1966 Deere D.U. y Miller R.P. - Eng Class and I Prop For Intact Rock

Best Available Copy
AFWL-T-65-6 AFWL-TR
4 )65-116
ENGINEEEiNG CLASSiFICATIONj
AND INDEX PROP2RTIES
FOR INTACT ROCK
D. U. Deere
R. P. Millor
University of Illinois
Urbana, Illinois
A. Contract AF 29(601)-6319
TECHNICAL REPORT NO. AFWL-TR-65-116

December 1966
AIR FORCE WEAPONS LABORATORY

Research and Technology Division
Air Force Systems Command
Kirtland Air Force Base
New Mexico p0
VC &Q'i76O1967
o 0014076RO1(5
AFWL-TR-65--116
Research andTechnology Division

AIR FORCE WEAPONS LABORATORY
Air Force Systems Command
Kirtland Air Force Base
New Mexico
When U. S. Government drawings, specifications, or other data are used for

any purpose other than a definitely related Government procurement operation,
the Government thereby incurs no responsibility nor any obligation whatsoever,
and ,he fact that the Government may have formulated, furnished, or in any
way supplied the said drawings, specifications, or other data, is not to be
regarded by implication or otherwise, as in any manner licensing the holder
or any other person or corporation, or conveying any rights or permission to
manufacture, use, or sell any patented invention that may in any way be
related thereto.
This report is made available for study with the understanding that
proprietary interests in and relating thereto will not be impaired. In
case of apparent conflict or any other questions between the Government's
rights and those of others, notify the Judge Advocate, Air Force Systems
Command, Andrews Air Force Base, Washington, D. C. 20331.
Distribution of this document is unlimited.
NIMTE
$EnTiON
IW,SEUTOI [
IFIMATN
U VOi
rUIUTOA, AYAFIAILITY C400

ST. AVAIL maud/it E
!a
?1
7-
AFWL-TR-65--116
ENGINEERING CLASSIFICATION AID INDEX

PROPERTiS FO2 I1"ChCT ROCK
D. U. Deere
R. P. Miller
University of Illinois
Urbana, Illinois
Contract AF29(601)-6319
TECHNICAL REPORT NO. AETL-TR-65-116
Distribution of this document

is unlimited.
AFm-TR-65-116
This report was preparcd by the Dci;:rtn7nt of Civil ELgincering at the

University of Illinois, Urbana, Illinoic under Contract AF29(601)-6319. The
research was performcd under Program Elc:nt 6.16.46.0l.D, Project 5710, Sub-
task 13.144, and was funded by the Defense Atomic Support ,.cncy (DASA).
Inclusive dates of research were 1 February 1964 through I April 1966.

The report was submitted 6 October 1966 by the AFUL Project Officer, Captain
Joseph J. O'Kobrick (WLDC). Captain Thomas E. O'Brien was the former Project
Officer for the Air Force.
Dr. Don U. Deere, Professor of Civil Engineering and of Geology, was the
project supervisor. R. P. Miller, Research Associate in Civil Engineering was
immediately responsible for all phases of the test program and preparation of the
written report. E. J. Cording, Research Assistan in Civil Engineering, assisted
in development of special equipment and testing techniques. J. T. Wilband,
Teaching Assistant in Geology, performed the petlographic analysis. J. H. Coulson,
Graduate Fellow in Civil Engineering, assisted in specimen preparation and testing,
and prepared the computer programs for data reduction and statistical analysis.
The drawings were prepared by E. E. Boatz. Acknowledgment is extended to the
Corps of Engineers, Walla Walla District, and the U. S. Bureau of Reclamation,
Denver, who furnished core samples; and also to the numerous rock quarries, most
of whom furnished block samples without charge.
This report has been reviewed and is approved.
J140~qH.O'KOBRICK
Captain, USAF
Project Officer
ALLEN F. DILL GEORGEC. DBY, JR.

CDR, CEC, USNR Colonel, USAF
Chief, Civil Engineering Branch Chief, Development Division
li
ABSTRACT
Physical and elastic properties of NX-size rock core from 27

localities were Investigated in order to develop an engineering classifi-
cation system for intact rock, and also to develop index proertiec related
to important physical and engincerino charocteristics. Thirteen rick types
are represented. Lnlcrawury tests were conducted on these rocks as follows:
urni, weight, Shore scleroscope hardness, Schmidt hammer hardness, abrasion
hardness, absorption, sonic-velocity stress-strain under cyclic loading
to 5,000 psi. uniaxial stress-strain t-) failure, and point-load tensile
strength. A total of 257 specimens with L/D ratios of 2:1 wee tetcd.
Statistical studies were conducted with the Il 7094 computer to determine
correlation and regression relationships for selected pairs of variables.
A system of engineering classification is proposed in which rocks are
classified on the basis of their strength and modulus properties either
obtained directly from laboratory tests, or approximately from index proper-
ties recommended herein. Application of the proposed engineering classifi-
cation system to data obtained by others is shown by individual charts for
each of several different rock types. Five charts are presented for esti-
mating the strength or modulus properties for intact rock from toe numerical
indices obtained by either the Schmidt hamneri Lie Shore sclerosccpe, or
the sonic pulse velocity, all used in conjunction with the unit weight of
the rock.
III
This page intentionally left blank.
iv
-II
CONTENTS
SECTIOiN PAGE
1. INTRODUCTICA I
I. Nature of the Problem I
2. Objectives of Investigation 2
3. Scope of Study 3
2. EXISTING SYSTEMS OF ROCK CLASSIFICATION 5
1. Cencral 5
2. Geological Classification 5
3. Classification by Physical Character of Rock Materials 7
4. Classification by In-Situ Character of Rock Mass 11
3. HARDNESS OF ROCKS 15
1. G.neral 15
2. Types of Rardne.s and Their Methods of Measurement 16
3. Toughness and Resilience 20
4. Hardness and Other Physical Properties 25
5. Hardness and Drilling 30
4. EXPRERIMENTAL STUDIES 33
I. General 33
2. Description of Rocks 33
3. Preparation of Test Specimens 38
It. Laboratory Tests 46
5. Test Results 67
5. DISCUSSION AND INTERPRETATION OF EXPERIMENTAL DATA 84
1. General 84
2. Statistical Procedures 84
3. Physical Properties 87
4. Elastic Properties 90
5. Relationships Among Various Physical Properties 99
6. Relationships Among Physical and Elastic Properties 108
7. Comparison of Static and Dynamic Properties 122
V
SECTION PAGE
6. ENGINEERING CLASSIFICATION AMD INDEX PROPERTIES FOR IITACT ROCK 136

1. General 136
2. Proposed Engineering Classification 136
3. Index Properties 160
4. Unit Weight as an Index of Strength and Deformation 161
5. Schmidt Hardness as an Index of Strength and Deformation 162
6. Shore Hardness as an Index of Strength and Deformation 171
7. Soni. Veiocity as an Index of Deformation 180
7. SUMMARY AND CONCLUSIONS 185
1. Summary 185
2. Conclusions :86
3. Suggestions for Further Research 189
APPENDIX A. PETROGRAPHIC DESCRIPTIONS AND THIN-SECTION
PHOTOMICROGRAPHS 191
APPENDIX B. TYPICAL TEST RESULTS 219
APPENDIX C. STATISTICAL METHODS OF ANALYSIS 276
BIBLIOGRAPHY 291
DISTRIBUTION 301
vi
LIST OF TABLES
Table No. Page
2.1 SCALE OF ROCK HARDIN.ESS 9

2.2 CLASSIFICATION OF ROCKS FOR ROCK tZCIClLANICS II
3.1 MOHS HARDNESS SCALE AND CRYSTAL STRUCTURE 17
4.1 LIST OF ROCK TYPES 34
4.2 SUMARY OF AVERAGE PHYSICAL PROPERTIES FOR ROCK 71
4.3 SUMMARY OF AVERAGE ELASTIC PROPERTIES FOR ROCK 75
5.1 NUMBER OF TESTS AND COEFFICIENTS OF VARIATION
FOR MEASURED ROCK PROPERTIES 86
5.2 EFFECT OF OVEN-DRYING ON THE DILATATIONAL WAVE
VELOCITY FOR ROCKS 99
5.3 PERCENT BY WHICH DYNMIC PROPERTIES ARE GREATER
THAN STATIC PROPERTIES FOR ROCKS 132
6.1 ENGINEERING CLASSIFICATION FOR INTACT ROCK 137
vif
LIST OF FIGURES
Figure Vo. Page
3.1 RELATION:SHr? &:T1:7EN1 RLATi,2 VALUES C7 li?-ACT

TOUGHINESS AN.I W'13 RESIST;.JCE FC. ROCK 22
3.2 RELATIONSHIP BEETZEN S1:ORE I'PRDNESS k.YD ALCSION
HARDNESS FOR ROCK 23
3.3 1KOULI OV RESILIENCE OF S0ORE TYPICAL ROCKS TESTED
BY U. S. BUREAU OF MINES 24
3.4 RELATIONSHIP BET1W,'EEN ABRASION HADDNESS AND ULTIMATE
COMPRESSIVE ST'ENGTH FOR ROCK IN UNIAXIAL CONPRESSXON 26
3.5 RELATIONSHIP BETWEEN SHCRE HARDNESS P.ND ULTIt,1ATE
COMPRESSIVE STRENGTH FOR ROCK IN UNIAXIAL CONIPRESSION 28
3.6 RELATIONSHIP BETWEEN SHORE HARDNESS AND ULTIATE
COMPRESSIVE STRENGTH FOR ROCK IN UNIAXIAL COilPRESSION 29
4.1 DRILL PRESS 39
4.2 BLOCK SAMPLE CLAMPED IN POSITION FOR CORE DRILLING 41
4.3 DIAMOND ROCK SAV4 43
4.4 CORE SPECIMEN CLAMPED IN SAW CARRIAGE 44
4.5 LAPPING MACHINE 45
4.6 CORE SPEC'iMEN, CYLINDRICAL CORE HOLDER, CONDITIONING
RING, AND PRESSURE PLATE 45
4.7 SHORE SCLEROSCOPE 48
4.8 LONGITUDINAL SECTION OF THE TYPE L SCH14IDT TEST
HAMMER 50
4.9 SCHMIDT TEST HAMMER BESIDE SPECIMEN CRADLE BLOCK
AND BASE PLATE 51
4.10 ABRASION APPARATUS 53
4.11 SCHEMATIC INTERCONNECTION DIAGRAM FOR SONIC PULSE
VELOCITY MEASURING EQUIPMENT 56
4.12 TYPICAL SONIC PULSE WAVE-FORM DISPLAYS 58
4.13 EQUIPMENT FOR MEASURING SONIC PULSE VELOCITY AND
STATIC STRESS-STRAIN DURING CYCLIC LOADING 60
4.14 CORE SPECIMEN IN POSITION FOR MEASURING SONIC PULSE
VELOCITY. STRESS AND STRAIN UNDER UNIAXIAL LOAD 61
4.15 STRAIN READ-OUT 62
4.16 COMPRESSION TEST MACHINE 64
4.17 CORE SPECIMEN IN POSITION FOR COMPRESSION TEST TO
FAILURE 65
viii
Figure No. Page
4.18 TYPICAL CC;:PRESSIVE FAILURES 66

4.19 SCHE,!.ATIC DIA,GPRmI FOR POIMlT-LC,lD TE[SILE TEST 68
4.20 POINT-LC%,D TEE2SILE TEST ECUIiEYT 69
4.21 CORE SFPCIICIMN IN POSITION FOR POIiNT-LO;AD TENISILE
TEST 70
5.1 REPRESENTATIVE STRESS-STRAIN CURVE FOR ROCK IN
UNIAXIAL CO;MPRESSION 91
5.2 TYPICAL STRESS-STRAIN CURVES FOR ROCK IN UNIAXIAL
COMPRESSION TO FAILURE 94
5.3 RELATIONSHIP BETWEEN AVERAGE VALUES OF DRY UNIT
WEIGHT AND 1LTIMATE COMPRESSIVE STRENGTH FOR ROCK
IN UNIAXIAL CCMRESSION 100
5.4 RELATIONSHIP BETWEEN AVERAGE VALUES OF SHORE
HARDNESS AND ULTIMATE COMIPRESSIVE STRENGTH FOR
ROCK IN UNIAXIAL COMPRESSION 102
5.5 RELATIONSHIP BETWEEN AVERAGE VALUES OF SCHMIDT
HARDNESS AND ULTIMATE COMPRESSIVE STRENGT' FOR
ROCK III UNIAX;,IAL COMPRESSION 103
5.6 RELATIC'"*!5P BETWEEN AVERAGE VALUES OF ADF.SION
HARDNESS AND ULTItrATE COMPRESSIVE STRENGTH FOR
ROCK IN UNIAXIAL COMIPRESSION 104
5.7 RELATIONSH!P BETWE.N ABSORPTION AND AVERAGE VALUES
OF ULTIMATE COMPRESSIVE STRENGTH FOR ROCK IN
UNIAXIAL COMPRESSION 106
5.8 RELATIONSHIP BETWEEN AVERAGE VALUES OF POINT-LOAD
TENSILE STRENGTH AW") ULTIMATE COMPRESSIVE STRENGTH
FOR ROCK IN UNIAX.AL COMPRESSION 107
HARDNESS AND ABRASION HARDNESS FOR ROCK 109
5.10 RELATIONSHIP BETWEEN AVERAGE VALUES OF DRY UNIT
WEIGHT AND TANGENT MODULUS OF DEFORMATION AT STRESS
LEVEL OF ONE-HALF ULTIMATE STRENGTH FOR ROCK IN
5.11 RELATIONSHIP BETWEEN AVERAGE VALUES OF [ILT&'ATE
COMPRESSIVE STRENGTH AND INITIAL MODULUS OF
DEFORMATION FOR ROCK IN UNIAXIAL COMPRESSION III
5.12 RELATIONSHIP BETWEEN AVERAGE VALUES OF ULTIMATE
COMPRESSIVE STRENGTH AND TANGENT MODULUS OF
DEFORMATION AT STRESS LEVEL OF ONE-HALF ULTIMATE
STRENGTH FOR ROCK IN UNIAXIAL COMPRESSION 112
HARDNESS AND TANGENT MODULUS OF DEFORMATION AT
STRESS LEVEL OF ONE-HALF ULTIMATE STRENGTH FOR
ROCK IN UNIAXIAL COMPRESSION 114
ix
Figure NIo. Page
5.14 RELATIC::ZNIP iETWEEN AVZGE VALUES OF SCHU7IDT

:%DuLU5 OF~ DEF~iATIC' AT
VK~t"ES A.'D' TAN!GENlT V'
STRESS LEVEL OF O1A-,'.1.F ULTiMtATE SThEGTfi FR
ROCK i: UAxIAL Cc1:?11SSIC-3 115
5.15 RELATIC',IESP11 BEZTtV2C AI V'llUES
OF T,',',ZNT K'ODULUS 0? c:Fcac:ArxC:., AT STR',SS LEVEL
OF ONIZ-11LF ULTI;IATE STRNGTh FOR ROCK IN UNIAXIAL
COlPIESS I0N 116
5.16 RELATICNSHIP BETWEEN AVERA-%GE VALUES OF DRY UNIT
WEIGHT AND DILATATIoIAL WAVE VELOCITY FOR ROCK
UNDER UNIAXIAL CO,2PRESSIVE STRESS OF 5)000 psi 118
5.17 RELATIONSHIP BETWEEN AVERAGE VALUES OF INITI!AL
MODULUS OF DEFORMATION1 ANID DILATATIONAL WAVE
VELOCITY FOR ROCK UNDER UNIAXIAL COMPRESSIVE
STRESS OF 100-150 psi 119
5.18 RELATIONSHIP BETWEEN AVERAGE VALUES OF TANGENT
MODULUS OF DEFORMATION AT STRESS LEVEL OF ONE-HALF
ULTIMATE STRENGTH AND DILATATIONAL WAVE VELOCITY FOR
ROCK U1'DER UNIAXIAL C0Ci1PRESSIVE STRESS OF 5,OO0 psi 120
5.19 VELOCITY-MODULUS VECTORS FOR ROCK UNDER INCREASING
UNIAXIAL COMPRESSIVE STRESS 121
5.20 SONIC STRESS WAVE SUPERIMPOSED UPON A STATICALLY
APPLIED STRESS 125
5.21 THEORETICAL CURVE RELATING CONSTRAINED MODULUS OF
DEFORMATION AND YOUNG'S MODULUS OF ELASTICITY FOR
DIFFERENT VALUES OF POISSON'S RATIO 126
5.22 RELATIONSHIP BETWEEN AVERAGE VALUES OF INITIAL AND
CONSTRAINED MODULI OF DEFORMATION FOR ROCK UNDER
UNIAXIAL COMPRESSIVE STRESS OF 100-150 psi 127
5.23 RELATIONSHIP BETWEEN AVERAGE VALUES OF TANGENT
MODULUS AT STRESS LEVEL OF ONE-HALF ULTIMATE
STRENGTH AND CONSTRAINED MODULUS OF DEFORMATION
FOR ROCK UNDER UNIAXIAL COMPRESSIVE STRESS OF
5,000 psi 128
5.24 RELATIONSHIP BETWEEN AVERAGE VALUES OF CONSTRAINED
MODULI OF DEFORMATION COMPUTED FROM STATIC AND
DYNAMIC MEASUREMENTS FOR ROCK UNDER UNIAXIAL COM-
PRESSIVE STRESS OF 100-150 psi 130
5.25 RELATIONSHIP BETWEEN AVERAGE VALUES OF CONSTRAINED
MODULI OF DEFORMATION COMPUTED FROM STATIC AND
DYNAMIC MEASUREMENTS FOR ROCK UNDER UNIAXIAL COM-
PRESSIVE STRESS UP TO 5,000 psi 131
5.26 RELATIONSHIP BETWEEN AVERAGE VALUES OF DILATATIONAL
WAVE VELOCITY COMPUTED FROM STATIC AND DYNAMIC
MEASUREMENTS FOR ROCK UNDER UNIAXIAL COMPRESSIVE
STRESS OF 100-150 psi 133
x
Figure No. Page
5.27 RELATICNSHIP CET',.'EN AVERAGE 0; DILAT.ATIOC.L

OrL.I
WAVE V~th!OCITY CC:'XUTED FR~C1 STATIC A2:r DY1"!,"lC
MEASURE.$ v ,UTS
n,
rCCIL;%L
U2'R cc:".ssn
STRESS UP TO 5,000 psi 134
6.j ENGXNEERIIG CLASSIFICATION FOR IT-,CT ROCK -
AVERAGE VALUES FOR 13 ROCK TYPES
(28 Locations) 139
6.2 ENGINEERING CLASSIFICATION FOR INTACT ROCK -
INDIVIDUAL TEST RESULTS (124 Specimens, 13 Rock Types) 142
EFFECT OF ANGLE OF BEDDING AND FOLIATION 143
SUMMARY PLOT IGNEOUS ROCK
(176 Specimens, 75% of Points) 144
DIABASE (26 Specimens, 8 Locations, Various Sources) 145
GRANITE FtJ4ILY (CO Specimens, 16 Locations,
Various Sources) 146
BASALT AF,D OTNER FLOWi ROCKS (70 Specimens,
20 Locations, Various Sources) 147
6.8 E.NGINEERING CLASSIFICATION FOR INTACT ROCK -
SUMMARY PLOT SEDIMENTARY ROCKS
LIMESTONE AND DOLOMITE (77 Specimens, 22 Locations,
SANDSTONE (82 Specimens, 18 Locations,
HARD SHALE (34 Specimens, 14 Locations,
POROUS TUFF (44 Specimens, 2 Locations,
SUMMAJY PLOT METPBORPHIC ROCKS
QUARTZITE (41 Specimens, 11 Locations,
xi
Figure N2o. Page
6.15 WGENICERING CLASSIFICATICO FOR INTACT ROCK -

