
in Fig. 16.9. required before a heap leach operation can be
Mined ore was hauled and dumped on to an commissioned. Thatcher et al. (1988) discuss
ore surge pile and fed in to a primary jaw the various regulatory aspects and permitting
crusher (Fig. 16.9). The 3/4 in (19 mm) prod- requirements for precious metal heap leach
uct passed over a two-deck vibrating screen on operations. These requirements include air
to an in-line agglomerator. Secondary crushing quality, and surface and groundwater quality
of the oversized ore was completed by cone permits as well as cyanide neutralization re-
crushers in parallel circuit. The crushed ore quirements. In the latter case, Nevada speciftes
was automatically sampled and weighed. detailed heap rinsing procedures and analytical
Ten pounds (4.5 kg) of cement was added to methods for cyanide detection as operating
1 t of ore prior to agglomeration. The agglomer- conditions in their water quality or discharge
ated ore was screened to remove the 3/8 in permits (Thatcher et al. 1988). In Nevada opera-
(10 mm) ftnes, and was then spread on the tors are expected to obtain cyanide neutraliza-
leach pad with a slough stacker. The tion levels of 0.2 mg per litre free CN  as a
700  1100 ft (210  335 m) leach pad was lined target concentration, but most operators are
with 60-mm heavy duty polyurethane (HDPE) unable to meet these standards. Similarly,
and divided into six cells. Each cell had a 183 t there should be only 1 ppm cyanide in the re-
capacity when stacked to 41 ft (12.5 m). Pri- sidual weak acid digest (WAD) or 10 ppm CN
mary and secondary leaching utilized drip feed in the soil. More realistically, they expect the
emitters providing a flow rate of 0.005–0.008 company to attempt to meet these criteria and
gallons per minute per square foot (200– then show that the cyanide remaining behind
350 ml min1 m2). Sprinklers were used during will not affect or put at risk the groundwater.
the ftnal stage of leaching and rinsing. The Trinity operators had several altern-
Pregnant solutions passed into the pregnant atives to consider with regard to cyanide
pond and were then processed by a Merrill– degradation and heap detoxiftcation. These are
Crowe Zn precipitator plant with a described in some detail by Smith (1988a),
1000 gallons min1 (3785 L min1) capacity. and the various methods are outlined here. The
398
[Link]
KK.G WHATELEY,, TT.. BELL
G WHATELEY
DRILLING & C.J. MOON
BELL & MOON
ORE CONTROL
Sampled and analyzed Broken ore
defined and flagged

SELECTIVE MINING

SURGE PILE STOCK PILE

Ag oxide ore Waste
sulfide ore
Ag bearing waste

PRIMARY CRUSHER 2 DECK VIBRATING +19 mm FEED

SCREEN

19 mm FEED 2 CONE SECONDARY

CRUSHER

AUTOMATIC SAMPLING
AND WEIGHING
Cyanide solution
emitters and wobblers

AGGLOMERATION LEACH PAD

2.4  8.5 m belt lined 22 mm ore pad cover on
agglomerator SLOUGH STACKER 60 mm HDPE liner
45 kg cement ton1 ore 6 cell pad 210  335  12.5 m
@ 10% moisture 200,000 t capacity

PRIMARY LEACH SECONDARY LEACH

Barren solutions

PREGNANT POND INTERMEDIATE POND BARREN POND

Cyanide Fresh water

CLARIFIERS DEAERATOR TOWER FILTER PRESSES

Flux
Zinc dust

DRYING MERCURY Filtered silver

RETORT precipitate

High grade slag

recycled

Waste slag to
FURNACES
heap

Doré to refinery

FIG. 16.9 A diagrammatic flow sheet of the mining and processing of the ore at the Trinity Mine, Nevada.
 16: TRINITY SILVER MINE, NEVADA 399

methods fall into two categories, namely nat- the pond so that the recirculated water is free
ural processes or chemical treatment. from cyanide. Both the last two methods are
quick acting, especially with the initial flush.
Thereafter there is a longer term diffusion
16.7.1 Natural degradation and
mechanism which completes the flushing
detoxification
procedure.
1 Passive abandonment. High UV levels and
strong winds for 9 months of the year would
16.7.3 Summary
make this a possible alternative. The natural
degradation of the cyanide would be fast in the There are advantages and disadvantages with
near-surface layers, but would slow rapidly each process as outlined above. The State of
with depth. Nevada discourages the use of hypochlorite
2 Rinsing with barren solutions. Rinsing to and peroxide neutralization methods (Thatcher
clean the entire heap has been tested empir- et al. 1988), but the INCO SO 2–air and the per-
ically in the laboratory. A column in the labor- oxide processes are thought to be more techni-
atory was rinsed with fresh water and the cally efftcient (Smith 1988a). However the ftnal
cyanide levels were reduced from 700 to conclusion may well be that each of the major
50 ppm. Unfortunately this took between 12 detoxiftcation processes in use has its applica-
and 21 months to achieve and extremely large tion in a particular set of circumstances which
quatities of water (of the order of 4500 L t 1) each operator must establish for each property.
were required.