GNEISS (43 Specimcns, 14 Locations, Various
Sources) 157
6.16 ENGINr-:::,w CLASSIFIC,:,TION' FOR It.TACT ROCK -
MAnLE (22 Specirnans, 7 Locations) 158
SCHIST (61 Specimens, 11 Locations, Various Sources) 159
6.18 RELATIONSHIP BETWEEN AVERAGE VALUES OF 7a.R AND
ULTIMATE COHIPRESSIVE STRENGTH FOR ROCK IN UNIAXIAL
COMIPRESSION 163
6.19 RELATIONSHIP BETWEEN AVERAGE VALUES OF )a-R AND
ULTIMATE COIPRESSIVE STRENGTH FOR ROCK IN UNIAXIAL
COMPRESSION (SEMI-LOG PLOT) 165
6.20 ROCK STRENGTH CHART BASED ON SCFaimDT HARDNESS 167
6.21 RELATIONSHIP BETWEEN AVERAGE VALUES OF ya.R AND
TANGENT MODULUS OF DEFORAtATION AT STRESS LEVEL OF
ONE-HALF ULTIMATE STRENGTH FOR ROCK IN UNIAXIAL
COMPRESS ION 168
6.22 RELATIONSHIP BETWEEN AVERAGE VALUES OF 7a 2
.R AND
TANGENT MODULUS OF DEFORtlATION AT STRESS LEVEL OF
ONE-HALF ULTIMATE STRENGTH FOR ROCK IN UNIAXIAL
COMPRESS ION 169
6.23 ROCK MODULUS CHART BASED ON SCHMIDT HARDNESS 170
HARDNESS AND SCHMIDT HARDNESS FOR ROCK 172
6.25 RELATIONSHIP BETWEEN AVERAGE VALUES OF 7a'Sh AND
ULTIMATF COMPRESSIVE STRENGTH FOR ROCK IN
6.26 RELATIONSHIP BETWEEN AVERAGE VALUES OF 7a.Sh AND
ULTIMATE COMPRESSIVE STRENGTH FOR ROCK IN
UNIAXIAL COMPRESSION (SEMY-LOG PLOT) 175
6.27 ROCK STRENGTH CHART BASED ON SHORE HARDNESS 176
6.28 RELATIONSHIP BETWEEN AVERAGE VALUES OF Ya-Sh AND
TANGENT MODULUS OF DEFORMATION AT STRESS LEVEL
OF ONE-HALF ULTIMATE STRENGTH FOR ROCK IN UNIAXIAL
COMPRESS ION 177
6.29 RELATIONSHIP BETWEEN AVERAGE VALUES OF Ya2 .Sh AND
TANGENT MODULUS OF DEFORMATION AT STRESS LEVEL OF
ONE-HALF ULTIMATE STRENGIH FOR ROCK IN UNIAXIAL
COMPRESSION 178
6.30 ROCK MODULUS CHART BASED ON SHORE HARDNESS 179
xii
Figure No. Page
6.31 RELATIONSHIP CETWEEN AVERAE VALUES OF 7a.Vp AND

TArNGENT MODULUS OF DEFOF.;11ATIQ:1 AT STRESS
LEVEL OF C=-;',.;LF ULTiIlWA,TE SYi;i:ST:I FOR ROCK
IN UNIAX;AL Cc:7,:ESSIG 181
6.32 RELATIONSHIP CETE AVRrAGE VALUES OF 7aVp AND
TANGENT KODULJS OF DEFC=.TIO AT STRESS
LEVEL OF ONE-HALF ULTIPIATE STR.ENGTH FOR ROCK IN
UNIAXIAL COMPRESSION (SEmI-LOG PLOT) 182
6.33 ROCK MODULUS CHART BASED ON DILATATIONAL WAVE
VELOCITY 184
B.0IA STRESS-STRAIN BEHA. IOR AND SCNIC PULSE VELOCITY

B.28A FOR ROCK IN UNIAXIAL COMPRESSION 220-274
8.18- STRESS-STRAIN CURVE AND POISSON'S RATIO FOR ROCK
B.2FA IN UNIAXIAL COMPRESSION TO FAILURE 221-275
C.1 CHARACTERISTICS OF NORMAL DISTRIBUTION 281
C.2 RELATIONSHIP BETWEEN VARIACLES X AND Y 281
C.3 EFFECT OF SAMPLING VARIATION ON REGRESSION ESTIMATES
OF POPULATION tMEANS 289
xiii
PRINCIPAL SY6"ZLS
a * Y-intercept of rcgrcssion line

Ah = abrasion hnrdns
b - slope of regression line, or ccoefficicnt of rcaression
BLH a Baldwin-Lim -Ec! Iton
CL - confidence limits
D - diameter of core specimen
DI - deviation, error or residual
e - 2.71828, the base of natural logarithms
E - Young's modulus of elasticity
Ed - Young's modulus of elasticity determined dynamically
El - initial tangent modulus of deformation
Es = secant modulus of deformation
Et - tangent modulus of deformation

F - total failure load
g - acceleration of gravity
G - modulus of rigidity (or shear modulus)
K - bulk modulus of elasticity
L - length of core specimen
I/D - length/diameter ratio of specimen
Mc - constrained modulus of deformation
Mr - modulus of resilience
Mt - modulus of toughness
n - number in group or sample
PI - weight of oven-dry specimen
P2 = weight of saturated specimen

r - coefficient of correlation
r2 - coefficient of determination
R - Schmidt hammer hardness
s - sample standard deviation
s2 - sample variance
Sy.x - sample standard deviation from regression, or standard error

of estimate (Y on X)
s5 - standard deviation of the means, or standard error
xlv
Sh - Shore sclcroscope hardness
t a "Student's" t-distrib.tion
T a point-load tensile strength
USOM - U. S. Bureau of Mines
US DR - U, S. Cureau of RoclI.:7tfon
VZ - coefficfcnt of variation
Vp = sonic pulse velocity or dilatational wzve velocity in an
unboundc medium (P-w~ve)
Vs - distortional or shear wave velocity (S-wave)
R - sample mean of X
X - Independent variable (abscissa)
- sample mean of Y
Y - dependent variable (ordinate)
Yest. - value of Y estimated from regression
Z - standard normal distribution
1 - level of significance
Y - unit weight
Ya = unit weight (air-dry)

dcr/dc - tangent modulus
c - strain
Cf - strain at failure
tan e - slope of linear stress-strain curve
P - mean of population
V - Poisson's ratio
i - 3.14159
i - angle of internal friction
P - mass density (2)
aa(Y) - yield stress

aa(ult.) - uniaxial compressive strength (ultimate)
a/f - secant modulus
a. - stress (in engineering mechanics)
or - standard deviation of population (in statistics)
E - st'rmation
xv
AEI-I-TR-65-116
This page intentionally left blank.
xvi
SECTIC; I
I NTRCDUCT IQOI
1. NATURE OF TIVE PROILFM
Traditionally, civil and military engineers have been involved with

the mechanics of rock bchavior on projects in rock such as dccp foundations
of dams, buildings, and bridges; dccp rock cuts fo." highway, railway, and
canal construction; and rock tunnels for power supply or vehicular use.
Mining engineers also have been keenly interested in the behavior of the
rock surrounding their operations for extraction of materials from beneath
the earth's surface. In recent years, the scope of projects concerned with
rock engineering has broadened extensively with ".he advent of underground
power plants, missile launch and control facilities, and other types of
r,;ilitary protective construction. These latter projects involve dynamic
or shock loading at high intensities of pressure as well as the normal
static loading.
Many early projects were successfully completed by applying procedures

and rule-of-thumb techniques gained from experience; however, this is not
possible with the current array of projects. If the engineering behavior
of rock under superimposed stresses is to be rationally predicted, knowledge
concerning the geological discontinuities of the rock mass, physical proper-
ties of the geological materials, and principles of mechanics. should be
ut ilized.
Physical discontinuities are present in all rock masse. in the form

of planes or surfaces separating intact blocks of rock. Geologically,
these discontinuities are recognized as joints, faults, bedding planes, or
cleavage planes. The strength arid deformation characteristics of a rock
mass are both influenced by:
i) the physical properties of the intact blocks of rock, and

ii) the number and nature of the discontinuities bounding the
individual blocks.
The problem is, therefore, two-fold. In order to develop, ultimately, a

meaningful system for evaluating in-place rock behavior at a potential con-
struction site, it is logical to investigate, initially' the properties of
the intact rock materials. Because of the discontinuous nature of a rock
mass appropriate reduction factors should then be determined for applica-
tion to the "upper limits" defined by the intact rock.
The present investigation is not concerned with the behavior and

classification of the in-situ rock mass with its inherent discontinuities.
Rather, it is limited to the ficst phase of the problem -- the physical
properties of the intact rock. Intact rock is defined herein as a poly-
crystalline sol id., consisting of a natural aggregate of minerals, the
properties of which depend upon:
i) the physical properties of the constituents, and

Ii) the type of bonding of these constituents to one another.
Ii ii ii i i
Rock corcs c, toinrd in e;,ploratcry drillinrg are normally the first
physical evid-nc- s frc;;i which the engine-r r'V;y get an indication nf the
nature of the substurfacc rock. The cores contain the intact rock materials
and also shcei cvidcrnce of soe-,a of the discontinuities. Consequently, proper
classification of thcse cores is an important step toward an understanding
of the behavior to be expected. both during and after construction. Unfor-
tunately, the qvcrl itative descriptions presently used (e.g., on boring logs
in design, and in construction specifications for rock engineering projects,
permit too bro-d an intcrpretation for universal engineering usage.
The identification and classification of the products of nature con-

stitute an artificial procedure, because these materials are infinitely varied
and do not lend themselves to separation into distinct categories (Peck,
Hanson, and Thornburn, 1953). As a result, various arbitrary systems of
classification have been developed, each with a particular purpose or appli-
cation to a specific locale. As attempts are made to refine any of these
systems, the system inevitably beccones more complicated and u;zimately more
cumbersome so as to defeat the purpose for which it was intended. To avoid
this difficulty, it is desirable to make use of relatively simple systems of
classification with small numbers of categories. Detailed information con-
cerning any given rock (or soil) can be sunmnarized by stating numerical
results, known as index .. operties of certain physical classification tests.
If the classification tests are properly chosen, rock (or soil) materials
having similar index properties, regardless of their geologic origin, are
likely to exhibit similar engineerin9 behavior (Peck, Hanson and Thornburn,
1953).
The importance and usefulness of index properties for soil have
already been well established in the field of soil ri)echanics and foundation
engineering. The desirability of similar procedures for rock is repeatedly
demonstrated in rock engineering problems. In order for an index property
to be useful, it must have the following three general characteristics,
whether it is used to describe soil, rock, or other engineering materials
(Pomeroy, 1957; Deere, 1963a):
i) It must be an index of a material property which is used

by an engineer in solving a design problem.
ii) The test to determine the property must be simple, inex-
pensive, and rapidly performed.
iii) The test results must be reproducible, within certain
limits, by different operators in different locations
using standardized equipment and procedures.
2. OBJECTIVES OF INVESTIGATION
The objective of this research was to develop an engineering classi-

fication system for intact rock, which has as its basis index properties
that are related to important physical properties of the rock. An experi-
mental investigation was conducted in order to determine the physical and
elastic properties of intact core specimens, representative of rock types
most commonly en:ountered in civil engineering projects in the United States.
Relationships among these various properti- iere studied, and the most
significant correlations were used in developing the desired indices.
2
Because of its widespread arplicatiorn in mztals, and the promise it
has shown for other investigetors in rock, h6rdness in its various forms
was of primary interest for possible use as en index property. The proper-
ties of toughness, resilience, strength, and elasticity, are all involved
to some degree in hareness tests. The latter two, strength and elasticity
(or deformability) , are the material properties which are of primary concern
in the majority of civil engineering projects. Therefore, IV.th the index
tests and the classification systcmn are directed towiard a relationship with
these properties.
3. SCOPE OF STUDY
In the following section, a -eview of existinq systems of rock classi-

fication il presented. These consist of: (1) the most common, geological
classifications, (2) systems which give consideration to classification on
the basis of the physical properties of the rock substance, and (3) in situ
class if icat ions.
Section Three deals with the concepts of hardness and their app ica-
tion to rocks. The three basic types of hardness and their methods of
measurement are presented, along with some of the relationships previously
found between hardness and other properties. The relationship of rock
hardness to drilling is also discussed.
In Section Four, a description of the 28 rock groups studied in this

investigation is given. The following 13 different geological rock types
are in,.luded: basalt, diabase, dolcmite, gneiss, granite, limestone,
marble, quartzite, rock salt, sandstone, schist, siltstone, and tuff. A
detailed explanation of sample preparation is presented, in addition to a
description of all tests performed. NX-size, intact core specimens, with
L/D ratios of 2:1, were subjected to the following laboratory tests: unit
weight, Shore (scleroscope) hardness, Schmidt hammer hardness, abrasion
hardness, absorption, sonic velocity stress-strain under cyclic loading,
stress-strain to failure, and point-load tensile strengtK. A summary of
all test data is given in tabular form at the end of Section Four.
The test results are discussed and interpreted in Sect ic Five.

Data plots are presented of those properties which show the most signifi-
cant degree of correlation with other properties, or are of general
interest. Least-square regression lines are shown on each plot, together
with the functional relationship, standard error of estimate, ead correla-
tion coeificient. A comparison of static and dynamic properties of rock,
as observed in this investigation, is also discussed.
Section Six presents the correlations from which the index proper-
ties are defined, and proposes a system for the engineering classification
of rock. This system is used to provide qual itative descriptions in terms
of compressive strength and Youngis modulus for intact specimens of rock.
Classification is determined from numerical data obtained from index pro-
perty tests, as described herein, or from actual laboratory measurements of
strength and modulus. Estimates of the general range of strength and
3
modulus vzitucs for various rock types can also be obtairnd from the charts
In Sect icon S ix.
In tlkz lest section, the sumciary and conclusions are presented.

Future rcsezrch vwhich appears warranted as a result of this investigation
is also suo-csted.
SECTION 2
EXISTIrWG SYSTEMIS OF RCIC CLA$SIFICATION
I. GENERAL
One of the primary purposes for rock classification in subsurface engi-

neering and construction projects is to provide an effective means of communi-
cation among the engineering geologist, laboratory and design engineers, field
engineer, and contractor. The records (f underground explorations shown on
contract drawings furnish the basic data for contractors' bids. and during
performance of the work form the basis from which all changed conditions must
be evaluated. The responsibility for presenting an accurate description of
the materials to be encountered rests with the design engineer and the engi-
neering geologist. It is desirable that the characteristics of the rock which
will influence the construction methods be clearly conveyed to the contractor.
For the engineer, the appropriate classification of both the rock

materials and rock mass would greatly assist in making the initial appraisal
of the project, and in pointing to the areas where additional engineering
information must be cbtained for use in arriving at the final solution to
the problem (Coates, 1964). Because of the important role of experience,
especially in subsurface engineering projects, the individual experiences of
all engineers should be summarized into a body of knowledge that can be
readily assimilated; otherwise their value is lost to the profession. Proper
classification of rocks (and other subsurface materials) into groups, within
each of which the significant engineering properties are somewhat similar, is
thus an important step in connection with any foundation project (Peck,
Hanson, and Thornburn, 1953).
This section is concerned with a description of some of the systems

which have been developed for rock classification. Geological systems are,
in general, most widely used. Other systems have been developed for particu-
lar purposes, but are often limited in their application. Some systems are
concerned with the properties of the rock material only, while others are
more concerned with the continuity, size, and arrangement of the intact rock
blocks (or slabs) within the rock mass. Still others give consideration to
both of these characteristics.
The following part of this section deals with geological classifica-

tion. The next part is concerned with systems which given consideration to
description and classification on the basis of the physical properties of the
rock substance. The last part discusses rock classification systems which
are primarily concerned with tlie in-situ character of the rock mass.
2. GEOLOGICAL CLASSIFICATION
In its broadest geological sense, a rock is any naturally-formed

aggregate or mass of mineral matter, whether or not coherent, constituting
an essential and appreciable part of the earth's crust (AGI Glossary, 1962).
5
To the Ic:y' n arid -the eniineer (and, in truth, to most geologists), the term
"rock" refers to any hard, sol id nwatter derived from the earth (Stokes and
Varnes, iI5).
A slnoIe scheme for clasrifying rocks thet Is both cc.2elte and

practical is presently rnot avaIla.le. Rocks ar, cl3s;ificd c6.2., ically,
petrogrCAhically, or goi-,ctically, ei;aonding on the purrose of the classi-
fication. Each basis has its own merits but none can fully cc-:nine the
advantages of all (NIuang, lf,2). A primary division of rocks into three
groups according to their genesis or mode of origin Is almost universal in
usage. I.nLois rocks are the primary rocks formed by cooling of molten
magmas, oi- by recrystallization of older rocks under heat and pressure
great enough to render them fluid. If magmas cool beneath the surface, they
form intrusive rocks; marmas reaching the surface form extrusive rocks.
Sedimentary rocks are the products of deposition of plant and animal
:1 remains (organic), and of m3terials formed by chemical decomposition
(chemical), and of physical disintegration of igneous, sedimentary, or
metamorphic rocks (clastic). Ietamorphic rocks are those produced by
Internal processes of heat, pressure, or permeation by other substances,
acting on preexisting rocks of any kind. Metamorphic rocks are usually
referred to as foliated or nonfoliated.
Petrographically, the most important properties utilized for identi-

fication and classification are texture, structure, and mineralogical
composition. Although there is no general agreement as to which physical
features of rocks shall be included as textures, and wh;ch are to be
regarded as structures, the term, fabric, is used to irnc!ude data from
both categories. Spock (1953) refers to texture as the size and shape of
rock constituents, together with variation in these properties. Structure
covers the distribution and grouping of minerals and also the immediate
effects of their arrangement, such as foliation and bedding. Mineral
composition is the fundamental basis in geological classification of rocks,
because the actual units (minerals) of which rocks consist are used
(Hunag, 1962).
Chemical classification is primarily useful for comparison of rocks

on the basis of chemical data in which each oxide is presented as a weight
percent. However, it is impossible to determine the character of a rock
from a chemical analysis alone, since rocks of closely related composition
may differ in origin, as well as in texture and mineralogy (Spock, 1953).
Detailed geological classifications are readily available in various

texts deal ing with rocks (Grout, 1940; Spock, 1953; Huang, 1962). Travis
(1955) presents in compact form the generally accepted conventions for
naming rocks on the basis of mineralogy and texture. Mielenz (1961) gives
a mineralgic and textural classification of igneous rocks only.
Unfortunately, the above geological grouping and classification pro-

cedures give little or no information concerning the engineering capabili-
ties of the rock. For example, some limestones have compressive strengths
of 6,000 psi, whereas others have strengths of 36,000 psi. Or for example,
the term "granite"may mean a hard, coherent rock. However, it is known
6
that granite can be extremaly variable from location to location, depending
upon the various wethering conditions to which it has bezn subjected. In
some tropical cl imatcs, granites are wrathcrcd to such a chroc that thWy
can be excavated by hand with an ordinory shovel. Srndstor::. lso, m, fall
under class if ictt iors susch as oorly cvcr:-.-ntod t:" to a.'tI 1ly cI.Iyd
sand" for the szrn,! rock, dzondinn u.clin thL backorounJ ;:J c;;.'d r ioncc of tkc
geologist assigning the classification. If the former of tVe two classifi-
cations were used in the contract documents, the contractor would be prepared
for excavation by blasting--and probably no bracing. If the latter classifi-
cation were used, the contractor would normally be prepared for excavation
with mechanical power equipment and extensive bracing. T1e cost and pro-
cedures for either case are entirely different. These and other examples
clearly illustrate the limitations of a strictly geological classification.
3. CLASSIFICATION BY PHYSICAL C T.E.R OF ROCK KATERIALS
Some of the earlier ground classification methods used in mining are

given by Harley (1926); he states that the factor that most influences all
underground operations is the physical nature of the "ground"with which the
mine operator has to deal. In the past, operations in mining were initiated
by using trial and error methods; many believed that the conditions under-
ground made this method of approach inherently necessary. Post general for
local usage were such terms as hard, medium, soft, easy-bret.king, et cetera.
Other schemes. based upon rock names., were unsatisfactory because they did
not consider the wide range of mineralogical composition. te.-ture, weathering,
and alteration, occurring in each rock type.
Harley (1926) proposed a scheme of classification which considered the

following physical characteristics:
i) unit weight,
ii) degree of hardness,
iii) degree of toughness, and
iv) occurrence of slips (or joints) in the rock.
This scheme was based upon the ft.-lbs. of energy required for drilling one
cubic inch of rock, which was correlated to a grinding resistance obtained
by a small grinding machine. The hardest rock in the drill test was assigned
a grinding resistance factor of 1.0 and a classification of A+; the softest
rock was classified as Do., and the grinding resistance factor was some
decimal part of 1.0. There is no evidence that Harley's proposed scheme
gained popular usage.
Head (1951) proposed a classification of geological rock formations

based entirely upon the relative efficiency with which the formations could
be drilled with a small rolling-cutter type of test bit. The micro-bit was
about 2 inches in diameter, and consisted of two rolling cutters (approxi-
mately 1-inch diameter) mounted on opposite ends at a slight angle to the
axis of the shaft. The test bit was designed to facilitate the replacement
of the cutters after each drillability test. The Drillability Classification
Number (DCN) was obtained by mounting the micro-bit in a lathe and measuring
7
the time interval, in seconds, required to drill 1/16-inch dcpth Into a
samrple of each forimation. Rotational speed was 110 rpm at a constant
bit thrust of 1417 pounds. By using the average result of at least two
tests, the relative drillabil ity wins establ ished for 15 rock formations
co,'ionly cc.untcrd In drill irn. The DC14 ran-nd from 1.9 for Wilcox sand-
stone to 555.7 for tosston quortzite.
To determine whether or not this classification was consistent with

results obtained from actual bits, drilling these same formations under field
conditionsP Ioad examined the performance of seven different types of bits
used in these formations in the field. He found that the rates of penetration
of the field bits fell in the same order as the rates of penetration of the
test bit (the DCN is inversely proportional to the corresponding rate of
penetration of the test bit). On the basis of these tests, Head concluded
that any rolling-cutter type of bit should drill all formations for which a
Drillability Classification Number has been established, in the same succes-
sion as the test bit, if chipping action occurs.
Hughes Tool Company, Houston, Texas, has developed an cmpirical, drill-

ability classification chart on the basis of micro-bit data from samples of
the same rock formations on which full-size bits have been run (Rollow, 1963).
The estimated drilling rate of a new bit (in the field) is obtained by enter-
ing the chart, using micro-bit drill ing rate and weight on the bit in pounds
per inch of diameter. The micro-bit is 1 1/4 inches in diameter and is of
slightly different design than the one used by Head (1951). The bit is placed
in a small, radial drill press under 200 lbs load at 55 rpm; air is used to
remove cuttings. Each test consists of drilling 3/32 inch, with drilling
time recorded for each 1/32 inch drilled. The average drilling rate and
cutter wear are calculated from the drilling tests.
The laboratory micro-bit test provides a basis for predicting rock

drillability from small samples. In oil-field work, where cores have been
checked with the micro-bit, the results have been accurate within 10%. The
results have been somewhat7 more erratic in the blast-hole industry, but
still usable. As Rollow points out, the micro-bit test is obviously valid
only to the extent that the micro-specimen is representative of the whole.
Also., the test procedure is reported to ignore several factors which affect
bit life. The major limitation of this test for universal rock classifica-
tion appears to be its specialized relation to particular types of equipment
and procedures for specific rock formations.
A scale of hardness applicable to rock material was given by the

Panama Canal Company (1947). This scale was slightly modified and used for
classification of rock during construction of the Balboa Bridge (1959).
The system is simple and is based upon the relative ease or difficulty
with which intact rock can be broken. Although qualitative in nature, it
uses terms familiar to all, providing a gradation from rock that can be
broken in the hand to rock that can only be chipped with heavy blows of the
hammer. This classification scale is shown in Table 2.1.
The U. S. Army Corns of Engineers (1961) present a 'tuide for Classi-

fication and Description of Rocks with Emphasis on Their Engineering Proper-
ties," which is essentially qualitative in nature. The basic classification
8
TABLE 2.1
SCALE OF ROCK Ha7,DN2SS
After Panama Caral Co-7.rny (1959)
R11-i Soft Slightly harder than very-hard ovcrburden, rock-

like character but crumbles or breaks easily by
hand. (Some clay-shales and uncemented sand-
stones and agglomerates.)
RH-I Medium Soft Cannot be crumbled between fingers, but can be

easily picked with light blows of the geology
hammer. (Some shales and slightly-cemented
sandstones and agglomerates.)
RH-2 Medium Hard Can be picked with moderate blows of geology

hammer. Can be cut with knife.
RH-3 Hard Cannot be picked with geology harnmer, but can be

chipped with moderate blows of the hammer.
RH-4 Very Ha-rd Chips can be broken off only with heavy blows of
the geology hammer.
includes a symbol and a name for each of 24 rock types encountered on Corps
of Engineers projects. The physical properties of the rock ar,- designated
on the boring logs, either by word descriptions or by using the appropriate
key number or numbers shown on the guide for the applicable physical
properties. There are 46 descriptive key terms which may be selected as
characterizing the following properties: bedding, lithology, hardness and
degree of cementation, texture, structure, degree of weathering, solution
and void conditions, swelling, and slaking. Of the nine properties listed,
only three are not directly concerned with material properties, i.e.,
bedding, I ithology, and structure. These three are characteristics of the
rock mass.
A simple, quantitative procedure which can be correlated with in-

situ and laboratory test results of the strength and compressibility of'
rock is described by Hamrol (1961). In this procedure, the index of
alteration (a measure of the short-term water absorption of rock samples)
is determined for a large number of samples over an entire site and, by
using the established correlations, the foundation area can be mapped in
terms of a particular engineering property (Rocha, 1964). This technique
appears to be applicable primarily In differentiating zones of weathered and
altered rock.
9
Deere (19$Tb) gives c,,,sis to those ccological fcaturcs which can
be observed in rock cores, and which appar to be significant in rock engi-
neering. The significant fecturcs include those whlch hove a direct bearing
on the ho.. .ene lty of the rock l,-ss with rcpc.ct to (1) var ict ions in hard-
ness, and (2) physical disccntiAities. In the: routine longing of rock
cores, it has bc.:n the custcrm to check tz;- [21rc!rdncss by nothir,,g rror than a
knife blade or geology pick. Since it is not thc sI ight diff"c
degree of hardness from one rock to anoth:er tL't is irysPortant, but rather
the large variations, a relative term for hardr -;may bo satisfactory.
In order to obtain a more meaningful nurmericcl hardneis value, Dere
(1963b) suggests the use of a small version of the S('imidt concrete test
hamm er. This procedure was recently reported by Knill and Jones (1965) for
granite cores from a dam in England. Although the impact numbers were not
shown to be directly related to any engineering property, they clearly
reflect the decrease In the effects of weathering in the rock with depth.
From an engineering point of view, Deere (1963b), also suggests that

a textural classification system eventually may prove to be of greater value
than the normal, mineralogical one. He has proposed that most rocks can be
fitted into one of the following three textural groups: interlocking, cemented,
laminated-foliated. Here, texture refers to the manner in which the consti-
tuent grains are arranged and bound together (the petrographer's fabric),
rather than reference to grain size.
Coates (1964) reviews the uses to be made of rock mechanics in engi-

neering work, and lists five characteristics or properties which he con-
siders to be most important for engineering applications. On the basis of
the five characteristics, he proposes a classification system which recog-
nizes that the rock substance (or material) has certain properties that can
be identified, but that the condition of the rock substance in-situ is also
of great importance. The three characteristics which deal with the proper-
ties of the intact rock material are listed as follows:
i) The uniaxial compressive strength is the property which
immediately indicates whether the rock substance is weak
enough, with respect to the application, to be a source of
trouble in Itself.
ii) The pre-failure information characteristics of the rock
substance Indicate whether creep of some nature could be
expected in the material itself, at stress levels less
than those required to produce failure.
Iii) The failure characteristics of the rock substance, i.e.,
brittle, or plastic, should influence the safety factor
that Is used in design' as well as precautions to be
token during construction.
The classification system proposed by Coates Is given in Table 2.2. It will

be noted that his system is based on both the material properties (Items 1-3)
and the in-situ geological discontinuities (Items 4 and 5).
10
TABLE 2.2
CLASS1FICATIGON OF ROCKS F0 ROCK 1ZCEAINICS
After Coates (1'1`1)
I. Uni axal cc'ressic,; strcrr,,.thi of thL rcck substancc

a. Vl-3k (less than 51C00~ pzI)
b. Strong (bctween 5,000 psi and 25,000 psi)
c. Very Strong (greater than 25,000 psi)
2. Pre-failure deformation of rock substance

a. Elastic
b. Viscous (at stress of 50% of uniaxial compressive strength the
strain rate is greater than 2 microstrain per hour)
3. Failure characteristics of the rock substance

a. Brittle
b. Plastic (more than 25% of the total strain before failure is
permanent)
4. Gross homogeneity
a. Massive
b. Layered (i.e., generally including sedimentary and schistose,
as well as any other, layering effects which would produce
parallel lines of weakness)
5. Continuity of the rock substance in the formation

a. Solid (joint spacing greater than 6 ft.)
b. Blocky (joint spacing between 3 in. and 6 ft.)
c. Broken (in fragments that would pass through a 3-in. sieve)
4. CLASSIFICATION BY IN-SITU CHARACTER OF ROCK MASS
In the process of assessing the engineering behavior of rock, the

engineering geologist and the foundation engineer are faced with the major
problem of translating the significant geological ervations into engi-
.
neering terms. The information normally required by the engineer includes

data on the strength, compressibility, and permeability characteristics of
the foundation rock. In addition, the existing state of stress, the ground-
water conditions) and chemical stability of the rock may also be required.
Of these characteristics, the first three and the last are properties of
both the rock material and the in-situ rock mass. The existing state of
stress and ground-water conditions deal only with in-situ characteristics.
A classification system to be used for determining the appropriate

kind and amount of support for tunnels in rock was proposed by Terzaghi
(1946). In this system, rocks are placed in the following groups in
11
decreasing order of cor,;-,ctcncy: intact, stratified, rnrodcratcly jointed,
blocky and seamy, crushed, squcezing, and swell ing. A similar classifica-
tion is given by Talobre (l&;7), except he has six categorics, zpparently
omitting modorately jointr~d rock. Neither systcm is particularly concerned
with the properties of the rock m,:terial. On the other hand, the description
of the rock moss is fairly informative frcm a qualitative standpoint.
The U. S. Bureau o' trinos (W3'2) rcport that "thre typce of rock, its
physical and geological cheracteristics, and the stress field before mining,
determine, in general, the sizc and shcpe of underground openings that can
be used and the amnount of artificial support that must be installed." A
classification system for underground openings in rock is given, in which
in-situ rock is divided into two main divisions as follows:
a. Competent rock -- rock that will sustain an opening without

artificial support.
(1) massive-elastic (homogeneous and isotropic)
(2) bedded-elastic (homogeneous, isotropic beds with thick-
ness less than the span of the opening and little cohesion
between beds)
(3) massive-plastic (rocks which flow under low stress)
b. Incompetent rock -- rock which requires artificial support
to sustain an opening.
As stated in the USBM report, the system is only for competent rocks, which
limits its application to other than the specific purpose intended. Its use
would also be limited to engineers with considerable experience because of
its qualitative, rather than quantitative basis. It provides no numerical
definition or other means for determining whether or not a rock is competent,
or to what degree.
A rock classification chart based on joint spacing and weathering

has been developed by the Austrian school of rock mechanics (John, 1962).
This chart relates two of the important factors governing the bulk strength
of a rock mass. These factors are the spacing of the geological joints, and
the degree of decomposition (weathering) and resulting compressive strength
of the rock material. The rock material is divided into four descriptive
classifications as follows:
Type I - Sound
Type II - Moderately sound, somewhat weathered
Type III - Weak, decomposed, and weathered
Type IV - Completely decomposed
Numerical limits for rock strength are assigned to each of the above cate-
gories. A chart of this type will be useful for classification if the
description of the rock composition can be made quantitative and reprodu-
cible for all rock types (irrespective of the project site), and can be
related to the engineering properties of the rock.
Estimation of the degree of jointing (or soundness) in a rock mass may

be obtained by use of seismic tests (Onodera, 1963). A laboratory seismic
(sonic pulse velocity) test performed on an intact rock specimen is unaffected
12
by the discontinuities present in the field. The field seismic wave, on the
other hand, will be affected by the discontinuities in the rock mass through
which it travels; the greater the degree of jointing, the lower will be the
seismic velocity. The ratio of the field seismic modulus (a square function
of the seismic velocity) to the laboratory seismic modulus can, therefore,
be used to estimate the degree of jointing. For an unjointed rock mass, the
ratio is assumed to be unity, i.e., the field velocity is the sa,-m as the
laboratory velocity (assumning no sanmpl ing disturbance and idcntical moisture
conditions). For a jointed rock mass, the ratio will be less than unity.
Onodera (1963) presents a classification which correlates the degree

of jointing and weathering with the seismic or "Jsoundness"E ratio. This
procedure includes both the properties of the rock material as well as the
discontinuities which affect the mass behavior. Descriptive geological
diagnostics are also given by Onodera in which the rock is graded from excel-
lent to bad, and assigned corresponding alphabetical symbols from A to E,
respectively. He also suggests that a similar concept may be appl ied to
rock simples from a given site in which an "ideal' specimen is assigned the
ideal modulus as determined from a laboratory, sonic velocity test. Other
specimens with lower modulus values are downgraded accordingly.
A Rock Quality Designation based on rock hardness and fracture

frequency, as determined from rock core examination, has been developed
in an attempt to general ize the average, in-situ rock qual ity (Deere, 1964).
In this method, the length of sound, relatively unfractured core per length
of core-hole is determined, by considering only core which has a certain
minirum hardness, and which is longer than 4 inches. The rock is then rated
in designations of Excellernt, Good, Fair, Poor, in descending order of rock
qual ity, based upon the percentage of sound core obtained from the boring.
A rock which is soft or weathered, a closely jointed rock, or a rock from
which the core recovery is low, would fall into the Poor to Fair category.
A quantitdtive method for evaluat'ng rock hardness is desirable in order
for this scheme to be rel iably used by inexperienced personnel.
Because of the complex relationships which exist between various rock

types at a dam-site in the Sudan, four separate grades of rock condition
were utilized by Knill and Jones (19G5) as follows:
Grade I : Fresh rock

Grade II : Sl ightly weathered rock
Grade III : Highly or moderately weathered rock
Grade IV : Completely weathered rock
They report that by this means it was possible to group togethcr rocks of
different types which have similar engineering properties. Slopes cut in
Grade IV rock disintegrate under wet conditions to angles between 25* and 30%
This rock can be excavated mechanically and, generally, can be dug manually.
Grade III rock was usually excavated by blasting; slopes up to 10 meters
high were stable, apart from minor spalling. Rock in both Grades I and II
was excavated by blasting and, in general, the sounder rock required less
explosive than that needed to excavate a similar volume of Grade II and III
rock. The four grades of this system are similar to the types proposed by
John (1962). For exploratory purposes, the method for assigning a rock to
13
any particular grcde, above, appears to be one based upon expcrience, thus
restricting Its universal applicability.
The difficulties to be encuntered in obtaining a universal engi-

neering classification system for rock cannot be over-emphasized. Geological
classification systcms, while providing a basis for naming and describing
rocks, contain little Information or relationship with the rock properties
important to their mzchanical b3hIrvior. The specific purpose for which a
classification system is developcd obviously plays an important role in
determining whather the ermphasis is placed on the material properties of the
rock substance or the continuity of the substance in the rock mass. It is
important that the process of LnIysis of a particular problem be made from
the point of view of the proposed service of the rock. However, the basic
data from which the engineering evaluations are to be made should be an
inherent property of the rock (both material and mass), and should be deter-
mined from standardized procedures. The same rock should have the same
classification regardless of how it is being used (Coates, 1964). For each
type of information, a systematic collection of the data, correlated with
engineering experience through a consistent system of classification, can
lead to improved methods for predicting rock behavior.
It is clearly seen that a universally acceptable engineering classi-

fication system for rock should indicate both the properties of the rock
material and also the in-situ characteristics of the rock mass. Because
the rock-mass cha.acteristics involve both the material properties and the
geological discontinuities, a meaningful system for in-situ classification
is not likely to be developed without also evaluating the behavior of the
rock substance. Thus, the investigation described herein is limited to the
first phase of Zne problem -- the physical properties of the intact rock.
This study is not concerned with the behavior and classification of the
in-situ rock mass with its inherent discontinuities.
The next section deals with one of the more important physical
characteristics of the intact rock material. The concepts of hardness
and their application-to rocks are presented. The three basic types of
hardness (i.e., abrasion, indentation, and rebound) and their methods of
measurement are discussed, along with some of the relationships previously
observed between hardness and other properties.
14
SECTION 3
HARDNESS OF ROCKS
1. GENERAL
For the past 65 years or more, engineers and metallurgists have been
making hardness measurements of metals as a means for assessing their general
mechanical properties. Because of the apparent connection between hardness
and other properties, simple hardness measurements on the surface serve as an
indication of the bulk properties of the material. This application has been
extremely valuable in metallurgy; its usefulness for rock materials also shows
considerable promise.
Hardness is one of the most investigated properties of materials,

and yet it is also cie of the most complex to understand. As has been
pointed out by many investigators (Obert et al. 1946), hardness is an
elusive property; the concept of hardness does not lend itself to exact
definition in terms of the customary physical units. A numerical value
of hardness is as much a function of the kind of test used as it is a
material property.
The term hardness, when used as a technological property of materials,

is primarily associated with the surface (Richards, 1961). Mineralogists were
among the first to be concerned with hardness because of its close association
with the resistance of prccious stones to surface scratchinq and abrasion.
The hardness cf a mineral is defined by Dana (1932) -as"'theresistance which
a smooth surface offers to abrasion." This connotation was carried over to
metals with the development of alloys that could be hardened for high
abrasion resistance. If the material is uniform in composition and structure,
the surface constitutes a layer only a few atoms thick and loses wuch of its
importance. Thus, in the usual means of measuring hardness, any such thin-
surface layer is penetrated and the hardness of the bulk of the material is
measured.
In general, hardness implies the resistance to deformation. Richards