16.8 GRADE ESTIMATION

16.7.2 Chemical treatment
Five types of chemical treatment are currently It is essential for reliable ore reserve calcula-
available and were considered by US Borax. tions to derive the best Ag grade estimate for an
1 Acid leach. The leachate thus produced orebody. The 5220 bench of the Trinity silver
would be collected in the ponds, neutralized deposit was chosen to illustrate the pro-
with lime, and the metals which precipitated blems and procedures of grade estimation. The
would be removed mechanically. 5220 level exhaustive dataset contains 1390
2 Sulfur dioxide and air oxidation method. blasthole samples. No spatial bias or clustering
(Patented by INCO). These are added in con- effects are evident due to the regular sample
junction with lime and a copper catalyst, but pattern (15 ft (4.5 m) centers).
special equipment is required and this method For ore reserve calculations a global estimate
is expensive. of mean silver is required. The global estimate
3 Alkaline chlorination. Either chlorine gas or must, however, be controlled by local esti-
calcium hypochlorite are used in the oxidation mates because of the spatial variation of silver
process but these methods have very high values. Measures of variability are important
reagent costs, and careful pH control with the in evaluating the accuracy of a grade estimate,
second method is required to prevent the for- especially within a spatial context, and have
mation of toxic cyanogen chloride. strong implications for mine planning.
4 Hydrogen peroxide oxidation. This method In many geological environments the prob-
is clean and nontoxic, but the H 2O2 is expen- lem of grade estimation is intrinsically linked
sive. This method is used by OK Tedi in PNG to the type of statistical distribution. Many
where H2O2 is added to the recirculation pond problems can be solved from an assessment of
and the cyanide-free water is then recirculated the univariate statistics, histograms, and prob-
through the heap using the original sprinkler ability plots (Davis 1986), but this assessment
system. can not be divorced from the spatial context
5 Ferrous sulfate. Chemical destruction of of the data (Isaaks & Srivastava 1989). The
the cyanide is used whereby 1 mole of ferrous following sections explore the problems of
sulfate ties up 6 moles of cyanide to produce grade estimation and derive simple solutions
Prussian Blue. This reaction takes place in for modeling skewed distributions.
400 M.K.G WHATELEY, T. BELL & C.J. MOON

16.8.1 Univariate statistics ftt a log-normal distribution. Although it is

possible to model a “best ftt” distribution to
For the purpose of grade estimation the dis-
the overall population, errors will occur during
tribution of assay values within the population
grade estimation. This is evident from the sum-
is a prime requirement, especially regarding
mary statistics (Table 16.4). It is clear from the
frequencies above a specifted lower limit, e.g.
frequency distributions that a global estimate
cut-off grades. Most statistical parameters are
based upon a normal distribution will be biased
applied to the Gaussian distribution, yet
toward high values (overestimated). Similarly
within a geological environment a normal dis-
the log transformed data will be biased by low
tribution is not always immediately evident.
values (underestimated) (Fig. 16.10d). Due to
It is common to ftnd a large number of small
the mathematical complications that arise
values and a few large values, the lognormal
from a log transformation, such a transforma-
distribution is then a good alternative (Isaaks &
tion is best avoided. It is a preferred alternative
Srivastava 1989). The histogram (Fig. 16.10a)
to try to establish Gaussian distributions
describes the character of the distribution. The
within the exhaustive dataset in order to im-
clustering of points at low values and the tail
prove grade estimation.
extending toward the high values indicates
The important features of the distribution
that the assays on the 5220 bench do not
are captured by the univariate statistics (Table
conform to a normal distribution. Most of the
16.4) and illustrate immediately the problem of
cumulative frequencies for the lower values
grade estimation for the 5220 bench. The sum-
plot in a relatively straight line, but departures
mary statistics provide measures of location,
occur toward the higher values (Fig. 16.10b).
spread, and shape. The mean Ag content of the
This is further evident from the curved nature
bench is recorded as 1.43 oz t 1 (49 g t1), yet
of the normal probability plot (Fig. 16.10c).
other estimates of central tendency, the
The departure from a straight line at lower
median (0.33) and the trimmed mean (0.48), in-
values on the log normal probability plot
voke caution in using the mean as an accurate
(Fig 16.10d) indicates that the data also do not
global estimate. The high variance and stand-
ard deviation describe the strong variability of
(a) (b)
the data values. The implication is that insufft-
40 100
cient conftdence can be placed upon the mean
Cumulative frequency %

30
80 as an accurate global estimate. A strong posit-
Frequency %

60
ive skew with a long tail of high values to the
20 right is evident from the high positive skew-
40 ness (10.62). The degree of asymmetry is also
10
20 supported by the coefftcient of variance (2.58).
Knudsen (1988) implies that if the coefftcient
0
0 4 8 12 16 0 4 8 12 16 of variance is greater than 1.2 a log-normal
distribution could be modeled. The Sichel-T
(c) (d)
estimator can be used to estimate the mean of
a log normal distribution, and is calculated to
Cumulative frequency %

Cumulative frequency %

99 99
90 90 be 1.28.
The frequency histogram (Fig. 16.10a) illus-
50
50
trates how the data are proportioned. It is
10 10 important to note that even though the data
1 1 range is 0–85 oz t1 Ag, only 5% is  7.2, 2%
 12.5, and 0.5%  20 oz t1 Ag. The upper quar-
0 4 8 12 16 0.1 1 10 100 tile indicates that 75% of the data lie between a
FIG. 16.10 (a)
Histogram of (cyanide leach) silver restricted range of 0–1.1 oz t1 Ag, and the mean
values (oz t1) obtained from blasthole samples which is higher (1.43) does not reflect the
on the 5220 bench, Trinity Mine, Nevada. (b) majority of the data. It is obvious therefore that
Cumulative frequency plot. (c) Normal probability the magnitude of the positive skew will have a
plot. (d) Log-normal probability plot. disproportionate effect upon the mean.
402 M.K.G WHATELEY, T. BELL & C.J. MOON

TABLE 16.4 Univariate statistics

of the silver values derived from Total Excluding 1.3–20
cyanide leach analysis of the blast population 20 oz t1 Ag oz t1 Ag
hole samples on the 5220 level,
Trinity Mine, Nevada. Number of assays 1390 1385 309
Mean 1.43 (49) 1.30 (45) 4.65 (159)
Median 0.33 (11) 0.33 (11) 2.30 ( 79)
Trimean 0.48 (16) 0.48 (16) 3.45 (118)
Sichel-T 1.28 1.22 4.52
Variance 13.59 6.93 16.36
Standard deviation 3.69 2.63 4.04
Standard error 0.10 0.07 0.36
Skewness 10.62 3.89 1.84
Coefftcient of variance 2.58 2.02 0.87
Lower quartile 0.16 ( 5) 0.16 ( 5) 1.90 ( 65)
Upper quartile 1.11 (38) 1.09 (37) 5.84 (200)
Interquartile range 0.95 (33) 0.93 (32) 3.93 (135)

Silver values are in oz t1 (g t1).

The initial statistical evaluation illustrates 20

a potential error in the global estimate result-
ing in possible overestimation and bias toward
erratic high values. More detail is required
given that the mean is only a robust estimate
if applied to a Gaussian distribution and a log
Frequency %

transformation is not adopted. It is important

at this stage to investigate the distribution 10
further. Having ascertained that the data do
not conform to a normal or log-normal dis-
tribution, the next step is to evaluate potential
modality and departures from a continuous
distribution.
The probability plot and histogram are very
useful in checking for multiple populations. 0
Although breaks in the graphs do not always 0 1 2 3 4
oz
imply multiple populations, they represent
changes in the character of the distribution FIG. 16.11 Histogramof (cyanide leach) silver values
over different class intervals. The initial step is (oz t1) obtained from blasthole samples on the 5220
to explode the lower end of the histogram by bench, Trinity Mine, Nevada, with class intervals
making the class interval smaller (Fig. 16.11) reduced to 0.05 oz t1. The higher values (5 oz t1)
and searching for potential breaks in the are omitted for clarity.
population. Breaks in the distribution are
chosen on the basis of signiftcant changes in
the frequency between class limits or repeated approximate normal distributions could
patterns reflecting polymodality. If subpopula- improve the global estimate.
tions are found, explanations must be sought Such an evaluation reveals a number of
and may depend upon sample support, geolo- breaks within the population of silver grade
gical control, or population mixing. The pos- on the 5220 level. The following class breaks
itively skewed distribution of the data may be have been interpreted: 0.4, 0.8, 1.3, 5, 10, and
a function of the overprinting of multiple 20 oz t1 Ag (14, 27, 45, 171, 343, and 686 g t 1).
populations. Deftning subgroups which may Those values greater than 20 oz t1 are
 16: TRINITY SILVER MINE, NEVADA 401

TABLE 16.5 Univariate statistics of the subpopulations of the silver values derived from cyanide leach analysis
of the blasthole samples on the 5220 level, Trinity Mine, Nevada.

Silver values in oz t 1 (g t 1)

0.0–0.4 0.4–0.8 0.8–1.3 1.3–5 5–10 10–20

Number of assays 762 212 1032 215 731

Mean 0.18 (6) 0.57 (20) 1.03 (35) 2.56 (88) 7.46 (256) 14.37 (493)
Median 0.17 (6) 0.55 (19) 1.01 (35) 2.25 (77) 7.34 (252) 13.60 (466)
Trimean 0.17 (6) 0.56 (19) 1.02 (35) 2.35 (81) 7.44 (255) 13.93 (478)
Sichel-T 0.20 0.57 1.03 2.56 7.46 14.37
Variance 0.01 0.01 0.02 1.04 2.48 10.79
Standard deviation 0.10 0.11 0.15 1.02 1.57 3.28
Standard error 0.00 0.01 0.02 0.07 0.21 0.58
Skewness 0.19 0.28 0.19 0.74 0.08 0.25
Coefftcient of variance 0.55 0.19 0.15 0.40 0.21 0.23
Lower quartile 0.10 (3) 0.47 (16) 0.89 (31) 1.71 (59) 5.98 (205) 11.18 (383)
Upper quartile 0.26 (9) 0.67 (23) 1.16 (40) 3.20 (110) 9.09 (312) 16.70 (573)
Interquartile range 0.16 (6) 0.20 (7) 0.27 (9) 1.49 (51) 3.11 (107) 5.52 (189)

considered outliers (5 out of 1390 samples) and univariate statistics and grade estimate (Table
are not evaluated. To assess the statistical sig- 16.4) despite only representing 0.5% of the
niftcance of the subgroups univariate statistics data. The mean is reduced by 10% to 1.3 oz t 1
are calculated for each group (Table 16.5). In (45 g t1) Ag while the variance and skewness
all instances normal distributions are appro- are reduced by up to half. However the distribu-
ximated with signiftcant reductions in the tion is still strongly skewed and the problems
variance, coefftcient of variance, and skewness. of estimation still remain. If the 1.3 oz t 1 cut-
Estimates of central tendency lie within statis- off is then also applied the distribution be-
tically acceptable limits indicating improved comes flatter, the global estimate more precise,
conftdence in the mean of each group. The 1.3– but accuracy is reduced because of increased
5 oz t1 group is slightly skewed and a further variability (Table 16.4). The 1.3 oz t 1 cut-off
split possible. However, improved detail may also reflects a major break in the population
not result in enhanced accuracy or conftdence and improved estimates may be a function of a
in the mean. separate population approximating to a normal
The exhaustive dataset can be split into sub- distribution. With a reduced skewness the vari-
groups according to grade classes and the grade ability of values become more symmetrical
estimate for each group is improved. The statis- about the mean.
tical and ftnancial evaluation undertaken by A further investigation of outlier removal
US Borax resulted in a similar grouping (section (Fig. 16.12) shows that as the tail effect of the
16.5). US Borax applied a 1.3 oz t1 cut-off to the distribution is reduced the mean grade and
data such that everything above 1.3 oz t 1 was standard deviation decreases in a linear fashion.
mined as ore. It is important to evaluate the The decline is more gradual for the total popu-
effect of cut-off on the global estimate and the lation than when the 1.3 g t1 cut-off is applied.
implications in relation to the subgroups. It is obvious that as values are included or
excluded the magnitude of the mean varies.
16.8.2 Outlier and cut-off grade evaluation This must be viewed in relation to the number
of samples, the effect of which is evident with
Those values in excess of 20 oz t1 (686 g t1) Ag values above or below 20 oz t 1 Ag. Although
have been classed as anomalously high values, the magnitude of the grade estimate can be
and if rejected have a signiftcant effect upon the increased by including outliers the rate of
 16: TRINITY SILVER MINE, NEVADA 403