(1961) defines technological hardness as "the resistance of a material to
permanent deformation of its surface." When metals are deformed or indented
the deformation is predominantly outside the elastic range and often involves
considerable plastic or permanent deformation. However, in dynamic hardness
measurements, the elastic properties may be as important as the plastic
properties. In all types of hardness, the properties of toughness,
resilience, strength, and elasticity are involved to some degree.
Rock hardness is considered to consist of the resistance of rock to

the displa':ement of surface particles by a tangential abrasive force, as
well as its resistance to a normal, penetrating force, whether static or
dynamic. Rock hardness depends on substantially the same factors as tough-
ness. Toughness is governed by the efficiency of the matrix in binding
together the grains or minerals comprising the bulk of the rock. In addition,
rock toughness is a function of the grain strength or mineral toughness. The
15
1i~ f ,.' ,, , , '% tJAL' _.,~bf.Wt !f
, ,r. ' - ... ' -- -lx
toughest rocks ccr,;rise those having strong minerals cf'cbeddcd in a strong

matrix or ccm,--nt (Shcpherd, 9I5l). Coth hardnesb and tov'hncss d2pond on
the type of binding forces betw:en atc:.;s, ions, or riilecules and increase,
like strength, with the rmonitud-, of t~ies forces. Coth prcertics are
also closely related to the yield strcngth of the mtcrial (Jastrzcbski,
2. TYPES OF w2ES'$S APD THEI R rE:TH4 OF IAMe24T1I
No other tests on materials approach in numbers the tests made under

the name of "hardness tests" as the criteria whereby materials .may be
classified or selected for special purposes (Williams, 1942). Hardness
measurements usually fall into three mnin categories: abrasion or scratch
hardness, indentation hardnessP and rebound or dynamic hardness. For
laboratory study of rocks, some form of abrasion hardness and the sclero-
scope rebound hardness have been investigated to the greatest extent.
a. Abrasion hardness
(I) Mohs' scale
Since 1824, Mohs' scale of hardness, although only relative in

nature, has been accepted universally as the method of measuring mineral,
and therefore, rock hardness. By use of Mohs' scale, the relative hardness
of a mineral can be determined. This scale consists of a group of minerals
having increasing hardness values from I to 10. Each mineral can be
scratched by those that follow it, and will scratch the preceding ones in
the scale. Minerals I and 2 can be scratched by the fingernail; those 3
to 5 by the point of a knife, 6 cannot be marked by a knife but can by a
steel file. Number 7 can scratch glass, 8 scratches the file, while 9is
only scratched by the diamond, which is assigned the highest hardness
number of 10.
Tabor (1954) showed totat the Mohs' hardness scale gives scratch
hardness values which correspond to fairly well-defined indentation hard-
ness values. Each increment on the Mohs' scale corresponds to a 60%
increase in indentation hardness, excluding diamond. which is anomalous.
This regularity in behavior suggests that Rohs did not simply choose "ten
common minerals arranged in order of increasing hardness," but experimented
until he had satisfied himself that he had obtained "equality of intervals."
Table 3.1 lists the minerals representing the ten steps of the.
scale, together with their crystal structure (Zwikker, 1954).
(2) Rock hardness based on mineral composition
Mohs' scale is obviously applicable only to minerals having

homogeneity. Each rock consists of several minerals; scratch tests based
on this scale are useless for determining the hardness values of hetero-
geneous bodies. Shepherd (1950) cites a method for measurement of composite
rock hardness, based on mineral hardness as obtained by the Mohs' scale.
16
TABLE 3.1
MOHS' HARDNESS SCALE AND CRYSTAL STRUCTURE
After Zwikker (IS-)
Mohs' I-ardness
Number, Hij Mineral Formula Crystal Structure
I Talc 3MgO.4SiO 2 .H2 0 Layer
2 Gypsum CaSO 4 . 212 0 Layer
3 Calcite CaCO3 Layer
4 Fluorite CaF 2 Ionic
5 Apatite CaF 2 .3Ca 3 P2 0 8 Ionic
6 Feldspar K2 0 .Al 203. 6SiO2 Mixed ionic-covalent
7 Quartz SiO2 Mixed ionic-covalent
8 TopaL (AIF) 2SiO4 Mixed ionic-covalent
9 Corundum Al.203 Covalent
10 Diamond C Covalent
This is done by multiplying the percentage of each mineral in a rock by

the Mohs' mineral hardness and dividing by 100 as follows:
H = S xM (3.1)
100
in which H = hardness of the rock
S - percentage of mineral present
M - Mohs' number for the mineral
Thus, in a rock containing 74% quartz (7), 6% feldspar (6). and 20% mica (3),
the rock hardness would be 6.14. There are obvious disadvantages attached
to this method as a difficulty lies in obtaining 3 true estimation of the
mineral content, and errors in measuring hard-mineral content may seriously
affect the final results. Moreover, the type and strength of the mineral
bonds, which are very important in determining aggregate hardness, are not
considered in the procedure.
(3) Dorry abrasion test
A standard test for determining rock hardness, which was developed

by the Office of Public Roads and Rural Engineering, U. S. Department of
17
Agriculture. is the Dorry test (Jackson, lPl6). The Dorry hardness test is
perforvzd on a cyl indrical s ;elI of rock, 25 mm. in diameter, which is held
against a revolving, cast-steel disk under a pressure of 250 gins. per sq. cm.
(total load 1,250 gs.). Standard crushed quartz, sized betvween 30- and 40-
mesh scrcens., is fed upon the revolving disk. Each end of the specimrn is
worn e,,::,:y in inverse ratio to its hardness. The loss in w.,icht, obtained by
everaging both ends after 1,000 revolutions of the disk, is an inde,.x of the
hardricss of the spoclr:n. In order to cc.Lcpare the results with other
abrasion tests, the hardness coefficient is calculated by subtracting one-
third of the loss in veight froa the constant, 20. Hardness values from
this test compare remarkably well with those obtained on the basis of
mineral composition (Gyss and Davis, 1927).
(4) Deval abrasion test
This apparatus consists of one or more hollow, cast-iron cylindrical

buckets, 20 cm. in diameter, and 34 cm. long. These closed buckets are mounted
together in a frame which supports them at an angle of 30 degrees to a hori-
zontal axis of rotation. The test sample usually consists of about 50 pieces
of broken rock weighing 5 kg. After revolving the cylinders 10,000 times at
30 rpm, the amount of material finer than 1/16 in. is obtained and expressed
as a percentage of the original weight of the sample. The French coefficient
of wear is likewise used in reporting results and is calculated by dividing
the number 40 by the percentage of wear. The standard method of test for
abrasion of rock by the Deval machine is given by ASTM Designation: D 2-33.
(5) Other abrasion tests
(a) Burbank (1955) describes a device for determining the relative

abrasiveness of rocks, minerals, and ores. It cc.nsists of a steel, single-
paddle impactor turning on a shaft at 632 rpm. The paddle strikes a column
of rock particles (4 lbs. per test) failing away from an outer, slowly
revolving, 13-inch-diameter drum. The loss in weight of the paddle, during
controlled test conditions, represents the abrasive action of the rock.
(b) A tentative method of test for abrasion resistance of concrete

is given by ASTM Designation: C 418-64T. This test consists of subjecting
the concrete to the impingement of air-driven, silica sand.
Wc) Tentative methods of test for resistance to abrasion of small-

and large-size aggregates by use of the Los Angeles Machine are given by ASTM
Designations: C 131-64T and C 535-64T, respectively.
(d) A tentative method of test for relative resistance to wear of

unglazed ceramic tile by the Taber Abraser is given by ASTM Designation:
C 501-62T.
(e) A standard method of test for resistance of transparent plas-

tics to surface abrasion is given in ASTM Designation: D 1044-56. This
involves the measurement of the optical effects produced on transparent
plastics by the Taber Abraser.
(f) A standard method of test for plastic materials to withstand

mechanical abrasive action such as rubbing, scraping or erosion, is given in
ASTM Designation: D 1242-56.
18
(g) A tentative test method for Bierbaum scratch hardness of
plastic materials is given in ASTH Dosignation: D 1526-58T. A known weight
is applied directly above an accurately-ground, diamond point, mounted on
the end of a leaf spring. The speciri,:n is moved laterally causing the
point to cut a groove in the surface. The index of hardness is calculated
from the weight appl icd end th- r.:asurcd width of the groove.
b. Indentnt ion h)r cn. F..
Since about 1900, a wide variety of penetrator sizes and loading

devices have been used for the static indentation hardness of metals, in
which the depth, area. or diameter, of an impression under a known load is
measured. Summar;es of these and comlparisons of several of the standard
tests are given by Devries (1911), Lea (1936), Williams (1942), and Tabor
(1951). Essentiallv, three penetrator shapes are used in all of these
tests; indenters are either spherical, conical, or pyramidal. Spherical
indenters are used in the Brinell and Rockwell tests; conical indenters are
used in the Rockwell and Ludwik tests; and pyramidal indenters are used in
the Knoop ard Vickers tests. Loading varies from as little as 7kg. in the
Vickers test (Brace, 1960) tc as much as 3,000 kg. in the Brinell test
(Tabor, 1951). A description of the standard method of test for Brinell
hardness of metallic materials is given by ASTM Designation: E 10-65.
Standard methods of test for Rockwell hardness and Rockwell superficial
hardness of metallic materials are given by ASTM Designation: E 18-65.
The standard method of test for diamond pyramid hardness of metallic
materials is given by ASTM1 Designation: E 92-65.
More recently, indentation hardness tests have been performed on

minerals and rock materials. Knoop, Peters, and Emerson (1939) determined
the Knoop hardness numbers for the standard minerals of Mohsa scale. The
hardness ratio of diamond and quartz on this basis is 8:1 rather than 10:7
as given by Mohs' scale. King and Tabor (1954) compared hardness with
compress iye strength of hal ite. Brace (1960) reports indentation tests on
polycrystalline, monomineralic. isotropic samples of four rock types using
a V'ckers indenter, with loads from 7 to 75 kg. Kraatz (1964) performed
Rockwell hardness tests on samples of 24 different rocks from the present
study.
Another type of indentation tester is the durometer. This is

described in the standard method of test for indentation hardness of plastics,
ASTM Designation: D 1706-61.
c. Rebound or dynamic hardness
Dynamic hardness of a material may be defined, by analogy with static

hardness, as the resistance to local indentation when the indentation is pro-
duced by a rapidly moving indenter. In most practical methods, the indenter
is allowed to fall under gravity onto the surface. It rebounds to a certain
height and leaves an indentation in the surface.
A review of the literature on rebound-hardness testing of materials,

indicates that several methods have been developed and put to various applica-
tions. Greaves (1909) describes the Shore scleroscope, in which the relative
19
height of rebound of a dlE.ond-tlpped her-.er fall ing freely on a horizontal,
plane surface, is a nmeasure of the hardness of the matcrial. Griffith (1937)
end 01ert ot al. (Slt;) used the sclroscop3 in obtaining hardness values in
their studies on the physical propertics of rock. U!olansky (1I04"C) reports use
of the sclcroscope in GCrmnny in an attempt to obtain rcliable and ccr;arable
values for the herdn.ess of rock, prticularly with respce:ct to drillability and
workability. Shepherd (010) used the Shore sclerosccc-'e for rock-hardness
measureTur, ts in stu,.dy ing physical prc,ertics and drill ebillity of rock.
Vluerkc.r (1553), in plotting data fro: the U. S. Bureau of tMines for a group
of more than 100 rocks, gives thrce equations for describing the relation of
compressive strength to Shore hardness. tie further suggests that since the
scleroscope test is fast and inexpensive, it could be useful for determining
strength and other physical data, outside of the testing laboratory. Kapadia
(1951) correlated scleroscope hardness with elastic and strength characteris-
tics of a limited number of rock types. Gilbert (1954) determined the sclero-
scope hardness of single crystals of 1Mohs scale mineral, and found an approxi-
mate, linear relationship, in the range of two to seven, inclusive. Harvey
(1963) describes a specially-designed instrument for determining the relative
impact resistance of limestones and dolomites of various textures. He con-
cluded that the resistance to impact for these rocks is controlled by the
porosity and size of crystals composing the rock.
3. TOUGHNESS AND RESILIENCE
Because of the apparent connection between hardness, toughness,

resilience, strength, and elasticity, it is appropriate to discuss briefly
the properties of toughness and resilience.
a. Toughness
Toughness primarily reflects the ability of a material to absorb energy

during plastic deformation. In a static test, this energy is measured by the
area under the stress-strain curve, which represents the work required to
fracture the test specimen. The specific property, modulus of toughness, is
the maximum amount of energy a unit volume of the material can absorb without
fracture; it is expressed in inch-pounds per cubic inch. The modulus of
toughness, Mt, can be estimated, for materials that indicate a parabolic-
shaped stress-strain curve, i.e., cast iron, concrete, et cetera, by mult;-
plying two-thirds of the ultimate strength, ca(ult.), by the strain at
failure, Ef, as follows:
Mt - 2 a (ult.) x ef (3.2)
It can be seen that materials of high toughness must have high strength and
high ductility. Brittle materials usually have low toughness since they show
only small plastic deformation before fracture (Jastrzebski, 1959).
A simple method for measuring the impact toughness of rock is

embodied In the Page impact machine. The impact test is made on carefully
prepared cylinders of intact rock, 25 mm. high by 25 mnm. in diameter. A
weight of 2 kg. Is permitted to fall vertically between parallel guides,
upon a spherical-ended plunger weighing 1 kg., which rests in contact with
the specimen. The height of the first blow is I cm. and each successive
20
blow thereafter is increased in hcight by 1 cm. The height in centimeters
of the blow at failure is recorded as the toughness of the material. The
standard method of test for toughness of rock by this method is given in
ASTH Designation: D 3-18. Obert et al. (19146) uscd ihis procedure, with
slight modification, in standardizing tests of mine rock.
From data obtained by the U. S. Dcpartn;2nt of Agriculture with the::

Page impact machine, Gyss and Davis (1527) have ccw-,piled a table co.Fyerring
relative values of toughness with relative wear (i.e., relative resistance to
wear). These data are plotted in Figure ^.. The property of rock toughness
is also seen to be related to rebound hardness. This is indicated by the
relationship in Figure 3.2 between abrasion and Shore hardness, since both
Figures 3.1 and 3.2 have a measure of abrasion hardness plotted as the
ordinate.
b. Resil ience
"11he capacity of a material to absorb energy in the elastic range is

designated as its resilience" (Jastrzebski, 1959). The modulus of resilience
is equal to the area under the elastic portion of the stress-strain curve and
is defined as the strain energy absorbed per unit volume when the material is
stressed to its proportional limit. The two properties, strength (here yield
strength) and elasticity, which seem to influence resistance to compression
most, both occur in the following formula for modulus of resilience:
Hr 2E (3.3)
in which M is the modulus of resilience in inch-pounds per cubic inch, and

a(y) is t~e yield strength. This equation indicates that for a high modulus
of resilience the material should have high yield strength and low elastic
modulus.
The term resilience should not be confused with modulus of resilience.

Within the proportional limit, the resilience is equal to the external work
put into a material during deformation. Thus the total resilience of a
material is the product of its volume and the modulus of resilience (Eshbach,
1952). A low resilience is desirable for good damping, and high resilience
for low internal heat generation (Richards, 1961).
The moduli of resilience in compression, of some typical rocks tested

by the U. S. Bureau of Mines (Windes, 1949), are represented graphically in
Figure 3.3. The sample of concrete of conventional mix is shown for purpose
of comparison. The values of the moduli of resilience of the various speci-
mens in the plot are:
Rock Mr (in-lb/in 3)
J -: *! ite 448.0
He,' tite ore 134.0
Amphibolite 124.5
Ma; lle 66.3
Sand'stone 62.5
Concrete 5.0
21
I) LirrzzIc' s1
for bch vcrlazz
- 0 Sc i'-tr rc2c - -
a I~.srccls
A
2.0
C,
0 0
0
0
I.2
A
I00
0 0
(Data from Gyss 8 Davis, 1927)

0 0 1 1 I I
0510.52.0 2.5 5.0
Relative Toughness (Pcge Impact Test'#
FIGURE 3.1 RELATIONSHP DETWEEN RELATIVE VALUES

OF IMPACT TOUGHiESS AIND WEAR RESISTANCE
FOR ROCK
22
o Scd~r.,:ntory roV!,s
o tgr~z..z 3 rocks3I
1 jt 4
-'-r-"~r~~s~rs48 4
h~l J
__0 i .J-a-__
o ID 0
- 00
I0
200
0 0l
0 0 0 A0a t
0 0 ) A A
C-5 _ M-4---- __
~o 00
0 A 0 a]
0 0 0 00
0
5~~b 0. -Q..
00
0 C 0 0 A
a_ _
(Doto from USCM, 944- 1956)
0 20 40 60 80 1OO 120
Shore Hordmn;:5, Sh
FIGURZE 32 RELATIONSHIP CETWEENJ SHO`ZUU ~DADR ASIO'N
AY'Z3
HARDNESS FOR ROCK(
23
Ito hlodui,, of
g 0 Josp i Ifte
Herna ftit, r
(448 )
so
~70
4,
40
30
20
to
0 -(66635
0
20--u
FIGURE
3.3Strain, incheslinCh 10-4' 2
FIGUFT3S3TEDJL
By Ri I IE
OF SO M'E TYPICAL
TueSTeD U-S. BUREsAU OF R~OCKS
Wuerer1953) WIVES (After
24
The concept of strain-energy expresscd by Mr, which is equal to the area
under each of the stress-strain diagrams herein, shows clearly the influence
of the two variables, strength and elasticity, on the failure of these parti-
cular rocks. Although the marble and the sandstone have about the same
modulus of resilience, it is evident from the generalized stress-strain
diagrams in Figure 3.3 that the former sustains a much higher load and a
smaller deformation. The overall energy in either case is theoretically
about the same. It should be noted thet only in exceptionl cases (in the
static determination of the mcdulus of elasticity of rocks) are straight-l ine
relations between stress and strain obtained (Wuerker, 1953).
4. HARDNESS AND OTHER PHYSICAL PROPERTIES
Because of the highly complex state of stress around an indentation,

there is no way of relating hardness measurements to each other, or to other
mechanical properties on the basis of theory. However, in view of their
similarities, different types of hardness numbers can be related to each
other empirically for any given material. Qualitatively, we might also expect
some sort of correlation between hardness and other properties involving
about the same amount of deformation (Richards, 1961).
a. Relationship between abrasion hardness and comnressive strength
Data obtained by the U. S. Bureau of Mines (1949-1956) are plotted in

Figure 3.4, and indicate a reasonably consistent proportional ity between
abrasion hardness and compressive strength. This plot is remarkably similar
in shape to Figure 3.2. It can be concluded from these plots that compres-
sive strength and Shore (scleroscope) hardness may be linearly related.
b. Relationship between indentation hardness, compressive strength

and modulus of deformation
Brace (1960) compared the indentation hardness and compressive strength

of limestone, marble, anhydrite, and artificial rock salt. He used a 1360
Vickers pyramidal indenter which produces an approximately hemispherical,
deformed zone beneath the surface indentation. By means of model analysis and
elastic theory, approximate stress and strain fields were determined for the
deformed zone, from which average values of stress difference, confining
pressure, and strain were determined. These three quantities zgreed fairly
well with a point on a stress-strain curve obtained from a confined compres-
sion test of a cylinder of the same material. Brace concluded that, for
nearly all incompressible isotropic materials, strength is about one-third
of the hardness (kg/mm 2 ); and that for rocks, this strength is the stress
difference which can be supported at a confining pressure of about one-sixth
of the hardness.
Kraatz (1964) performed Rockwell tests using a major load of 150 kg.
and a 1/2-inch diameter steel-ball indenter. He found an approximately
linear relationship between compressive strength and Rockwell number for
rocks having strengths of less than 12,500 psi. Similarly, a linear rela-
tionship was indicated for tangent modulus values below 6 x 106 psi. Above
these limits. the Rockwell number remains essentially constant irrespective
25
I 0 Sz ityrcs,.Is 1
0 a
25- -
a~ a
c -- 0 -a _
0 a 0 00
Q - - - -0 13-A O- ~
A 1 00
10 - 0 - D- -j - - -I -
o VoA. 0 0
aaa
2 0- - c
o 0 A 0
20
0 (Data from US8M, 1949-1956)

0102 040 50 60
Compressive Strcngth, o%(ult), psi x le~
FIGURE 3.4 RELATIONSHIP BEETWEEN AB3 SZCM.N HARlDNESS AN40 Ul-TINATE
COMPRESSIVE STRENGTH1- FCR R0, ;li' W:: X;AL CCOfESS ION
26
of increased strength and modulus. Kraatz concluded that the effective
range of the Rockwell equipment was exceeded above these limits.
c. Relationship between rebound hardness and comrpressive strenqtbh
Griffith (1937) obtained data on the physical properties of approxi-

mately 100 typical American rocks, and formulated much of these data analyti-
cally into simple working forrnules. In correlating the uniaxial com~pressive
strength data with rebound hardness, as determined by the Shore scleroscope,
Griffith suggested the following relation:
Sc - 300 H (I + ) (3.4)
in which S. = compressive strength in psi

H - scleroscope hardness in scleroscope units
The data obtained by Griffith are replotted in Figure 3.5. Griffith's
suggested "average" line is shown, although different symbols have been
used for the two parameters as follows:
%a (ult.) - 300 Sh (3,5)
in which Ua (ult) = compressive strength in psi
Sh = Shore (scleroscope) hardness in scleroscope units
The same variables have been plotted from a group of nearly 250 rock
specimens, tested by the U. S. Bureau of Mines (19/49-1956). Wuerker made
the same plot in 1953 for the first 100 of this group. Figure 3.6 indicates
that the average relation between compressive strength and Shore hardness is
described by the equation:
a (ult.) 400 Sh (3.6)
Griffith's formula forms a lower limiting range for these rocks, while the
upper range, as suggested by Wuerker (1953), is formed by the line:
% (ult.) - 500 Sh (3.7)
Many of the very high strength rocks tested by the U. S. Bureau of Mines lie
above this. and suggest (as Wuerker did) that a parabola would probably fit
the plotted values better than a straight line.
The L/D ratios of the USBM specimens were 1:1. If we assume that
Griffith's specimens were 2:1, and apply the correction recommended by ASTM
Designation: C 170-50 for relating the compressive strengths of specimens
of different L/D ratio's, the average line for the Griffith data is:
% (ult.) - 340 sh (3.8)