10
Mean Ag (>1.3 oz t1 / 45 g t1)
7 Mean Ag
SD (>1.3 oz t1 / 45 g t1) 9
SD
6 Mean (Total) 8
SD (Total)
5 7

Ag oz t  1
Ag oz t 1

6
4
5
3
4
2 3

1 2
(a)
1
0
0 100 200 9
Outlier cut-off grade (Ag oz t1) Mean Ag
8 SD
FIG. 16.12Graph showing changes in mean grade and 7
standard deviation when different outlying (high)
grades are removed from the database of the silver Ag oz t  1 6
values from the 5220 bench, Trinity Mine, Nevada.
5
4
improvement in the ore reserve estimation is 3
reduced. A grade estimate must also evaluate
2
how the values fluctuate around the mean, (b)
therefore the relationship of the standard 1
deviation to the mean is important. 0 0.5 1 1.5 2 2.5 3 3.5 4
Within a mining environment the con- Cut-off grade (Ag oz t  1)
ftdence of the grade estimate is derived from FIG. 16.13 Graphsshowing the average silver grade
the standard deviation. The value of the stand- when varying cut-offs are applied to the database
ard deviation, however, is subject to the group of the silver values for the 5220 bench, Trinity
of values from which the estimate was Mine, Nevada: (a) total population; (b) with values
obtained and the spatial context. The standard 20 oz t1 removed.
deviation expresses conftdence in symmetrical
intervals, therefore when the standard devia-
tion exceeds the mean the magnitude of under- expands, variability increases. The grade cut-
estimates is signiftcantly different from the off curves reinforce the relationship of the
magnitude of overestimates. This is illustrated mean to the standard deviation for the total
in Fig. 16.12 for the total population such that population (Fig. 16.13a) and the effect of remov-
the global estimate would exceed the local ing the 20 oz t1 outliers (Fig. 16.13b). As
estimate in the majority of cases. The degree extreme values of the distribution are altered
of variability and error increase in magnitude the degree of variability declines. A point is
as the tail effect of the distribution is more achieved where the number of local underesti-
prominent, therefore the above effect becomes mates and overestimates as compared with the
more pronounced. global estimate balance each other. As low or
The relationship of the standard deviation to high values are included or removed the mean
the mean changes when the same evaluation is changes accordingly and when an approximate
applied to ore grade material. The graph (Fig. normal distribution is modeled the variability
16.12) indicates that the estimate for the ore reaches a constant (Fig. 16.13b). At this point
grade material is more reliable and subject to there is no improvement in the conftdence of
less fluctuation, although as the distribution the grade estimate and, in terms of reserve
404 M.K.G WHATELEY, T. BELL & C.J. MOON

1500
Pit outline

1400

1.8
1300
1.3

1200

0.4

0.8
1100

0 250 500 750 1000

contour map of the silver grades (oz t1) on the 5220 bench, Trinity Mine. Contours are coded
FIG. 16.14 A
according to grade classes.

estimation, it is the spatial distribution of estimates can be applied to the contoured

silver that becomes important. zones. A further assessment of local estimates
can be achieved by using moving windows
(Isaaks & Srivastava 1989, Hatton 1994a).
16.8.3 Spatial distribution
Statistics were calculated for 100 ft (30 m)
Although it is possible to improve grade esti- square windows overlapping by 50 ft (15 m)
mates by evaluating the statistical distribu- (Fig. 16.16). The contour plots of the mean and
tion, the spatial features of the dataset are standard deviation for each window (Fig. 16.17)
important. A spatial evaluation will focus upon show that the average silver values and vari-
the location of high values, zoning, trends, and ability change locally across the area. Zones
continuity. A contour map of the 5220 level of erratic ore grades can be located and flagged
(Fig. 16.14) shows an E–W trend to the data for the purpose of mine planning and grade
which fragments and changes to a NW–SE control.
trend to the west. Zoning occurs and discrete In general the change in variability reflects
pods of ore grade material are evident. Low the change in the mean, although at the edge of
grade values less than 0.8 oz t1 (27 g t1) cover the high grade zone to the east, the variability
a greater area. The closeness of the contours increases at a greater rate. High variability is
especially at the 1.3 oz t 1 Ag (45 g t1) contour a function of the mixing of two populations or
indicates a steep gradient from high to low val- the transition from one population or zone to
ues. The abrupt change from waste to ore grade another. This is seen in the proftle taken along
material marked by a steep gradient will be N1300 (Fig. 16.18a). A proftle passing through
reflected by a high local variance. the high grade zone (N1200, Fig. 16.18b) shows
It is obvious that the subgroups previously a reduced change in variability compared with
deftned are not randomly distributed but con- a pronounced increase in the magnitude of the
form to a zonal arrangement. The indicator mean. The strong relationship between the
maps (Fig. 16.15) support the spatial integrity local mean and variability (Fig. 16.18) is re-
of the subgroups, illustrating the concentric ferred to as a proportional effect which implies
zoning and isolated groupings of the data. The that the variability is predictable although the
statistics of each subgroup can then be applied data do not reflect a normal distribution. The
to speciftc areas of the 5220 level and local contour plots (Fig. 16.17) show that in certain
 16: TRINITY SILVER MINE, NEVADA 405