This is in somewhat better agreement with the USBM data for the lower Shore
values; however, equation (3.6) provides the better average relationship for
all values.
The Shore (scleroscope) hardness test is rapidl' performed, inexpen-

sive, and nondestructive. These data suggest the desirability of attempting
to establish the general application of this test, or one of similar nature,
27
.0 -o
0 ,edi.-i""'tcry
-__
rodks
. ---------
. .- 1
050 - --
- - - _ _ _ - - - -
cp
500
0.0
.13
C . O r a \,,
E 30
o so
0 A 0 0
c 00, 30 A
CPA . ~ 0 0 0
o" 0 1 %
FI wUR ( 0oto from Griffith, 1937 )
0 20 40 60 80 too 10
Shore ,iTrdna25, Sh
FIGURE 3.5 RELATIONSHIP BETWEEN SHORE HARDNESS AND ULTIMATE
COMPRESSIVE STRENGTH FOR ROCK IN UNIXIAL COMPRESSION
28
90
0 Sedirnntary roc hs
0 Igncous renc
eoo
700 0
o .60
'- I
1
CL -- - __ A,
50- ------ 0 0 0 0 o,
0 00 c~
01 0
AJ 0
40
0A 0
E0o Mb
ooooC 0 X0 . ,o,
. -
0 O0 0O AA
20 10t] ,
08A 0&
A0 0
10
Shor
0 4od,00, A
'0
0 040 60 80 100 120

Shore Hordrins, Sh
FIGURE 3.6 RELATIONSHIP CETWEEN SHORE HARDNESS AND ULTIMATE
COMPRESSIVE STRENGTH FOR ROCK IN UNWlAXIAL COMPRESSION
29
for estim.,tIng other physical prc.mrtkes of rock. The significant relation-
ships clht tt-en be used as a basis for an engincering classification for
Intact rock speclmnns.
Economiic and efficient brcnkinq of rock is governcd to a very great

extent by the successful application of rock drills, whether they consist
of the percussive or rotary type. Cecause of the many variables involved,
this subject has been scilewhat difficult to interpret by fixed rules governing
the amount of energy that must be used to drill any particular rock stratum.
So-called '"ard rocks" have been drilled easily compared with softer rocks
because of their more brittle and chipping characteristics. For example, a
hard igneous or metamorphic rock may be drilled more efficiently than a com-
pact I imestone (Shepherd, I"50).
The word hardness as used in the drilling industry, has a variety of

meanings which depend upon the type of drilling method employed. The term,
"hard rock", is generally used to describe a geological formation which for
any reason is difficult to drill, In diamond drilling, hardness is inter-
preted as resistance to abrasion. In percussion drilling, the term implies
resistance to impact or indentation; while in rotary drill ing, hardness is
considered as analogous to compressive strength. Although these may be con-
flicting variables for different types of rock, the term hardness is applied
indiscriminately by the industry for the resistance of rock to penetration by
any type of drilling technique (Mather) 1951).
The volume of blast- and bore-hole drilling done in the mining, oil,
and construction industries is exceedingly great. Hence, the problem of
increasing drilling efficiency takes on a great deal of significance.
Drilling efficiency depends on the mechanical properties of rocks, the
drilling tools, and the men who do the drilling (Protodyakonov, 1963).
Extensive research has been conducted by manufacturers of various drilling
rigs, drill steel, bits and other equipment. in order to determine the most
suitable types for drilling in certain strata. However, until the last 15
years or so, very little research has been concerned with the more fundamental
aspects of drilling and the influence of the physical characteristics of rock
on the vague property "drillability."
Some of the studies which have dealt with this problem have been
concerned with the basic processes by which energy is transmitted through
the system into the rock during percussive drilling (Hartman. 1959; Fair-
hurst., 1961a). Others discuss the mechanism and fundamentals of rock
failure under impact and percussion (Pennington, 1954; Singh and Hartman,
1961); and the mechanics of rock penetration and fragmeretation with both
percussive and rotary drilling systems (Fairhurst and Lacabanne, 1957;
Cheatham, 1958; Appl ard Gatley, 1961). Correlation of physical property
tests with force-displacement data was used by Reichmuth (1963) as the basis
of a new theory concerning the sequences of failures produced by penetration
of a wedge into brittle materials.
30
Some investigators have felt that although a kncriiledg: of the physical
properties is helpful, In general, physical prcperties are not a reliable
guide to drillabil ity. Rollow (1963) sucoested that the best way to measure
formation drlllability, with respect to rolling cutters, is to "drill with
roll Ing cutters." C. describes a I cboratory test, w!,Trcby drillability can
be pred icted on the bas is of drill ing srnl II sr:ti;l1,s w ith a I I/,I,- inch dic-:<,rtcr
micro-bit. Cy drill ing roczk-fornnttion s Ircs with
%,I this micro-b it., and
measur ing drill ing rate and tooth t'czr, est inrtes cc:a be rm.ade of large bit
performance. rirst use of the micro-bit was reported by Scott (19!,6) to
assist in correlating the drillabil ity of rock with the crushing strength.
Head (1951) used a micro-bit in establishing a drillability classification
for 15 geological formations.
Mather (1951) indicates that the principal cause of confusion in the

literature dealing with rock drilling can be attributed to varying defini-
tions of the hardness factor of rock and cutting media. Although none of
the above referenced research was directly concerned with a study of hard-
ness properties as sucb, use of the micro-bit actually involves all three
types of hardness, i.e., abrasion, indentation) and dynamic or impact.
Other investigators have felt that hardness is closely connected to

drillability and have made attempts to discover this relationship, with
varying degrees of success. Protodyakonov (1963) states that it is essential
to know the hardness, plasticity (deformability), and abrasivity, in order to
characterize the mechanical properties of rocks in drill ing. For determining
the strength of rocks, he describes a pounding technique for breaking up rock
in which an empirical strength coefficient is obtained. He estimates defor-
mation from Shore hardness tests, by use of equation (5.6) herein. Hardness
and abrasivity are evaluated by means of a specially-devised drilling
apparatus. (Protodyakonov does not make a clear distinction between his
various uses of the word, hardness.) Head (1951) conducted limited tests
using a Knoop indenter, in order to determine the relationship between the
drillability of seven different geological formations and the hardness of
specimens from these formations. He concluded that no consistent relation-
ship exists between the hardness and drillability of the tested formations,
but, rather, that drillabil ity is more related to the manner in which the
hard crystals are bound together.
Rebound hardness tests have also been investigated to determine

their application as an index of drillability. Wolansky (1949) made use
of the scleroscope in Germany for obtaining reliable and comparable values
for the hardness of rocks, particularly with respect to drillability and
workability. An interesting comparison between results obtained by th6 use
of various makes of scleroscope was made and it was concluded that the Shore
instrument gives the most reliable results. It was further concluded that
the scleroscope can be useful for indicating drillability, if comparative
values are emphasized, rather than absolute numbers. Shepherd (1950, 1951)
performed original research on the d.-illability of rocks, in order to com-
pare the resistance to penetration and to correlate the results with the
physical properties. He discusses the use and limitations of hardness tests
when applied to drilling, and concludes that scleroscope readings, unless
correctly analyzed, give no useful guide to rock drillability.
31
The next section describes the laboratory studies which were per-
formed in order to obtain specific values for comparison of physical
properties of selected rock types under controlled conditions. Among the
numerous tests conducted were two types of rebound hardness measurements,
and a new (but as yet unperfected) nethod for measuring abrasion hardness.
Indentation hardness tests for the majority of rocks from this study were
performed by Kraatz (l&!'.); therefore these tests were not repeated in the
present study.
32
SECTION 4
EXPER IMENTAL STUD IES
The cc,-1ressivc strengths, elastic properties, sonic velocities,

various types of hardness m-,2asurements, and other physical prcoprties, wore
determined for rocks from 27 localities. Thirteen geologically distinct
rock types are represented. Selection of the rocks was primarily limited to
the common types most likely to be encountered in civil engineering projects
within the United States. Other rocks -- marbles, rock salt, tuff, and
Solenhofen limestone -- were selected because of their-research interest,
either in the laboratory or in the field.
All tests reported herein were performrd on intact, cylindiical rock

specimens, macroscopically homogeneous and free from fractures, joints, and
seams. The specimens were prepared from diamond-drill cores from field
projects, or from core drilled in the laboratory from rough, quarry blocks.
All cores were standard NX-size, having a nominal diameter of 2 1/8 inches
and., for the most part, oriented at right angles to the bedding planes. The
cores were cut approximately to length with a diamond cut-off saw, and then
the ends were lapped. All test specimens had a standard length to diameter
ratio of 2:1. A total of 257 specimens were prepared and tested in this
invest igat ion.
2. DESCRIPTION OF ROCKS
A list of the rock types and the number of specimens of each type are
given in Table 4.1. These are listed in numerical order by group numbers
which were assigned and maintained for identification throughout the program.
The group number is somewhat arbitrary, but consists of assigning consecutive
integers to each geologic rock type listed alphabetically (basalt, diabase,
etc.). Rocks from different locations, but of the same type, are given a
second number following a decimal point in the group number. Thus, for exam-
ple, the Berea sandstone is numbered 10.1 and the Navajo sandstone is 10.3.
The town or city nearest to the project site or quarry from which the rock
was obtained is also given in Table 4.1.
Petrographic analyses were conducted on thin sections oriented at

right angles to the axis of a representative core specimen from each group.
The thin sections were ground to 30 microns ( 5 microns tolerance) and
covered with 22 x 40 mm. glass cover-plates. The thin sections were examined
with a Zeiss petrographic microscope. A mechanical point counter was used to
determine the mineral percentages of most of the rocks, except for some of
the essentially monomineralic sedimentary and metamorphic varieties. The
spacing of the counter, and the number of counts per section were varied at
the discretion of the operator, depending upon the Grain size, so that the
best statistical averages were obtained.
33
TACLE 4.1
LIST CF r , .!4T-
Greup Vo.
1.1 Basalt (Lo,.r Cr Ite) Pulrn;, V]shlngton 12

1,2 Basalt (Little Coose) Walle i:,;lla, Lzashingtcn 6
1.3 Basalt (John Day) Arlington, Oregon 6
2.1 Dlabase (Palisades) West Nyack, New York 12
2.2 Diabase (Coogins) Culpeper, Virginia 6
2.3 Diabase (French Creek) St. Poters, Pennsylvania 6
3.1 Dolomite (Crcota) Kasota, Minnesota 12
3.2 Dolomite (Lockport) Niagara Falls, New York 6
3.3 Dolomite (Donne Terre) Bonne Terre, Missouri 6
4.1 Gneiss (Oworshak) Orofino, Idaho 12
5.1 Granite (Pikes Peak) Colorado Springs, Colorado 6
5.2 Granite (Pikes Peak) Colorado Springs, Colorado 6
5.3 Granite (Barre) Barre, Vermont 12
6.1 Limestone (Cedford) Bedford, Indiana 12
6.2 Limestone (Ozark Tavernelle) Carthage, Missouri 12
6.3 Limestone (Solenhofen) Solenhofen, Bavaria 6
7.! Marble (Taconic White) West Rutland, Vermont 12
7.2 Marble (Cherokee) Tate, Georgia 6
7.5 Marble (Imprerial Danby) West Rutland, Vermont 13
8.1 Quartzite (Baraboo) Baraboo, Wisconsin 7
9.1 Rock Salt (Diamond Crystal) Jefferson Island, Louisiana 6
10.1 Sandstone (Derea) Amherst, Ohio 12
10.2 Sandstone (Crab Orchard) Crossville, Tennessee 9
10.3 Sandstone (flavajo) Glen Canyon, Arizona 12
li. Schist (Luther Falls) L Unknomn Origin 12
11.2 Schist (Luther Falls) ii Unknown Origin 6
13.1 Slltstone (Iackensack) Hackensack, New Jersey 12
14.1 Tuff (NTS-E Tunnel) Mercury, Nevada 12
Total 257
34
Fhoto.icrogr,-,hs of the thin scct ions vw.r'e fIlkzd on a Zeiss "'iftr,-
phot I I" c-era-rficrosccpe using Pol aroid 4 x 5 sflots (type 55 positlve/r.2,a-
t lve). The petrographic descriptions and photc-icro r,-..,hs are presentced in
Apcnd ix A.
A brief description of a typical core specir,.n, from ec.ch of the 23

rock sa,:;les studicd in this invcstis::tion, is given in the folc'aing par6o
grapThs :
1.1 B~alt(V',r :" r:nte), rul lr.:n, V'::sh irton
Dark gray, mrssivc), oract to vesicular basalt; interlocking,
crystalline texture. Core samples from dam-site on the Snake
River.
1.2 Basalt (littleGe2eL !alla Walla, Washington
Very dark gray, massive, compact to vesicular basalt; Interlock-

ing, crystalline texture. Core samples from dam-site on the
Snake River.
1.3 Basalt (John Day,) Arlington, Oregon

Very dark gray to black, massive, compact to vesicular basalt;
interlocking, crystall ine texture. So,, flows of this rock are
very highly vesicular; however, the specimens selected for test-
ing contained few or no vesicles. Core samples from dam-site
on the Columbia River.
2.1 P, Vest Niyack., New Ycrk

D iabase -(Pa1lsa e'
Black with I ighter gray, spockled appearance, medium-graincd,
dense, massive diabase; tightly-interlockina, crystall ine
texture. Block sarp;)le from NLew York Trap Rock Corporation
quarry, located in sill intrusion in Newark Group of Late
Triassic age.
2.2 Diabase (Cogqins), Culpeper, Virginia

Dark gray with lighter gray, speckled appearance, medium-
grained, dense, massive diabase; tightly-interlocking,
crystalline texture. Block sample from Coggins Granite
Industries quarry in Allegheny Mountains near Culpeper.
2.3 Diabase (French Creek), St. Peters, Pennsylvania

Dark gray to black with lighter gray, speckled appearance,
medlum-grained, dense, massive diabase; tightly-interlocking,
crystalline texture. Block sample from French Creek Granite
Company, Inc.
3.1 Dolomite (Oneotja,, Kasota, Minnesota

Buff, fine-grained, massive, porous dolomite; slightly mottled
appearance from small, irregular, white veins of calcite; pre-
dominantly granular, interlocKing texture. Block sample froin
The Babcock Company.
3.2 Dolomite (Locnort L Niagara Falls, New York

Gray to dark gray, very fine-grained, slightly porous, massive
dolomite, containing small infrequent blebs of anhydrite;
35
texture intcrmediate hotween ccrrrntcd-granular and inter-
locking crystalline. Block sa.ple from tuiagnra Stone Division,
McLain Industries, Inc.
3.3 Doleriltc (onneTTer Tre Missouri
Yellowish buff, extra fine-grained, compact dolomite, containing

numarous, narrow to very fine calcite-filled fracture veins that
are well-healed; interlocking, crystalline texture. Block
sample frcm Valley Dolomite Corporation.
4.1 Gneiss(>'orsh. '. Orofino, Idaho

Black and white, fine- to medium-grained granodiorite gneiss,
with fol iation making an angle of 450 with the ax.s of the core.
Somewhat interlocking, foliated, crystalline texture. Core
samples from dam-site located in the Orofino series on the North
Fork Clearwater River.
5.1 Granite (Pikes Peak) Colorado Springs, Colorado

Predominantly pink with small, irregular, dark patches of bio-
tite mica, coarse-grained, weathered granite; interlocking,
crystalline texture. Core samples from NORAD project,
Cheyenne Mounta in.
5.2 Granite (Pikes Peak), Colorado Springs, Colorado

Gray and pink, fine- to medium-grained, fresh, dense granite;
tightly-interlocking, crystalline texture. Core samples from
NORAD project, Cheyenne Mountain.
5.3 Granite (Parre)rr Barre, Vermont

Uniform gray, black, and white, medium-grained, fresh, dense
granite; interlocking, crystalline texture. Block samples from
Rock of Ages Corporation.
6.1 Limestone (Bed ford). Bedford, Indiana

Very light grayish buff, slightly porous, oolitic bioclastic
limestone; cemented texture of rounded fossil shells. Block
sample from Ind~ana Limestone Company, Inc.
6.2 Limestone (Ozark Tavernelle), Carthage, Missouri

Light uniform gray, fine-grained, compact limestone, containing
numerous fossil shell fragments; cemented, crystalline texture.
Block sample from Carthage Marble Corporation.
6.3 i . .(Solenhofen), SolenhJofen, Bavaria

Liglco yish bsiff, extremely "fine-grained, massive, lithographic
) ~c., containing lighter, thin, uniformly distributed 'bed-
g.vorc~*. Tight, interlocking, crystalline texture. Block
4(.. v Ut. S. Goologicai Survey necessitated coring parallel
Sb:. bceuse of size and shape.
1-17' It') Wast Rutland, Vermont

" ... ifcorm, fine-grained, massive, saccharoidal
l c:lo.klng,
I: crystall ine texture. Block
7.2 En rbie
I heroY
Lre) Tate, Georgia
Light grayish white, medium- to coarse-grained, massive marble;
tightly-interlocking, crystalline texture. Block sample from
the Georgia Mlarble Company.
7.5 Larble (t ,nlfl nbL st Rutland, Vermont

Vey)
Oft-white, rredium-grained, rr assive mrrble, containing infrequent
patches of derk ir:Qpurities; ti~htly-interlocking, crystalline
texture. Clock scrple fro-m Vcrront tsrble Co:: 2 ,.
8.1 uartzitc (!5rahL_ Baraboo, Wisconsin

Pinkish gray to purple, fine-grained, brittle, semi-vitreous,
massive Pre-Cambrian quartzite, containing numerous undulatory
beddin3 surfaces; tightly-interlocking, crystalline texture.
Block sample from Baraboo Quartzite Company, Inc.
9.1 Rock Salt (Diamond Crstal), Jefferson Island, Louisiana

Grayish white, translucent, very coarse-grained, massive rock
salt; interlocking texture. Block sample from Diamond Crystal
Salt Company.
10.1 Sandstone (Berea Amherst, Ohio

Light gray, fine-grained, massive, slightly porous sandstone,
containing very fine, light orangel uniformly to randomly dis-
tributed flecks; cemented to partially-interlocking texture of
subangular to rounded quartz grains. Block sample from Cleveland
Quarries Company.
10.2 Sandstone (Crab Orchard). Crossville, Tennessee

Light reddish brown, very fine-grained, compact quartzose sand-'
stone; tightly-interlocking texture. Core axis parallel to
bedding planes. Block samples from Luther Falls Stone Company,
Urbana, Illinois) and Crab Orchard Stone Company, Inc.
10.3 Sandstone (Navajol., Glen Canyor, Arizona

Orange-red to brown, porous, fine- to medium-grained, friable
sandstone; loose hemat;te-cemented texture of subrounded quartz
grains. Core samples from dam-site on the Colorado River.
11.1 Schist (Luther Falls) .L1 Unknown origin

Light to dark gray quartz-mica-schist. Extreme crenulations
visible on sides of core specimens, which were cored perpendicu-
lar to the foliation. Numerous garnets visible throughout.
Interlocking, crytall ine quartz texture, interspersed with
micaceous foliation planes. Block sample from Luther Falls
Stone Company, Urbana, Ill inois. Exact origin is unknown;
however, this schist resembles the Manhattan schist in essen-
tially all respects.
11.2 Schist (Luther Falls). 11, Unknown origin

Description as above except axis of each core parallels folia-
tion planes.
37
13.1 Siltstc, (!zkrr'rrk) t t'w Jersey
-ackensack,
Dark rcddish brown, massive, compact, clayey siltstone; some
more sandy and soma more shaly phnses; clay- and hematite-richJ
cemented texture. Core samples from frozen-in-ground gas
storage facility In Hackensack Kc.dows, located in Ulewark Series
of Tr lass ic are.
14.1 Tuff UTS-F Tonnl r:!rcury, FUvada