16
(a)
15
14
10 2

13
16
12
(d)
15
1
1 14

10 2
0 2 50 500 750 1000
13
16
(b) 12
15
1
14 1
10 2

0 250 500 750 1000

13
12 16
1 15 (e)
1
0 750 10 2 14
500
16 13
50
1000(c) 12
15 2
1
14
10 2

1
0 500 750
13
12 1000
1
1 0 250 500 750 1000
250
maps of the Ag grades (oz t1) of the 5220 bench demonstrating the spatial integrity of the
FIG. 16.15 Indicator
data. The grade boundaries are at (a) 0.4, (b) 0.8, (c) 1.3, (d) 5.0 and (e) 10.0 oz t1 (14, 27, 45, 171, and 343 g t1).

locations the change in local variance mirrors 13 to 8 oz Ag 2. This again demonstrates the
or is greater than the change in the mean (the effect that the few high values have on the
proportional effect). However elsewhere the variability.
variability remains roughly constant whilst Directional semi-variograms show different
the local mean fluctuates. ranges in different directions (Fig. 16.19a–d).
Continuity was greater in the E–W direction
(300 ft) (Fig. 16.19c) compared with all other
16.8.4 Semi-variograms
directions. The minimum continuity was 90 ft
Semi-variograms are used to quantify the in a N–S direction (Fig. 16.19a). An E–W elipse
spatial continuity of the data (see Chapter 9). measuring 300  90 ft (Fig. 16.19f) would be the
The range (a) of the semi-variogram deftnes optimal moving window shape and size to give
a radius around which the local estimate has the lowest variability for a local estimate.
the least variability. The omnidirectional semi- The NW–SE directional semi-variogram
variogram of the 5220 bench (Fig. 16.19a) (Fig. 16.19d) shows a “hole effect” (Journel
shows a range of 120 ft (35 m). When samples & Huijbregts 1978, Clark 1979) which results
larger than 20 oz t1 (686 g t1) were excluded from alternate comparison of high and low
(Fig. 16.19b) the sill was reduced by a third from zones in the NW half of the deposit.
406 M.K.G WHATELEY, T. BELL & C.J. MOON

(a)
1600
0.3 0.4 0.3 0.1 0.0 0.0

1500 0.2 0.8 0.9 0.2 0.1 0.0

0.3 0.3 0.4 1.4 1.5 0.6 0.2 0.1 0.1

1400 0.4 0.4 0.5 1.2 1.5 1.0 0.3 0.3 0.3 0.2 0.1 0.1

0.5 0.6 1.0 0.9 0.6 1.0 1.4 0.8 0.2 0.2 0.1 0.1 0.1 0.1 0.2 0.4 0.3 0.1 0.1 0.5

1300 0.6 0.9 1.0 0.5 0.7 1.4 1.6 0.9 0.2 0.3 0.9 2.2 3.6 2.3 1.8 1.1 0.3 0.2 0.4 0.5 0.5

0.5 0.6 0.5 0.9 0.8 0.6 1.4 4.7 8.1 7.1 7.3 5.2 3.6 2.8 1.5 1.9 1.9

1200 0.4 0.4 0.3 1.0 1.4 0.9 1.0 3.3 3.4 5.7 6.6 6.3 7.7 0.1 3.0 3.3 2.2

0.3 0.4 0.6 0.5 0.6 1.1 1.3 1.2 1.3 2.1 3.8 3.1 2.0 2.2 1.9

1100 3.3 0.1 0.1 0.1 0.2 0.2 0.2 0.2 0.3 0.3 0.2 0.2 0.6 1.0 2.8

0 250 500 750 1000

(b)
1600
0.1 0.4 0.2 0.1 0.0 0.0

1500 0.1 2.6 2.6 0.3 0.0 0.0

0.2 0.2 0.6 2.5 2.5 0.6 0.2 0.1 0.2

0.5 0.5 0.5 0.9 0.9 0.9 0.3 0.4 0.4 0.2 0.1 0.0
1400
0.3 0.4 0.9 0.9 0.5 0.9 2.1 2.1 0.1 0.1 0.1 0.1 0.1 0.1 0.1 0.7 0.6 0.1 0.1 0.3

0.3 0.8 0.9 0.5 0.7 1.0 1.9 1.8 0.1 0.3 1.1 4.1 6.6 4.3 3.8 3.1 0.5 0.2 0.3 0.3 0.3
1300
0.5 0.6 0.4 1.2 1.2 0.5 1.1 4.4 6.7 5.4 5.6 5.2 5.5 4.8 1.8 2.6 2.8

0.3 0.3 0.2 1.4 1.4 0.8 0.8 0.8 4.7 5.0 5.1 4.2 5.5 5.5 2.0 2.5 2.8
1200
0.2 0.8 1.0 0.7 0.5 1.4 1.9 1.9 1.7 3.1 5.2 4.6 1.7 1.6 4.2

16.4 0.1 0.0 0.1 0.2 0.1 0.3 0.1 0.1 0.2 0.2 0.2 0.5 0.4 6.6
1100

0 250 500 750 1000

FIG. 16.16 Window

statistics performed on the silver grades of the 5220 bench, based on a transformed grid
oriented E–W with 30 m square, moving window overlapping by 15 m: (a) mean Ag oz t 1; (b) standard
deviation.
 16: TRINITY SILVER MINE, NEVADA 407

(a)
Pit outline
1400

0.6
Fig 16.18a
1300

Fig 16.18b
1200
0.6
0.4
1100
(b)
2.0
Pit outline
1400
0.4

Fig 16.18a
1300
0.6 2.0
Fig 16.18b
1200
0.4

1.0
1100
0 250 500 750 1000

FIG. 16.17 Windowstatistics performed on the silver grades of the 5220 bench, based on a transformed grid
oriented E–W with 30 m square, moving window overlapping by 15 m: (a) contours of mean Ag oz t 1,
(b) contours of standard deviation. (From Bell & Whateley 1994.)

(a) (b)
6 Mean Ag 7 Mean Ag
Standard deviation Standard deviation
5 6

5
4
Ag oz t  1
Ag oz t  1

4
3
3
2
2
1
1

0 0
100 300 500 700 900 1100 100 300 500 700 900 1100
Distance (m) Distance (m)

FIG. 16.18 Proftles

1 and 2 (a and b respectively) of the window mean and standard deviation showing the
proportional effect and highlighting areas of expected high error. (From Bell & Whateley 1994.)
408 M.K.G WHATELEY, T. BELL & C.J. MOON

(a) (d)
13 11
11 10
10 9
9 8
Gamma (h)

Gamma (h)
8 C0  0.5 7 C0  1.0
6 C  12.0 6 C  10.3
5 a  110 5 a  80
4 3
3 2
1 1
0 0
0 100 200 300 400 500 600 0 100 200 300 400 500 600
Distance (m) Distance (m)

(b) (e)
13 11
12 10
11 8
9 7
Gamma (h)

Gamma (h)
8 6
7 5
5 C0  1.3 4 C0  1.5
4 C  11.8 3 C  8.2
3 a  310 2 a  110
1 1
0 0
0 100 200 300 400 500 600 0 100 200 300 400 500 600
Distance (m) Distance (m)

(c) (f )
11
9 Anisotropy factors
8 1.4 3.9
7 310 m
Gamma (h)

6 110 m
1.6
5 C0  1.0
4 C  8.1 1.0 130 m
3 a  130 80 m
1
0
0 100 200 300 400 500 600 Ellipse of anisotropy
Distance (m)

FIG. 16.19 Semi-variograms

(spherical scheme models) for the silver assays of the 5220 bench: (a) N–S
direction; (b) NE–SW direction; (c) E–W direction; (d) SE–NW direction; (e) omni-directional; (f) ellipse of
anisotropy and anisotropy factors. (From Bell & Whateley 1994.)

16.8.5 Conclusions understand what constitutes the population,

in order that sensible subdivisions into sub-
Attempting a global estimate of the grade for populations can be made. It is incorrect to
the entire bench is irrelevant in a mining situa- evaluate an exhaustive dataset if the data are
tion because geologists and mining engineers derived from different episodes of mineralisa-
are only interested in ore grade material. The tion, when their physical and chemical charac-
skewed population means that ore grade ma- teristics differ. If the global estimate is used to
terial will be underestimated. It is essential to make local estimates of grade, then the grade
 16: TRINITY SILVER MINE, NEVADA 409

will be overestimated, because no account is carried out, each with a different conftdence
made of the spatial relationship of the samples level. The conftdence in the reserve estimate
or of the high variability. grew as the amount of borehole data increased.
To improve the grade estimation it is neces-
sary to try to establish normal distributions
16.9.1 Evaluation of initial exploration data
within the whole skewed population, using
geology, structure, mineralisation, alteration, Immediately after exploration was initiated in
etc., rather than transforming the whole popu- early 1982, it was realized that a signiftcant
lation. The variability within each subpopu- tonnage of disseminated silver existed. During
lation is thus reduced, which increases the the 1982–83 exploration program the oxide
conftdence in the subpopulation mean. mineralisation had not been fully evaluated,
The high values cause extreme variability and the initial reserve estimates only took the
which influences the mean. By excluding them sulftde mineralisation into account.
the mean and variance are reduced and the
estimation becomes more reliable.
16.9.2 Evaluation of additional exploration
By applying these techniques to the data
data
above cut-off grade (omitting outliers), a popu-
lation which approximates a normal distribu- Additional exploration drilling took place in
tion is achieved. This improves the conftdence 1986 with the intention of identifying addi-
in the estimation of the mean. By subdividing tional sulftde mineralisation, and to collect
the data above cut-off into subpopulations, samples for metallurgical and engineering
variability is further reduced. studies. In mid 1986 it became apparent that
Indicator maps show that subgroups, as a small area of high grade silver oxide miner-
deftned by statistical parameters, are spatially alisation could possibly support a heap leach
arranged. High local variance is a function of operation. The reserve potential of the oxide
boundaries between these populations. Appli- zone was calculated by several members of
cation of the moving window statistics should the joint venture, using different methods and
therefore be viewed in terms of the spatially cut-off grades. Table 16.6 summarizes the
distributed groups, trying to avoid crossing results.
subpopulation boundaries. The tonnage calculations vary from 0.932 to
From the univariate statistics and spatial 1.304 Mt where a Ag cut-off grade of 1.5 oz t1
distribution of the data the conclusion is that (50 g t1) was used. Where the cut-off grade was
a global estimate must be derived from local raised to 2 and 3 oz t1 (69 and 103 g t1), as one
estimates rather than the whole population. would expect, the tonnage was reduced but the
In hindsight, the best ore reserve estimation average grade increased. These differences in
could be derived using the average of the esti- tonnage and average grade were accounted for
mates from discrete zones or the average of the when thin, lower grade intersections used in
estimates derived from the areas between con- earlier calculations were omitted, and modifted
tours of the grades. Bell and Whateley (1994) polygons were used. No reference was made to
undertook global estimation of the grade by the geology of the deposit in any of the above
applying a number of estimation techniques calculations.
to the original exploration borehole data of the This demonstrates the difftculty in calculat-
5220 level. They compared the results with ing reserve estimates in the exploration phases
the blasthole dataset and concluded that linear of a program (Bell & Whateley 1994). Different
interpolation provided grade estimates with methods, by different people using differing
the lowest variance. cut-off grades, produces a variety of estimates.
It is not possible to quantify the error of estima-
tion using these 2D manual methods of reserve
16.9 ORE RESERVE ESTIMATION estimation. It is not even practical to compare
the early reserve estimates with the ftnal ton-
During the life of the project, from the explora- nage and grade extracted, because in the ftnal
tion stages to the feasibility and mine design estimate a cut-off grade of 1.3 oz t1 (45 g t1)
studies, different ore reserve calculations were was used.
 16: TRINITY SILVER MINE, NEVADA 411

TABLE 16.6 Summary of the ore reserve calculation for the silver oxide zone, Trinity Mine, Nevada.