Very l ishit pink, porous tuff, containing uniformly-distributed,
white, dirk gray, and a few random, brown, lithic fragments;
cemented texture and a very lcw-density rock. Block saample
from Nkevada Test Site.
3. PREPARATION OF TEST SPrCI11ENS
Rock samples were obtained either as NX-size cores from the field, or
as block quarry samples which were then core-drilled in the laboratory. In
both cases, the samples were in an air-dried state. The standard NX-size
core, as accepted by the Diamond Core Drill Manufacturers Association, has a
diameter of 2.155 inches .005 inch. The field cores received for this study
were in most instances slightly below these limits; the smallest diameter
being 2.070 inches. Laboratory cores were almost uniformly 2.160 inches in
diameter; the largest size cores were 2.165 inches.
Block samples were in nearly all cases quarry-sawed, having dimen-

sions on the order of I-foot square and approximately 5 to 7 inches in
thickness. Samples which were irregular or larger than this were cut to
the desired dimensions by means of a 30-inch-diaraeter, water-cooled,
diamond-abrasive, cut-off wheel.
a. Laboratory core drilling.
Coring of the b-lcck samples was accomplished by a large, modified

shop drill press, fitted with a special d;amond-core bit. The drill press,
which is illustrated in Figure 4.1J was manufactured by The Fosdick M. T.
Co.., Cincinnati, Ohio. It is equipped with a 24-inch-square table with edge
drains, and has a 2 1/2-inch-diameter spindle with 6 inches travel for any
one run. Both the head and the table may be raised and lowered as desired
for maximum flexibility in sample size. The throat dimension is 12 inches.
The drill is powered with a 1-horsepower, 3-phase, 60-cycleY 220-volt,
1730-rpm, Allis-Chalmers induction motor. Drill speeds can be varied when
required as follows: 225, 370, 480, 800, 1085, and 1800 rpm.
The drill pre;s is fitted with a large, water-coolant swivel '.,33

Jacobs taper on spinV_'1 and high-pressure hoses for supplying large
volumes of water diresrly from the domestic, water-supply line. Recircula-
tion o' the coolant is ssible by means of an electric pump from the drill-
press sump; however, th- 6;d not prove to be satisfactory because the. volume
and pressure were too lPo. An 8-inch.-diameter aluminum pulley is mounted on
the hand-feed wheh-1 b-.. of which a cable and a weight system provide
constant load on the dril l as desired.
38
FIGUIEZ 4.1 DRILL PRICSS
39
For successful core drilling, ve that the block sample
be securely clamped in place. A cdan which was found to be
entirely satisfactory is illustrated in figures 4.1 and 4.2. This
consists of two 10- to 12-inch steel 'nairpin" clamps, bolted to the slotted
table-top with two 12-inch long bolts. Each clamp is supported at one end
by a 2 x 4 wood block; the opposite end holds down the sample. In order to
help align the block sample and provide rotational resistance, sLeel angles
are placed on two sides of the block and bolted to the table-top as well.
The drill bit used for all laboratory-drilled core was manufactured
by Anton Smit & Co., Inc., New York. This is a core-matic "glass" drill (so
named because of its designed use for cutting holes in plate glass). An
annular, diamond-impregriated sing extends 3/16 inch below the bottom of the
hollow, steel-tube. drill body. Because the thickness of this annular
cutting ring is approximately 1/16 inch, a minimum amount of power and
drilling pressure is reqcuired. Difficu!ty was experienced, initially,
during drill ing of soft limestone and marble specimens with this bit.
Circumferential ridges were obtained all the way down the sample, even
though constant loading was applied through all ranges of drilling speed.
The ridges were found to be caused by irregular, weld-metal surfaces on the
inside root of the annular diamond cutting ring. This was remedied by first
coring several specimens of quartz-bearing rocks (granite or quartzite) in
order to "clean up" the inside surface of the bit. Later, the manufacturer
was careful to correct each bit before shipment. The drill-bit size, used
throughout the investigation. was 8 inches in length and 2.155 ( .005)
inches inside diameter.
The most satisfactory drilling speed was generally found to be 225

rpm (the slowest available), particilarly for the softer rocks such as the
Bedford limestone. The denser diabases appeared to drill a little more effi-
cienly at 370 rpm, although no method to evaluate this was established.
With a reasonably new bit and a weight of 45 pounds on the pulley cable, a
6-inch long specimen of limestone generally required 6 to 8 minutes for drill-
ing. The same size specimen of diabase, with 75 to 100 pounds weight,
required 15 to 25 minutes. A specimen of Barre granite required 20 to 45
minutes with the same weight. The most difficult specimens to drill were
those from the Baraboo quartzite. With a maximum of 135 to 150 pounds
weight, these specimens required approximately 30 minutes to I hour to drill
with a new, sharp bit. The fifth or sixth consecutive specimen with the
same bit required up to 2 hours.
Water was used as a coolant for drilling all rocks except one. Com-
pressed air was fed into the coqlant swivel for drilling rock salt. This
proved to be relatively satisfactory, although the bit heated considerably
(too hot to touch). The operator was equipped with goggles and a respirator.
A large, commercial vacuum-sweeper. intake nozzle was installed above th3
drill bit to minimize the dust during this operation. Because of the neces-
sity for stopping intermittently to allow the bit to cool, approximately 30
to 45 minutes were required to drill each specimen of rock salt, 6 inches
in length.
40
i I
FIGURE 4.2 CLOCK SAI\&PLE CLAkIPED

IN POSITION FOR CORE DRILLING
41
b. SIwijrf tu lirnqth
All core sample-, were cut to a length of approximately k:.38 inches

with a diamond saw blade. The saw, which is illustrated in Figure 4.3, is
a Model J-3 Rock Slab Saw, manufactured by the Highland Park Manufacturing
Co., South Pasadena, California. This saw has a If-inch diameter, 750-rpm
constant-speed, diamond blade, and is equipped with a 3-speed, power feed
and automatic, cut-off switch. Power for the saw is provided by a 1/2-horse-
power, heavy-duty, General Electric motor. The cutting solution and coolant
for the blade consists of a mixture of soluble oil (20%) and water (80%).
The coolant is placed in the saw bo:t cr tank to a depth sufficient to immerse
the rim of the blade 3/8 inch.
In order to prevent (or at least minimize) the cutting solution from

being absorbed by the sample during sawing, all samples were soaked in water
for at least 15 minutes prior to this operation. Those samples which had
just been drilled were effectively soaked under pressure. This procedure
proved to be quite satisfactory for all rocks except the more porous, Navajo
sandstone. A bar of hand soap was used to prov;de a temporary, relatively
impermeable coating ove." these specimens. The soap was removed by soaking
in water following completion of sawing. Tne rock salt specimens were cut
"dry," after lowering the level of the coolant in the saw box below the
bottom of the bladc.
Just as in drilling, the specimen must be rigidly secured during

sawing. This is effectively accompl ished by the vise on the saw carriage,
a close-up of which is shown in Figure 4.4. Rate of sawing can be adjusted,
depending upon the abrasive or other hardness characteristics of the rock.
The softer limestones and tuff are cut at the faster rate, which requires
about 6 minutes for a single cut through an NX-size specimen. The quartz-
bearing granites and quartzites are cut at the slowest rate, and require
about 12 to 15 minutes per cut. The diabases, basalts, and harder sandstones
are cut at the medium rate of about 10 minutes per cut. Use of these pro-
cedures provided smooth, right-angle cuts, relatively free of saw chatter
marks.
c. Lapping the ends
In order to obtain polished, flat surfaces, the ends of all specimens

were lapped. A Lapma~ter, Model 12 lapping machine, manufactured by the
Crane Packing Company, Morton Grove, Illinois was used for this purpose. In
the lapping process, a compound consisting of No. 1800 aluminum-oxide abra-
sive in an oil-based, lapping vehicle is spraycd onto the surface of a
circular, serrated, rotating lap plate. One specimen is placed in each of
three, close-fitting, aluminum, cylindrical core holders, which are slightly
shorter than the finished specimen. The aluminum cylinders containing the
3peci-)ens fit loosely inside the three conditioning rings, which rest directly
on the rotating, serrated lap plate. A circumferential lip on the aluminum
cylinders rests or, the shoulders of the conditioning rings, thus holding the
cylinders off the lap plate and allowing only the specimens and conditioning
rings to be in contact with it. A felt pad is placed on top of each speci-
men, followed by a 4-pound pressure plate. After lapping one end, the speci-
men is turned over.and the pad and plate again replaced for lapping the
opposite end. The Lapmaster a.nd various lapping accessories are illustrated
in Figures 4.5 and 4.6.
42
FIGURE 4.3 DIAMOiNDl ROCKI $A1,W
43
S7:' I'll,
FIGURE 4.4 CORE SPECIMEN CLAMPED IN

SAW CARRIAGE
44
FIGUl"%Z e.5 LAPPI'G
"J2
FICURE 4.6 CORE SPECIMEN, CYLINDIRICAL CORE

HOLDERl, CONDMTQ~:ING RINGs AND PRESSURE
PLATE
45
Prior to lapping, each specimen (except rock salt) was soaked in
water at least 15 minutes to fill up accessible pore spaces, and reducZ the
amount of lapping vehicle that might enter the sample. In general, 20 to 30
minutes were required for lapping each end of a specimen for nine of the rock
samples. The quartz-bearing granites and one diabase required 45 to 60 min-
utes. The Berea sandstone specimens were lapped from 60 to 90 minutes on
each end. The softer dolomites, limestone tuff, and one marble required
approximately 5 to 10 minutes lapping time. Because of their friable nature,
the Navajo sandstone specimens could not be lapped.
By use of the above procedures, flatness and parallelism of the tes'

specimens were maintained predominantly within 0.0015 inch. After lappirn,
the specimen ends were rinsed in be izine solvent, then soaked for 30 minutes
in a water solution containing Tide soap. This procedure, which was folloked
by oven-drying at 100C. for 24 hours, was very effective in removing all
foreign cutting oil, lapping vehicle, arid moisture from the test specimens.
4. LABORATORY TESTS
Following oven-drying, all specimens werf allowed to remain open to

the ambient conditions of the laboratory for at least 2 weeks before under-
going any tests. Because of the variety of porosities represented by the
selected rock types, ar~d the various moisture and pore fluid conditions to
which the specimens were subjected during preparation, the drying procedures
used were considered to be the most desirable for providing a uniform basis
from which to begin the test program.
In recommending that testing of rock be accomplished in the 2-week,

air-dry condition, conclusions reached by Obert et al. (1946) state that
"oven-drying often produces pronounced and sometimes erratic changes in
both the elastic constants and the specific damping capacity - - - - more-
over, this effect is often irreversible." However, the data from which these
conclusions were drawn indicate that in going from an initial air-dry to
oven-dry condition the following physical property changes were observed:
i) Dynamic E and G decreased by less than 15%.

ii) Uniaxial compressive strength increased, on an average, 6%."
iii) Shore hardness showed no change for marble, granite and one
sandstone. Another sandstone increased by 20%.
iv) Specific damping capacity (ratio of energy dissipated per
cycle vibration to the total vibrational energy) showed
inconclusive results. There was a two- tcp three-fold
increase in some cases and a 50% decrease in another.
On the other hand, in going from an initial, air-dry state through various
moisture conditions to the saturated state, the following changes were observed:
i) Dynamic E and G generally increased 19% to 35% for some rocks

and decreased by the same or greater percentage for others.
ii) Compressive strength decreased, on an average, 12%.
iii) Shore hardness decreased 10%.
iv) Specific damping capacity increased three- to ten-fold.
46
Colback and Wiid (1965) present data to show'i that the moisture ,con-
tent has a major influence on the compressive strength characteristics of
some rocks. In tests on specimens of quartzitic shale and quartzitic sand-
stone., they found that the compressive strength under saturated cornditions
(subriiergcd in water) is on the order of 50% of tlhat under dry conditions
(dried for specified tim- over CaCI 2 or P 2 0 5 ).
It Is therefore a:parcnt that the potential errors and discreparicics

resulting from "wet.' variable[, and unknc;,.n roisture conditions are likely
to be much greater and further from the air-dry state (as recommended by
Obert et al.) 19 t465, than the potential errors resulting from the initially
oven-dry state as used herein.
a. Unit wei ht
All specimens were weighed on a Model K-4T ?1ettler, precision balance

to the nearest 0.1 gram imr-mediately following oven-drying, and then again
after remaining 2 weeks in the ambient air of the laboratory. The latter
values are reported and are used in all computations. During this period,
the length and diameter were measured. Lengths were measured by means of an
Ames dial comparator to the nearest 0.0001 inch. The average of five dif-
ferent readings, four at opposite points on the circumference and one on the
center axis of the specimen, was recorded as the correct length. The diam-
eter was taken as the average of three different measurements with a standard
shop caliper. one in the middle and one near either end of the specimen.
b. Shore sclerosco.e hardness
The sclerosco.e hardness was determined for all specimens by means of

a Model C-2 scleroscope, manufactured by the Shore Instrument & Mfg. Co.,
Jamaica, New York. The Shore scleroscope is a nondestructive, hardness-
measuring device which indicates relativ- values of hardness by the height
.& r-bound cf a small :'amonj-pointed hi.,irer dropped vertically onto the
test surface, from a distance of approximatel, 10 inches. The hammer falls
and rebounds within a close-bore glass tube. Air pressure, suppl ied by hand
compression of a rubber bulb, operates a catch which releases the hammer.
The bulb is connected by a rubber tube to a cylinder, containing a piston,
at the top of the instrument. The vertical ascent of the hammer after a test
is effected by squeezing the bulb, the hammer again being suspended by the
catch at the top. The height of rebound is read from a scale of 0-14O
divis ions.
This instrument is illustrated in Figure 4.7. which shows the position

of the specimen for taking readings along the sides. The scleroscope is sup-
ported on a cantilever, jointed-arm and may be freely moved to any position
over the specimen. The instrument is lowered into contact with the test sur-
face by means of the hand wheel on the left.
Two types of observations were made in this Investigation:
I) Twenty readings, spaced at approximately 0.2 Inch were taken

along each of four opposite sides of the specimen, making a
total of 80 side readings per specimen.
47
FIGURE 4.7 SHORE SCLEROSCOPE
48
ii) Five readings, equally s3ace6 on each of four separate
diameters, were taken on each lapped-end surface of the
specimen, making a total of 40 end readings per specimen.
The Shore hardness is taken to be the average value of the readings obtained
by either method. Both values are reported in the test results. Before a
series of tests on any day, the scleroscope wos checked by taking a minimum
of ten readings on a ;tee! standard, supplitd 1y the manufacturer.
b. Schmidt hardness
The Schmidt hardness was determined by mears of a Type L Concrete Test

Hammer, designed by E. Schmidt, Basel, Switzerland. This instrument has been
developed for use in estimating the strength of concrete, and is particularly
suitable for concrete already in place within a struc-ure. Use of the Schmidt
hammer for rock testing, as described herein, is relatively new, although
several authors have recently reported of its appl ication to rock for various
purposes. Roxborough and Whittaker (1964-65,1 report the use of the Schmidt
hammer for assessing variations in coal hardness along a mine face in rela-
tion to changes in mine-roof conditions. Knill and Jones (1965) used the
Schmidt hammer for determining the relative hardness of granite cores, as
noted in Section Two. Also Hucka (1965) suggests the use of the Schmidt
hammer for rapidly determining the strength of rock, in-situ in a coal mine.
The Schmidt hammer is essentially a nondestructive, portable test

devike which expends a definite amount of stored energy from a spring, and
indicates the degree of rebound of a hammer mass within the instrument,
following impact. Initial tests were conducted with the Type N hammer,
which has an impact energy of 1.63 ft.-lbs. These tests were unsatisfactory
because all but the strongest specimens broke upon impact. The Type L hammer
has an impact energy of 0.54 ft.-lbs., which is only 1/3 that of the Type N
hammer. This was found to be extremely satisfactory as only occasional
weaker specimens, were broken during impact.
Because of the possibility of cracking, or other damage to the speci-

mens during Schmidt hardness tests, representative specimens from each rock
group were set aside for this test series. These specimens were all initially
subjected to the nondestructive Shore hardness and sonic velocity tests, but
were not used for determining compressive strength and static modulus proper-
ties.
A longitudinal section of the Schmidt Test Hammer is shown in Figure

4.8., in which the vprious component parts are indicated by number. Figure
4.9 shows a photograph of the hammer lying alongside the steel "cradle"
which was developed for holding the specimens during testing. The cradle
consists of three main parts:
A steel block, weighing approximately 35 lbs., with a semi-

i)
cylindrical machined slot of the same radius as the rock
spec men,
ii) steel U-clamps for holding the specimen, if desired, and
iii) a flat steel plate, also weighing 35 lbs., to which the
block may be bolted, for use outside of the laboratory where
a firm base is not availab!e.
49
I. Impact plunar -11
2. Test r,pccimen
3. Housing 12 - _210
4. Rider with guide rod -20
5. Scale 23- 7 4 -22
6. Pushbutton -
7. Hammer guide bar
8. Disk
9. Cop
10. Two-part ring

I1. Rear cover
12. Compression spring
13. Pawl 4
14. Hammer moss 5-
15. Retaining spring 19-
16. Impact spring
17. Guide sleeve --3
18. Felt washer 14
19. Plexig!oss window
20. Trip screw
21. Lock nut
22. Pin
23. Pawl spring
A. Front fixation of
impact spring
B. Rear end of impact spring is
engaged to hammer moss 1 A
117
9
Note: 1
(0) Plunger (I) in
impacted position
(b) Impact energy - 0.075 mkg (0.542 ft-lb)
FIGURE 4.8 LONGITUDINAL SECTION OF THE TYPE L

SCHMIDT TEST HA11%NAER
50
FIUE
. SHID ES ESD
SPCAJE CRDL OLC"A, AS LT
51N
Test cbservations are rrncde by placing the specimen in the cradle
(itlanved or unclr;0od) and imrctir.9 ," hO.cr at vwrious points along the
topmost side. Tle h:erer is held vertically in both hands, and should be at
right angles to the a;xis, and tangent to the surface of the specimen. The
plunger Is placed against the spot to be tested. The hanm;er Is impacted by
gradually depressing the plunger into the housing. After impact, the ha-mer
mass rebounds by an amount related to the hardness of the rock. This is indi-
cated on the scale by a sliding rider. The rider position gives the rebound
value In percent of the tota: forward movement of the hammeor mass. By sim;ply
removing the hcnmer from the spot tested, it s reset for a furthcr test and
at the same timn the reading is cancelled.
In this investigation, three separate readings were made at equal

spacing along a line on the upper side of each test spec men. The specimen
was then rotated 450 and three more readings taken. This procedure was
repeated until a total of 24 readings for each specimen was obtained. The
average of these readings, for all specimens of any given rock sample, is
tha Schmidt hardness. Any obviously erroneous readings (such as improper
holding or slipping of the hrmmer) were disregarded and replaced by readings
In an adjacent spot.
Schmidt hardness readings could not be obtained on core specimens of

the Diamond Crystal rock salt, because they fractured during impact. Instead,
the top of a block sample was marked into 1-inch squares and one reading was
taken in each square. The average value of these readings is reported as the
Schmidt hardness for rock salt used in this study. P,'eliminary tests on
selected block samples of the other rock types proved this procedure to be
valid. There was found to be no perceptible diffe-ence in Schmidt hardness
values from block samples or from NX-size core, as long as the core or block
remained intact. The problem of fracturing cores during impact was also
encountered to a lesser degree in the Navajo sandstone, particularly in the
more friable specimens and those containing bedding planes. In this case,
additional core specimens were tested.
d. Abrasion hardness
A special device was developed to measure the abrasion resistance

qualities of the various rock types studied in this investigation. This
device was designe2d to be a mechanized, self-cleaning, "scratch" hardne.s
tester. An attempt was made to provide some degree of control over the
variables of force, abrasive, and time. while making this test, and to average
out the various mineral hardnesses encountered within any particular rock
specimen.
Two views of the abrasion apparatus are shown in Figure 4.10. As

can be seen in these photographs, the steel cradle block utilized in the
Shore and Schmidt hardness tests was used as a base for the apparatus, which
is primarily constructed of aluminum. A 6-inch long, adjustable rod is
mounted in a square pivot-block which is free to rotate on two pin-button
bearings. The bearings are mounted in a U-shaped yoke which is screwed to
the base plate. A steel axle-shaft is mounted horizontally in teflon bear-
ings., at right angles to the outer end of the adjustable rod. At one end of
the axle is a 7/8-inch diameter abrasion wheel. and at the other end is a
pulley of the same diameter.
52
r
4--
I - - -- ) I
- r
-
/
r r ji
Co) Side VIM?
I- 7-----;
/
rb;>
V -
V I 7$'
.5 -J/
(b) Three-Quoflcr View