Method Tonnage Composite Ag cut-off Tonnage Average Total Ag

factor length grade  106 Ag grade oz (g)  106
ft3 t1 (SG) ft (m) oz t1 (g t1) oz t1 (g t1)

Polygons–USB 13.3 (2.41) 20 (6.1) 1.5 (50) 0.967 6.95 (238) 6.22 (213)
N–S cross-sections, USB 13.3 (2.41) 10 (3.0) 1.5 (50) 1.304 6.16 (211) 8.03 (275)
E–W cross-sections, USB 13.3 (2.41) 10 (3.0) 1.5 (50) 1.293 5.90 (200) 7.63 (262)
N–S cross-sections, USB 13.3 (2.41) 10 (3.0) 1.5 (50) 0.932 7.69 (264) 7.17 (246)
Polygons, Santa Fe 13.3 (2.41) 20 (6.1) 3.0 (100) 0.669 9.10 (310) 6.09 (209)
N–S cross sections, 13.3 (2.41) 10 (3.0) 2.0 (69) 0.870 8.00 (274) 6.96 (240)
Santa Fe and USB

16.9.3 Reserve estimate for mine planning with the early ore reserve estimates). This may
be partially a result of an edge effect where the
In 1987 a computerized 3D block modeling
majority of any given block at the edge of the
technique was used to estimate reserves and
deposit may lie outside the mineralized area,
the data thus generated were used for mine
but because of the shape of the boundary they
planning. A conventional 2D inverse distance have been included in the reserve estimate. An
squared (1/D2, see section 9.5.1) technique was
additional factor is the use of 13.7 cu ft t 1
used to interpolate grades from borehole infor-
(2.34 t m3) as the ore and waste speciftc grav-
mation to the block centers, using a 150 ft
ity, to calculate in situ tonnage compared
(46 m) search radius (Baele 1987). The blocks
with only 13.3 cu ft t1 (2.41 t1 m3) used in the
were given a 25  25 ft (7.6  7.6 m) area and exploration evaluation.
were 15 ft (4.6 m) deep. Borehole assays were
collected at 5 ft (1.5 m) intervals, but were
composited by a weighted average technique 16.9.4 Updated tonnage and grade estimate
over 15 ft (4.6 m) to conform to the block
model. Geology and alteration were incorpor- The grade and tonnage of the ore mined from
ated into the block model by assigning a rock the ftrst three benches was signiftcantly differ-
and alteration code to each block. Tonnage and ent from that predicted by the 1/D2 block
grade was estimated at various cut-off grades model. An updated reserve estimate was hand-
(Table 16.7). calculated using cross-sections to evalu- ate the
Despite the geological control, this method discrepancy. The geology and structure from
of reserve estimation appears to have overesti- the mined-out area were plotted on the cross-
mated the reserve potential (in comparison sections and used to project the orebody

TABLE 16.7 Reserve estimates on the silver oxide zone, using the 1/D2 block model, Trinity Mine, Nevada.

Tonnage Tonnage Composite Ag cut-off Average

 106 factor length ft (m) grade oz t1 Ag grade
ft3 t1 (SG) (g t1) oz t1 (g t1)

2.65 13.7 (2.34) 15 (4.5) 0.5 (17) 3.05 (105)

2.05 13.7 (2.34) 15 (4.5) 1.0 (34) 3.76 (129)
1.65 13.7 (2.34) 15 (4.5) 1.5 (50) 4.46 (153)
1.26 13.7 (2.34) 15 (4.5) 2.0 (69) 5.19 (178)
0.88 13.7 (2.34) 15 (4.5) 3.0 (100) 6.33 (217)
0.63 13.7 (2.34) 15 (4.5) 4.0 (137) 7.52 (258)
410 M.K.G WHATELEY, T. BELL & C.J. MOON

TABLE 16.8 Updated reserve estimate on the silver oxide zone, Trinity Mine, Nevada.

Method Tonnage Composite Ag cut-off Tonnage Average Total

factor length ft (m) grade  106 Ag grade Ag oz (g)
ft3 t1 (SG) oz t1 (g t1) oz t 1 (g t1)  106

N–S cross-sections 13.7 (2.34) 15 (4.5) 1.5 (50) 0.899 6.94 (238) 6.24 (214)
Bench plans* 13.7 (2.34) 15 (4.5) 1.6 (55) 0.859 7.33 (251) 6.29 (216)
Blocks 13.7 (2.54) 15 (4.5) 1.6 (55) 0.963 6.55 (225) 6.31 (217)
15  15  15 ft
(4.5  4.5  4.5 m)

* These estimates exclude the ftrst three benches which had already been mined out.

geometry downwards (Reim 1988). Reim felt 1 Geological control. It is most important to
that the unsatisfactory grade control given by establish the geological and structural control
the 1/D2 method could be accounted for by of mineralisation as early in a project as pos-
the inadequate structural control in the 1/D2 sible. In the example used here, the 1/D2 block
model, variability in grade, and paucity of model had good geological control, but the
sampling data within the southwest portion structural control was lacking. This resulted in
of the oxide ore body. Additional drilling overestimation of reserves.
increased the sample density in poorly repres- 2 Sampling methods. It is important to control
ented areas and this helped deftne the orebody the sampling method to ensure consistent
geometry. The reserve estimate calculated and reliable sampling, e.g. down-the-variation
using the updated cross-sections is shown in in silver values derived from either reverse
Table 16.8. circulation or percussion drilling may show
For mine planning purposes a mineable re- disparity.
serve was calculated on a bench by bench basis. 3 Cut-off grade. This varied between estima-
Mineralized zones were transferred from the tion methods from 3.0 oz t1 (103 g t1) in the
cross-sections on to 15 ft (4.5 m) bench plans. exploration assessment to 1.3 oz t1 (45 g t1)
Polygons were constructed around each during mining (Tables 16.6–16.8). Compari-
borehole and a reserve calculated (Table 16.8). sons of reserve estimates using different cut-off
grades can be achieved if grade–tonnage graphs
are constructed. These are difftcult to con-
16.9.5 Grade control and ore reserve
struct during the exploration phase as data are
estimation using blastholes
sparse. Normally economic criteria are used
Mine grade control was exercised by sampling to establish the cut-off grade (Lane 1988). US
and assaying the blasthole cuttings. Mined Borax assumed a commodity price of $6.50 per
blocks were 15  15  15 ft (4.6  4.6  4.6 m) oz and expected silver recoveries between 65
and only blocks 1.3 oz t1 (45 g t1) were mined and 79% (amongst other criteria) to establish
as ore. A ftnal reserve was calculated using the 1.3 oz t1 (45 g t1) cut-off grade used during
these blocks and a slightly higher reserve with mining.
a lower grade was estimated (Table 16.8). 4 Sample mean. The mean grade is dependent
upon the method used to estimate it (section
16.8.2). The arithmetic mean probably over-
16.9.6 Factors affecting ore reserve
estimated the mean in this study but this
calculations
could have been improved by establishing
A number of factors influenced the validity of subpopulations with normal distibutions and
ore reserve estimates, namely geological con- averaging the estimates from these discrete
trol, sampling method, cut-off grade, sample zones.
mean, analytical technique, speciftc gravity 5 Analytical technique. In order to compare
and dilution factors: samples used to estimate reserves using
 16: TRINITY SILVER MINE, NEVADA 411

exploration boreholes and blastholes, a simple Borax initially expected a 20% dilution, but
linear regression of cyanide-extractable silver this was later modifted to 7% after further
on total recoverable silver was performed. A assessment during mining.
correlation coefftcient of 0.94 indicated a
strong positive relationship. Above 10 oz t 1
(343 g t1) Ag, the conversion of total silver to 16.10 SUMMARY AND CONCLUSIONS
cyanide-eachable silver was less accurate.
Changing analytical procedure during an ex- The Trinity oxide orebody deftned a small but
ploration or feasibility exercise, without under- economic silver deposit, amenable to cyanide
taking an overlapping series of analyses which heap leaching. The orebody was fairly continu-
can be used for correlative purposes, introduces ous to the northeast but became fragmented
an unnecessary risk of error. and fault controlled to the southwest. The ore
6 Specific gravity. Initially, densities of 13.0 zone was deftned by stacked tabular to lenticu-
and 13.3 ft3 t1 (2.34 and 2.41 g cm3) were used lar bodies dipping steeply to the west and
to calculate tonnages. Density was determined northwest and offset by N–S and NW–SE
on a number of different rock types and cross-faulting. Mineralisation tended to be
an average of 13.7 ft3 t1 (2.34 g t1) was calcu- controlled by faulting, alteration, and fracture
lated, which was used for the ftnal tonnage density. The sharp ore grade boundaries along
calculations. the footwall and hanging wall suggest geo-
7 Mining dilution. Dilution of ore by waste logical control. These boundaries may reflect
during mining is inevitable but careful grade change in host rock permeability or sharp fault
control will minimize the risk of dilution. contacts.
Dilution usually occurs in ore deposits where The Trinity deposit illustrates the estima-
grade is highly variable, as at the Trinity Mine. tion problems encountered in deposits which
The assumption that a blasthole in the center have highly skewed data. The problem is
of a 15  15  15 ft block (4.5  4.5  4.5 m) is compounded when trying to estimate grade
representative of that block is not necessarily and resource potential from a relatively low
true. The arbitrary boundary between a block number of exploratory boreholes. Recognizing
showing a grade less than cut-off and a high the importance that a few very high values
grade block is drawn half way between the two may have upon the estimation of the global
holes. The low grade ore may extend farther mean is important. If the variability of the
into the high grade block than one expects, data can be reduced the conftdence in the mean
thus adding to the mining dilution problem. US goes up.
 17: EKATI AND DIAVIK DIAMOND MINES 413

17
412 M.K.G WHATELEY, T. BELL & C.J. MOON

Diamond Exploration –
Ekati and Diavik Mines
Charles J. Moon

The discovery of major diamond deposits in its use in jewellery that is most important for
the barren lands of the Canadian Arctic was the exploration geologist (Fig. 17.1). Although
one of the great exploration successes of the 90% of industrial diamonds are produced syn-
1990s. Targeting of the remote discovery area thetically, it is, at present, uneconomic to pro-
resulted from remarkable persistence by a duce the much more valuable gem diamonds
junior company, although two major mining artiftcially.
companies undertook the detailed exploration Diamond deposits have many similarities to
and development. industrial mineral deposits in that the value
of the raw material is only a small proportion
of the value of the ftnished product. The world
17.1 WORLD DIAMOND DEPOSITS AND rough diamond market in 2001 was worth
EXPLORATION $US7.9 billion whereas the retail diamond
jewellery market was worth $US56 billion of
Diamond is the hardest mineral known and has which $US13.5 billion was the value of
signiftcant industrial uses, e.g. in the cutting of polished stones (Diamond Facts 2003). Unlike
other materials and diamond drilling, but it is other minerals, wholesaling of gem diamonds

FIG. 17.1 The

ftnal product:
diamonds from the Diavik Mine.
(Courtesy Diavik Diamond Mines
Inc.)