I
FIGUflE 4.10 A6RASON ArA5AATUS
53
- -5' -_________
* - fl-;S.*J-------------
5
The abrcsioc"i tL:,,pratus Is pc-.,e rcdl by a l/0he15c~, l-volt A.C.,
rorev~m clectric'i,:::tor. It hns a sned raneo u;) to l4,000 rn, t~hich Is con-
trolrl.d by a fcot rowr Is su~pl icd to the epparatus throujh tw:o,
I/8-inch 0-ring Lc<ts. In order to m31ntain fi,-.d tension in the drive-belt
syste.- during cuttinC, tw'o pullcys are mounted at tha rotational axis of the
pivot-block, noccssitlng th3 tw:o-bolt systc;. Vlbrmal force cppl ied by the
cutt ing wh'Xl Is vnr ied b'y mz::ns of the extons ion arm end shot bucket M.,unted
on the pivot-block. A Vc.r ccuntar is lnstnllcd adjacent to tho cutting
whnel, in ordcr to ,::nure t% nuser of revolutions of the whcel daring a
test.
Each test was pcrformzd by cic.ping a speclinmn in the cradle block
end mounting a new, thin, double-cutting, dcntal carborundum disc on the
device. ihe disc rests directly upon the specimen surface, colinear with the
radius and in a line paralleling the axis of the specimen. A short weight of
55 grams was used with an extension arm length of 4 inches and a distance of
3 Inches between the cutter-axle and the pivot-point. The instrument was
supplied with full powecr for a timed period of 3 seconds. No attempt was made
to allow for frictional starting resistance, or inertia, at the end of the
timed period. These two unknowns were assumed to balance out. Tests made at
5-second and 10-second periods proved this assumption to be reasonably correct,
as no significant differences were noted in the abrasion resistances, when
compared with the 3-second interval. The counter provided a means for quanti-
tatively checking the relative amount of slippage in the belt-drive system
during a test, and indicating the validity for comparison with the other test
data.
In general, one test was performed on each of four opposite sides of

a single.test specimen from each group. A new carborundum disc was used for
each test. The depth of each cut was measured with the Ames dial comparator,
using a specially-built foot which was the exact size of the carborundum
abrasion disc. Areas of the circle segments thus cut were comnputed and aver-
aged to obtain a relative, abrasion-resistance number. Since the softer rocks
had deeper cuts (thus .higher numbers), the abrasion hardness is reported as
the reciprocal of the computed area cut. Abrasion hardness values were not
obtained for the Berea (10.1) or Navajo (10.3) sandstones, because their
coarser grains blocked and clogged the cutting wheel, thus stopping the test.
e. Absorpt ion
The absorption (or porosity) characteristics were obtained for one

representative specimen from each rock type by using a vacuum-saturation
process, similar to that suggested by the U.S. Bureau of Reclamation (1953).
This process consists of:
i) oven-drying the specimen for a period of 24 hours at 105IC.,

II) placing the specimen in a bell Jar under a vacuum exceeding
20 Inches of mercury for 24 hours,
iii) Immersing the specimen In water while continuing with vacuum
for another 24 hours,
iv) removing the vacuum and exposing the water containing the
Immersed specimen to the atmosphere for at least 48 hours.
54
This proccd ure el IImir r,. ir frcm: t rock porcs rd e-1 Trs tV3 wcter;
?-a,
r.tmosphcr Ic
prbssuro t k, n cr irs v,,- t.cr ir teo tLh rock, t ,u.s tur ing snsture-
t ion. Upon complet icn of tlI-. 5-d:y cycIr, tYD sc Ih-r Is rc.ovc d frc- tho
water, surface dr lcd, cnd t:;cn wr~lihz' to the r.crost 0.1 r. The absorp-
tion Is computed by tL: follor.,,j iq tion:
P2- P
Albs o r Pti on .... x 100
in which P n wvi.kt of saturated yc' n

P12 - weight of ovwn-dry sp~ci2;.en
Absorption was not determined for 83araoo quartzite (g.1) and Diamond Crystal
rock salt (9.1).
f. Sonic pulse vclty rreasure ... nts- (Pt so..tln _0
Standard pulse techniques were used for the detcrmination of the

sonic velocity for all rock specimens while subjected to a loi, axial stress
(seating load) of 100-150 psi. This technique Involves the generation of a
short-duration stress pulse of low anmlitude at one end of a speciman, and
accurately measuring the time taken for the arrival of the first recogniza-
ble energy-pulse at the opposite end. The velocity, V . which is the velocity
of dilatational waves in an unbounded mcd ium, or the bclk ccmpressional
velocity, is then corm;puted by simply dividing the length of the specimen by
the measured travel time.
The schematic diagram in Figure 4.11 illustrates the interconnection

between the major cor.ponents used for applying a pulse and measuring its
travel time through a rock specimen, subjected to a uniaxial static load.
The two stainless-steel transducer cups, referred to as the 'receiver" and
"Istransmitter, are identical in construction and are, therefore, interchange-
able. Each of the transducers contains a I/8-inch thick, barium titanate
crystal, which has a natural frequency of 200-300 kc/s. The crystal Is
cemented directly to the interior face of the transducer cup. Coupling with
the test specimen is greatly improved with a thin coating of Dow-Corning
high-vacuum crease, smeared between the face of the transducers end each end
of the specimen.
A short-duration, high voltage (I,000 volts maximum) pulse Is applied

to the transmitting crystal from a high-gain, high-pass signal amplifier.
This pulse generator causes the crystal to expand and contract rapidly, thus
Imparting stress pulseq to one end of the sample. In this way, ?ntermittent
trains of ultrasonic waves are propagated through the rock specimen. the
arrival of the pulses at the opposite end of the specimen, causes mtchanical
contraction of the receiver crystal, generating a voltage across the crystal
faces. The induced electrical signals are amplified and displayed on the
cathode-ray tube of a Tektronix Type 535A Oscilloscope, equipped with a Type
8 Plug-In Unit.
The oscilloscope display is formed by a repetitive sweep of the input
wave-form across the oscilloscope screen. In order to examine the character-
istics of the wave-form in detail, it is necessary for the repetitive wave-
form to appear stationary on the screen. This type of display requires that
55
Qo 0Q Co
o0 04 0
00 0 (9 00
-. (>
0000,
x0 -
IL z
0' 0
0-0
w
m~ C)
- - - ~ r - - - -.
56
the start of the sweepbr Vr (definite, f ixad-t iri- relat iorsh p (hence,
synchronized) with tU,: c,,!:,,ocrcnce of tie inmput wzvc-form This is
accomplished in the oscilloscope by triezring (sta'rtlng) the svwep with
the displayed wave-fckm:.
The calibrated sv:ecps of the Type 535A Oscilloscopc ceuse any hori-
zontal distance on the screen to reprc~osnt a definite, knc,%n interval of tin>e.
Because of this f0t... th: t irh::
lck betwecn twto events d l:rielyed on the
screen, i.e., the be-ifeirro, end the "first arriv3l" of a so.nic puIsoe at thoe
opposite end of a sric.-,rn :y be accurately rTcwesured. TL mOthods are
used for time measurc>,','.nt:
I) Using the graticule on the screen, the horizontal distance

between the two displayed events is measured directly. The
distance measured in centimeters is multiplpied by the
appropriate time/centimeter control setting (which determines
the horizontal spread of the displayed wave-form). Photo-
graphs of typical records obtained in this way are shown in
Figure 4.12. The first arrival of the dilatational wave
occurs at the first break (bend) in the horizontal portion
of the displayed wave-form. Since we are Interested in the
first arrival only, the entire wave-form is not displayed.
ii) Use of the delayed sweep feature of the oscilloscope provides

a very high degree of accu' cy. The delay-time multiplier
control is rotated until the leading edge of the tracking
pulse just merges with the first break In the horizontal line
of the wave-form. The reading from the delay time control is
then multipl ied by the tire/centimeter control setting.
Although both methods are used simultaneously as a check., each on the

other., all time measurements were made in this investigation by the second
method. This method eliminates parallax errors, and those reulti.ng from
visual determination of the location of the beginning of the sonic pulse.,
which are inherent in the first method (using the graticule only). Correc-
tion for instrument delay time must be made by subtracting this predetermined
"constant" from the time readings obtained by either of the above procedures.
A unlaxial seating load is applied during this test by placing the

core specimen., along with the transducer cups, between the loading heads of
a hydraulic load press. The seating load is applied only to obtain better
coupling.of the transducers to the specimen. Load is measured with a stand-
ard steel-column, SR-4 type, Baldwin load cell. The strain Indicator is set
to a predetermined value of load, computed from the load-strain calibration
of the load cell. A hydraulic hand pump is used to maintain the desired load.
g. Sonic pulse velcicity and static stress-strain measuremernts (durin,

cycl ic load ino)
Approximately hIl6f of the test specimens were individually subjected

to simultaneous measurements of sonic pulse velocity, Poisson's ratio. and
modulus of elast!city, at various axial stress levels during cyclic loading
under static conditions. Sonic pulse techniques as previously described were
57
(a) Varao acndslene ur cr: unlaxlcI m-prczr~v
(b) Oeroo sajndston undenr unlaxiol comproazfva

stress of' 5,000 pei
F1~24.12 TYMIAL SONIC PU1LCM WAVE -Po-z, DISCPLAYS
58
?loyd
I for r,,s-'rce-:nnt o'f the d Ilatat iorn wve velocit ies. .21licat ion
and mriesurcri:nt of unia ial eds o, were acce:?lis ?<d by the procedures dis-
cussed in the precediig pra,reph. Vertical and lateral strain nme.:surcr.ents
were obtained by use of resistance strain c;:-.s bncrid directly to the speci-
men. Equipment for perfor.ilng this test series is illustrated In Figture 4.13.
A specimen which hias an L/D ratio of 2:1 may be considered to be

under almost uniform vertical stress unc!ur the middle 5/6 of its hainht,
during compression testing (Fairhurst, lI1b. In this test series, two
Baldwin S2-4 Ty0 e A-I stro in gca:s, he:vino a ;. lernqth of 13/IG inch are
placed vertically on opposite sidzs at mid-heLait of thc specimen. Thele are
wired in series for vertical strain measurcmrrnts. A second set of two gages
are placed horizontally at mid-height on the opposite diameter to the vertical
set, and are also wired in series for lateral strain measurements.
All four strain gages are securely bonded to tbe specimeo L"y means of
Eastman 910 cement, which is pressure-sensitive and sets up In approximately
30 seconds. This is a cyanoacrylate adhesive, which is manufactured by the
Tennessee Eastman Company of Kingsport, Tennessee, a division of Eastman Kodak
Company. The combination of rapid setting action, high bond strength, and
ability to bond to dissimilar materials, all at ambient temperatures, make
this cement an excellent bonding agent for this purpose. There is an extreme
reactivity associated with cyanoacrylate, and any water, or water vapor is
sufficiently basic to cause deterioration of the adhesive over a period of
time (McCarthy, 1964).
The surface of the specimen is first cleaned with acetone, and the
back of the gage is coated with a special catalyst. Following zpplication of
the cement to the specimen, the gage is moutsnted and -oderate thumb pressure
is appl ied through a sponge rubber pad, which is prevented from sticking by
means of a small sheet of teflon. The gage is finally coated with a plastis
water-proofing (in an organic solvent) to kecp out moisture. Both the water-
proofing and catalyst accelerator are produced by the Budd Company, for use
in conjunction with the Eastman 910 cement for strain gage application.
A close-up view of the specimen, strain gages, sonic pulse trans-

ducers, and load cell within the hydraulic loading frame is shown in Figure
4.14. Measurement of strain in both sets of strain gages is accomrplished
with one SR-4 strain indicator, by means of a single-pole, double-throw
switch. Strain read-out is illustrated by the photograph in Figure 4.15.
In this investigation, each specimen was subjected to two complete

loading and unloading cycles up to 5,000 psi, except rock salt (9.1) and
tuff (14.1). Because of their low strengths, these two latter groups were
cycled to 1,500 psi only. Loading proceeded at a uniform rate by smooth,
steady strokes of the hydraulic hand pump. Loading was stopped and maintained
at constant values while sonic velocity and strain readings were made. These
readings were obtained at zero axial stress ard every 1,000-psi stress inter-
val for both loading and unloading, following an initial seating load of
100-150 psi. Readings were taken at stress intervals of 250 psi for the
rock salt and tuff samples. Efficient performance of these tests require the
combined efforts of two operators. By use of the procedures described above,
approximately I hour is required for performance of one complete test through
two loading cycles.
59
-I
'S I----I

.. 0 I
FIGUriE 4.13 EQUIPMENT FOR FEASURING SONIC

PULSE VELOCITY AND STATIC STRESS-STRAIN
DURING CYCLIC LOADING
60
?,7
FIGUflE 4.14 C07E SPECU2IN IN POSITION FGR

MEASURING SONIC PULSE VELOCITY, STRESS
AND STRAIN UNDER UNIIAXIAL LOAD
61
top,
FIGURFE 4.15 STRAIN READ-OUT
62
Al I spec ir,: rs wh ich hd bscin tcstcd ure:r cycl Iic I:c,4,j co.JJr It ions.
descriL.d aLove, wcrc also tested In static cr.,w,.rcssioa to failure. A 3C0-kip,
univorsal, Richle hydraulic testing rt:ch lr w- s us,,d fcar lc.dliv t,'-, cc3.:as.
This mcchine has six load ranges. The lo-:e-st renr,-:: uscd for balancing the
machine and the ranges were progressively incrceed, as th,' so.ir:.n aprozched
failure, in order to insure mIn;!irmcciir'r, irn tf. z;:x .i lc-,; -: urc,:.:nts.
A photo~rc;)h of tho static cc:: 2rcgsl, tc,:, r chc is sY... in F1
Vertical ,ndlateral stra.in r!:,:urc :. ;it s-::rc iw d by r : es of tL'..

SR-4 resistance strain :,es provyiously uscrd in the cyclic lz-nding, tests.
However, in order to maintain a continulus rate of load eppl ication, strain
read-out was facilitated by two strain indicators. one for each set of gages,
rather than by us ing the much slower, s0,witching procceuf-e as in the cycl ic
load tests. In order to provide a mans for checking the vertical resistance
strain gage readings, one dial indicator was placed batwen the ioading platens
of the compression machine. as close as possible to the test spncim:,e. A close-
up view of the specimcn as instrumented and positioned in the test machine Is
shown in Figure 4.17.
All compression tests were conducted on uncapped specirmens, as reconm-

mended by Obert et al. (19L6). Care was exercised in centering the specimens
in the testing machine and zero-load, strain readings were taken fro,. the
strain indicators. An initial load of 100-150 psi was thcn applied very
slowly to permit adjustrm,,!nt of the sphevrlcal cc,,ression block. After appli-
cation of the "seating" load, initial rca,dings were taken on the machanical
dial indicator, which was graduated to 0.0001 inch per division. Loading
proceeded continuously to failure, with all strain readings obtained sirul-
taneously at predetermined load intervals based upon the esti;L tod co:.oressiva
strength of the specirm,,n. Since loading w3s manually controlled, rate of
loading was not the samn,: for all rock types. In general, the averag2 loading
rate was 10 to 25 psi/sec for the low to medium strength specimens, 35 to 50
psi/sec for the medium to high strength specimens, and 60 to'70 psi/sec for
the very high strength specimens. Essentially constant loading rates were
maintained for specimens from any one particular sample. Typical compressive
failures are shown in the photographs of Figure 4.18.
The compression tests constituted the third loading cycle for those
specimens previously subjected to the cycl ic loading tests. Selected addi-
tional specimens from each rock group were tested in compression to failure
without strain readings. These were conducted at the same rates of loading
as those specimens with strain measurements. Ultimate compressive strengths
were computed by dividing the maximum load at failure by the average Initial
cross-sectional area of the specimen.
I. Point-load tensile testF
Measurement of point-load tensile strengths of representative speci-

mens from each rock group was accomplished by methods described by Relchmuth
(1963), and D'Andrca, Fischer, and Fogelson (1964). The point-load tensile
strength is determined by applying compressive point-loads to the curved sur-
face of a cyl Indrical core specimen., of which the axis is placed horizontally;
the specimen is tested between the loading heads of a load press. The point
63
NW_ * _
-.
I
FIGURE 4.16 COMJPRESSION TEST MACH1INE

- fI
FOR CO'iPRE S SION TEST TO FAULUR
4 .
.. .. ..... .... .. .
- . ,i l il ,
65
F
- 41'
4 4-.
It
9
(a) Dtcxzht Cwha
4
113]i
Ib) KcSers;z
FISUCZS 4.10 T'CLZAL ccz:x20c;v: FAILWTZS
66
loads are obtc,11%
C r, r-nznns of sr:l 1-d rclIicIta
stc7: I~rrc'
orknda r ~t. xi ~cs to0tL czx i s o f t 1he spc Ii- n . Th is prc du;. inltcrn'c
tens ile strcssoc n iua to thei londing ax is. A scL'-,.'7t ic dc In i
Ficure 4.19 lllvuctrcntcs the gencral Szcci--,try of thism tcst. risurcs 1".423 Cnd
4.21 are photo;:-r Lh of the actual cniparatus uscd.
Test splnc ir.-cas for this ser los wcnre i&zrt lcd- in aillr e with-
those used in ot;itcsts. Thcy wsre scelctccl frcci all groups pre r.cda',. in it i-
ally for this irx.eit ion. I"C~, z
n ' F,7 t r Li _t lyve
t c.1 a hen
i-dlbnC i ,c d
for each spc i., Je cd
'a ing, fveedjns, i th 32 rc
wc~i: tr
harde:,ned selrol lcrs v.,crc e~ c and~ instal lcd: d ircctly on tNO, iced clII
and hydraul ic.'J7c!k piston of tho lornd press. The (ehvri loaid cell wes raised
an add itijonal tCirce inchas with flat, stool bear ing blocks. This redUce-d the
amount of Jack e;xtcnsion to a minimum, effect ively providing mraxir.-,,um lateral
support for the jack piston.
A Speed-0413x, Type GP strip recordar was used for determining the

maximum failure load, The recorder was adjusted for full-scale reandings of
500, 1,000, and 2,030 micro-inchcs/inch (de-pending u'pon the estitiated strength
of the speciraLn) by me-ans of a CLH Strain Indicator. Since the load-coill
calibration was kno'wn to be 9.6 lbs/p-in/in, the total load could be accurately
computed by multiplying the proportion of the full-scale reading roached at
failure by 9.6.
Each speciman was loaded mid-way between the ends; the load was
appl ied by smooth., even strokes with tile hydraul ic hand pump. The point load
tensile strength, T,. was Lomputed from the following cecpiricao expression
given by Reichruth (1963):
T - 0.96 -.. (4.2)
In which F - total failure load in pounds

D - core diameter in inches
Two tests in this series were conducted from each of 19 rock groups. One test
was conducted for each of the other nine groups.
5. TEST RESULTS
The physical and elastic properties for 28 different rock samples
studied in this Investigat ion are summarized in Tables 4.2 and 4.3. Division
of the data Into two separate tables is primarily for convenience In presenta-
tion,.because of the large number and kinds of variables Involved. Both
tables, though slightly different in organization, present the same type of
Information for each variable:
1) the average value of each variable,
11) the number of tests from which the average was obtuined, and
111) the coefficient of variation, V/%, for the variable.
All raw test data were put on standard keypunch card: and reduced by means of
an IBV4 7094 computer. Individual test results for -.achi gpecimen were subjected
to statistical analyses, as described in Appendix C, In order to obtain aver-
age values and dispersion characteristics for each property of each rock group.
67
00
V 0 CLi
to-
emS
a 4- - u0
I w
0:
I ~LI
0 I(
r
F->
I
I
I
I
I
FtGUfl 4.20 POIT-L0AD TZL TEST EQUIPJT
69
_____________________________ ______________________________
FIGURE 4.21 CORE SPEMCI112- IN POSiTION
FOR POINT-LOAD TENZILE TEST
70
- t in-
t 1o- 0 P
es ~ )P
::; 0;
d4 ff 04n4 V_
I
0e V4
04 od ; m C
a.. SO
g d
X:xmmuI
9 ___ H I ~:
____
____
~ Ufl U
L 1
4 .9.9 .90 .9~~~0 i 9 .0 .0 -
P4 - ____71
r4 0
*1
-v .4 a . J
v -4
44 to1 0 re m ; kA00P
0* w
4 fta 0 fr
ON N r
fm
- 0 N S 4S 00 0 N P4 V
caI N0 P4 0 4. M M N 0 O
- N N - N
fg3 ~ 4Z
I-I - ~0-
N - N N
o a 72 -
- N N N 0
C~~t ~ -N * 4 *4 ed
VN
IN --
U) fq
IS.
el o n a, 0
.0 0; 0v N t F. ID
0, -- N NN
114 to J U ~- N U)
_ _ _ 73 W U)
m fm C-4
tof
$f- 44f i- P4 ~
046 n ~
ft ft P4 ow 14 4444
00-n
3 ~0 ~ -s ~ ~ I.q a1 c u . * u
-~~~t -4/ 6 6-bF bn N
tIZ-'5 74 -- - ~
13 9-
qlj e.
DI O O np N 1
S S
or vi
60D ~ D 0 ~ O f
* IUD
-5-;
Ira
**n N 0
S.D
9.. _ __ __ _ _
D75
*A ,r, , - v--'- - t tr -st Vl t tr S
-~o
~ ~61 mi
w~P ~ ~ 2 ~ 2 .~a
t, V; 0. * 4)
r1 0- f%
- % -% 0
04 4L
0 40 40 em f. 4 - Ao t 0 o--
Ca'at
* 0 - N Pt a Nht 01
4 ~U ~N 76 N
CA a Iue t- ~ '
40
10. m -4
. . -I' .
04 M- -" 0
to _ _ 9 * _
.cc
j a. ft a,-
-i4 U0 .. . .
ri-
.
-NO
-.
U
fl
0A
4WQ
-NmN
~ Ucm
N - - N
i ___FA
- _ __ __ 77- - . 2 ~
a-
6 6
2 0 Is a0
: -* - * * " - * ", .,
SV
P F I- a a -
-!- Af---4M 4 0
- P) 40
,, , ,
78
a W,4 0 4 . co1 cl 0, 000 ftA
'SA
MA S m --.
*~~~ S 0fq*
S .0 . ~iU 5In
t~
P: 14 - - a- 4 4 0 0 f
04-
I-
N.~~ 10 0 4 f
o U _ ____ _ __ __79
e 0 * -W... ..~ O.. . (. .. .... ....
V0
t- In
ii
14,' I' # - "= - -
It
gin
80
.71
PbN to 0P0P -
- l s ~
3LI I
, 1:-1 -1 _ _ _
- Na.
h. ci F 0 4l P
I *~~~~~ 2~ f -40~P 0- ~ @ - ~ .'-f
0 0- v - U -. N U e. wb~ C Dq O 0 0 0
040~
2 x x 2 0 - 01
IL f
A~ C4f
Uc
~.. ~ ~U~ V id s
81W 4 00 P -
TLIeb 2~ 2~ iza ttis :. prc:' rt ics v.h ich rc.;y L c:j i1e:,
rd.riz.rc or
lc.s~~ic~1v
e ior "conWtnts" for eccch rock ty>. I co:t~t YL
p r 0;rtirt in
1)I .V 11.3 have several valucs, dcpondnirJ U:.;C ,n1A strcs Isvo I
invo v.I. (v-,-1:n of tLse term.s is not r riorc-s, but Is int,,od for urvoc
of Scn,:Aral cl,'orif ic,-,tion.) Table 4.3 Is dividod into to soctions:
i) Elastic Properties (First Loading Cycle)

This section ccntains both static and dynn;Cnlc values for
thoase [.,rcprrtics cc:'..:utcd frc;,i data obtain.vd cd-ring tVNO first
loadirn cycle nnly of the tvo-cyclc, sonic volocity strcss-
strain tests to 5,000 (and I,00) psi, doscrilbd in p.rarc.ph
(g).
ii) Static Elastic Properties (Loading to Failure)
This section contains those properties computed from data
obtained during the compression tests to failure (third loading
cycle), described in paragraph (h).
All values reported for E and v are tangent values computed from. a tangent to
the corresponding curve at the reported stress levels. Secant values are not
reported in these tables.
Appendix 8 contains graphical summaries of typical data obtained for

one representative specimen from each of the 28 different rock samples. Two
consecutive sheets., A and B, constitute the sunmary for each specimen. Infor-
mation within the square box in the upper corner of each sheet identifies the
rock type and specimen, and includes certain physical data for that particular
specimen. The Shore hardness is given for the average values obtained from
the lapped ends of the specimen. The sonic velocity, Vp, given in the data
box on shoet 3 of each set, is the dilatational wave velocity at a seating
load of 100 to 150 psi.
Sheet A of each graphical data set, summarizes the data obtained

during the static cyclic loading tests to 5,000 (and 1,500) psi. Plot (a)
shows the axial stress-strain behavior, during both loading cycles. Plot
(b) shows the variation of Poisson's ratio with stress, as determined from
the SR-4 strain gages, for the first loading cycle only. Both the tangent
and secant values are shown thereon. Plot c) shows the variation of the
dilatational w~ve (sonic pulse) velocity with stress, during both loading
cycles. Plot (d) shows the changes in the various moduli of deformation,
during the first cycle of loading. The tangent and secant moduli values, Et
and Es res ectively, are computed from the static stress-strain data given
in plot (a). The constrained modulus values, Mcy are computed from the sonic
pulse velocity data given In plot Cc) by the equation:
M
- V2 (4.3)
In which p - mass density

Vp - dilatational wave velocity in an unbounded medium
Sheet B of each graphical data set presents the data obtained during
unlaxial compression to failure. Plot (a) shows the axial stress-strain
behavior. 8oth the Initial modulus valua, El, and the tangent modulus, Et.
82
which wns cc *rt-, at a stress equal to on:-half tko ult ircte ccc;)rCssive
strenflth, arc r.ccrcd thereon. Strain w:casurcrmzcnt: d&;to.riirnd from the
SR-4 strain s arc shovin to start from zero, althoi<h this test is
actually th. third looding cycle for the spec irnn. These tests wcre porforf;;od
a minimum of 2 to 3 wceks after the cycl ic load tests, euring which time no
atterwjt was m:c to ri.acsure strain rcbound. Little or no .oraa3nsnt set w3s
measurcd during tC cyclic loading tests, howo,;cvorJ, cxccpt for tl"ie speciri,,r::ns
of rock salt (.I). Plot (b) sh1i'.,s tho variation of Po i:sra is ratio during
loading to fc ilurn, cc,.:?utcd for both the, turnrnt and s:c;:,nt valucs. Thi,,
theoretical Iimit of Poisson's ratio is 0.5, and hence, rio valucs are SI';;:n
above this l imit.
Relationships among various pertinent physical and elastic properties

are presented grqphically in Section Five. Significant correlations used in
development of index-property dasignations, and consequent rock classification
charts; are given in Section Six.
83
................. ....... I I
DISCU$SIM
I ZD, If2 iZ:TAT ICN O
,1TL ME "RIE6!6TML DATA
In this scction the erperimnntal data are examincd statistically and

tha correlations ar.ong the various rock properties are given. The difficulties
encountercd in r.:zsuring so:m- of the properties and the bases for the inter-
pretation of the test data are discussed. A summary is given of the procedure
used in the present statistical correlations including a discussion of the
necessary replication (samsple size) for statistical inference.
2. STATISTICAL PROCEDURES
a. Required snmole size
When a treatment or test procedure occurs more than once in an

experiment, it is said to be replicated. According to Steel and Torrie
(1960), the functions of replication are:
i) to provide an estimate of expe.lmental error,

ii) to Improve the precision of an experiment by reducing the
standard deviation of the sample mean for a particular
property,
1ii) t~o increase the scope of inference of the experimental data
by selection and appropriate use of a wide variation of test
samples, and
iv) to effect control of the error variance.
An estimate of experimental error is required for determination of
significance and for confidence-interval estimation of a given set of test
data. When there is no method of estimating experimental error, there is no
way to determine whether observed differences indicate real differences or are
due to inherent variation. In such cases, any inference must be based upon
prior experience.
The degree of precision required is probably the most important factor
affecting the number of replicates. Increased replication usually improves
precision, decreasing the width of the confidence intervals and i,,creasing
the power of statistical tests.
iplication does not reduce error due to faulty technique or equipment.

Neither does mere statistical significance give information as to the practical
Importance of a departure from the null hypothesis (i.e., a statistical hypo-
thesis fornulaWi for the sole purpose of rejecting or nullifying it). The
magnitude of a real departure that Is of practical significance can be judged
only by one with technical knowledge of the subject; this maqnitude, together
with a measure of.how desirable it is to detect it, serves to determine the
required precision and, eventually, the necessary number of replicates (Steel
and Torrie, 1960).
P8.
Rock test data, published by the U. S. Oureau of fKinzs (l49-I%0) are
reported to have a probability of 75 chances out of 100 or better of being
within 20% of the population mean (i.e., the mean approached as more and mre-
observations are made). For standardized physical property tests, the Bureau
of Mines (Blair, 1955) rccommends an optimum and minimum nuebcr of specimens
for each test as listed below:
Mo. of ?pecirnmns per test
(I) Compressive strength 10 5

(2) Shore hardness 10 5
(3) Sonic velocity
(dynamic-elastic constants) 6 3
(4) Modulus of elasticity
(static-elastic constants) 4 3
From the definition of limit of accuracy, it is possible to determine

the sample size, n', required to predict the population mean within a certain
definite range for a certain degree of accuracy. Since limit of accuracy,
L - tj then r ts or the required sample size:
I 2
ni (t (5.1)
where s - sample standard deviation, and

t - "Student's" t, for n-I degrees of freedom.
Thus the required sanmle size depends on the specified confidence interval and
level of significance, which can be altered to fit various experimental condi-
tions (Liu and Thorrburn, I>$5).
In the present investigation, the number of specimens or measurements

performed to determine each property was not the same for each rock group
because of the type of test involved and the number of specimens prepared for
each group. A summary of the average (or usual) number of tests per group and
the average coefficient of variation for each measured property is given in
Table 5.1. In general, either 6 specimens or 12 specimens were prepared for
each group. The number of tests from the 6-specimen groups is given on the
left of the middle column of Table 5.1, and the number from the 12-specimen
groups is on the right of the middle column of this table. For exampl3, for
compressive strength, 3-8, means that in general, there were three tests per-
formed for the 6-specimen groups, and eight tests for the 12-specimen oroups.
The average coefficient of variation was computed as a weighted average on the
basis of the number of tests and the corresponding coefficient for each group.
With two exceptions, the compressive strengths of the population means,

of each reck group in the present investigation have a 95% probability of
being within 25% of the sample means, R. In other words, the number of tests
performed for 26 of the 28 rock groups is greater than the number of tests
required by equation (5.1) at the 5% level of significance and a limit of
accuracy (confidence interval) equal to 25% R. For the standard deviations
obtained in this study, Group 3.3 requires nine tests for compressive strength
85
TAI.C !2
f C=F
R, JTS f:, 1"
CC FFICJ11 VlS 0F
Averao o
No. Tests Confficient of
~rft;Lr~9,1r,-.ti e. r. variain. V 7,
Un I t ifh t 6-12 .5
Cmpressive strength 3-8 7.6
Shore hardness (sides) 480-9S0 12.7
Shore hardness (ends) 24o0-4 11.8
Schmidt hardness (all readings) 72-144 7.7
Scfl.midt hardness (high 50%) 36-72 4.7
.Ibrasitn hardness 4 15.3
T,l' strength (point-load) 1-2 12.0
Initial tangent modulus (Ist cycle) 3-6 22.4
Tangent modulus (at 50% "a ult.) 3-6 11.3
Initial Poisson's ratio (1st cycle) 3-6 82.4
Poisson's ratio (at 50% aa ult.) 3-6 19.5
Dilatational wave velocity (100-150 psi) 6-12 2.7
Dilatational wave velocity (5,000 psi) 3-6 1.9
and only three were performed. Group 11.2 requires 18 tests and only three
were performed.
For the modulus of elasticity at 50% ultimate strength, three groups

do not meet the above precision criteria. Group 3.2 requires five tests and
three were performed; Group 9.1 requires 19 tests and only five were per-
formed; and Group 13.1 requires eight tests and six were performed.
The unit weight, Shore hardness, Schmidt hardness, and sonic velocity
are well within the above precision criteria. Indeed, with few exceptions the
population means of these properties are within 10% of the sample means Pt the
5% level of significance as computed by equation (5.1).
b. Correlation studies
Statistical methods of analysis were used for obtaining the least-

square regression line, the standard error of estimate, and the coefficient
of correlation for each set of variables. The statistical methods are
desrribed in Appendix C and are similar to those employed by Judd and
liuber (1961).
86
The cvcre:e a:ta govi:n in Sr..ction rc.r for cnch rock prc;v.rt,"y ::r, pvt
on standard keyi:;unch cards for co:<;,utCr C,1,lysis. Correlation and rcrssion
prrgrems written by the Statistical Scrvicc Unit of tho University of Illinoils
id described In V:Inual of Cc.:auter Prcgraias for Statistical ft-ilysts,"
were used in this study. SSUPAC consists of a collection of ICA ". coj-,:utlr
progrms stored pernmncntly on th; Il: 1301 Disk StcrCe; File. 1cclud-3 In tel
collection are rn-:ny co:,.:mon statisticalA procneures, a cre-3pcto sc'rt ol bzc
matrix an;d " .r prcc ..
ratio;-, and os for tV r,: nipul ,r', of c L or"r
to call a proOgrc;: frc-- tie dlUk into t.. coi r:::ery r. dAy for u-' it is neces-
sary only fcor th user to sut:ply a pr-o-j.r..cl,, I card o v: I c' i i::cted te
name of the progrcm, follo,.d by whatever specific projrz.,:; paraz,,etgrs and data
cards are rcquired.
The program output lists the mn-ns and standard deviations of all vari-
ables, and the simple-correlation coefficients between all pairs of variables.
The simple linear-regression coefficients and coordinote-axes Intercpts are
also listed. from which the least-square regression lines and corresponding
equations are obtained. By transformaticn of the variables Into lo'jerithrtlc
and square functions and denoting each of these as another variable. nonlinear
relationships, in many cases, were rnduccd to linear relationships c id sub-
jected to the same methods of analysis. The SSUPAC program Is based on natural
logarithms, rather than comwan logarithms. Hence., another transfori,-ation to
the base 10 Is required when using this program. In all, 74 variables were
analyzea in this manner.
The covariance matrices and standard errors of estrimate were obtained

for selected pairs of variables by use of the SSUPAC I'ultiple Ccrrclation Pro-
gram. In this program, it is necessary to specify In advance which variable
is to be indepcndent, and which is to be depcndent. N'umerous other statistical
matrices are included In the printed output of the program.
Each of the data plots presented In this section shows one %est-line"
as determined by the method of least squares. The equation of this line is
given on the face of the plot, along with the standard error of estimate, Sy.x
and the coefficient of correlation. r. The coefficients of correlation are
all significant at the 1% level. In most cases the X-variable is taken to be
independent and the Y-variable as dependent. Thus, the regression equations
and standard errors of estimate are given for "i on X". There are one or two
exceptions in which "A on Y" is given, but these are readily apparent and shown
as sxoyy
3. PHYSICAL PROP.E,TIES
a. Unit weight
The unit weight was the simplest property to determine and also showed
the least variation, equal to 0.5%. Unit weights ranged from the lower value
of 101 pcf for tuff (14.1) to the upper value of l1l pcf for diabase (2.3).
The average unit weight for all rock groups included in this study is 162.3 pcf.
87
The ult cc>rc::.rssive str,:'th of a test specimcen Is defined
herein as tho ratio of tVe l1c_'.d at failure to the cross-sectional
area of tho sccu.rn Leforc tho test. Failure was considered to have taken
place when a sdn &crp in epplicd led was observed and no further load
could be suninortol. Thr mrjority of t!;e compressive strength tests in this
study can bc cliin.sficd as cither 'cr'"or shear failures. The latter is
sometiriu:s t, rd"Ai . l" in wvich a plane of shear is inclined to the
axis of thC s - : (!>r&), l2 .ccpt
. for spccimens cored parallcl to
beddln or felleticn
- (n.g., schist 11.2 end limestone 6.3), "splitting" or
longitudinra fz1lures rarely occurrcd. Even in the case of sandstone (10.2),
cored parallel to leddling, cone failures wcre usually obtained. however, in
numerous instances, I.e.,, for hI~h-strcngth basalts and diabases, the failure
surfaces Intersected the ends of the specimen near the edges. This is an
Indication .that the loading heads offered little or no end-confining
restraint to the specimens. This was verified by placing a set of lateral
SR-4 strain gagas near the end of a specimen, as well as at mid-height.
The lateral strains obtained from those two sets of gages were identical.
By measuring the slope angle between a horizontal plane and the

shear plane, the angle of internal friction, 6, can be approximated from the
Mohr-Coulomb failure theory. For the high-strength rocks, the computed
friction angle was found to be generally about 50. The lowest values of
200-250 were computed for the schist (11.1), cored perpendicular to folia-
tion. Friction angles for intact rock have been reported to range from 270
to 58% with average values of approximately 480 (Deere., 1963a). Theoreti-
cally, in order for the failure plane to be entirely located within the
length of a specimen, having a friction angle of 480 the L/D ratio should
be 2.6. Thus for NX-size cores, the required length would be on the order
of 5 1/2 inches, rather than approximately 4 3/8 inches as used in the
present study.
The effect of rate of loading on the compressive strength was not a

part of this study. According to Hardy (1959), there is no significant dif-
ference In ultimate compressive strength determined by tests on specimens of
similar material using rates of loading of 100J 200, and 400 psi/sec. As
reported by Wuerker (1959), the effect of rate of stres5 application on
concretes was investigated by Watstein (1953). He found les, than a 5%
Increase in strength when tested at 100 psi/sec as compared to 10 psi/sec.
The increase was less than 10% even up to a stress rate of 1,000 psi/sec.
(The same comparison and percentage increases were obtained for modulus of
elasticity.) The average rates of loading used in thepresent study varied
from a minimum of 4 psi/sec to a maximum of 66 psi/sec. Average rates for
each rock group are given in Table 4.2.
An indication of the uniformity of testing procedures and homogeneity

of the rock specimens Is given by the coefficient of variation. A minimum
variation of less than 1% was obtained for Barre granite (5.3), and a maximum
value of 25% was obtained for the Luther Falls schist (11.2);, the latter
cored parallel to the planes of foliation. The average coefficient of
variation was computed to be slightly less than EM for the compressive
strengths measured for all rocks in this investigation.
88
c. S!or P :.
The noneostructivo !orc Lrdn.cs tests, for which a larger nurL;ber of
observations wcere c'.taaIncd tLin for any other test, has an avorage ccefficic.it
of variation of 12-13%. The avwrn-e coefficient of variation for the "as-
cored" sides is not significantly different frcm that obtained for the lacnpcd
ends. In gener., th3 end readn. ars her than th3 sides; tV.
magnitude of this difference vcrs frc:m s to sa.:l.
tole Tho nucr of sid
observatios Is th :r of tLhose ta!,en cn tC'3 end's. Pcat, c of
these factors, c.-A the hic!icr Cc.-js of correlation of the results of sic,
observations wit othr prcpertL.sr, te side rc.dinC; arc uscd in t -hw,gr.. pi,
ical relationships and discussiois prescntcd hcrcafter In this study.
d. Schmidt hardness
Inasmuch as erroneous readings tend to low values (SosmanP 1927), it

appeared to be desirable to eliminate automatically a portion of the observa-
tions obtained with the Schmidt h:cmer. Since the readings are taken uniformly
over all the specimen surface, It was somewhat arbitrarily decid'ed to select
the high 50% of observations as being representative of the Scr midt hardness
for any particular rock group. The average values for all readings., and also
the high 50%, are both reported in the sumniary tables and show, somewhat sur-
prisingly, very little difference. Average cocfficients of variation are
approximately 8% for all readings, and 5% for the high rO0" of readings.
Average values of Schmidt hardness, for rocks in this study. range from 23
through 59.
e. Abrasion hardness
The abrasion hardness tests exhibited the largst variations of all

the physical tests conducted. eO%:3ver, the average cocfficient Is on the
order of 15%, which should be acceptable by most experimental standards.
Unfortunately, time did not permit the modification and perfection of the
equipment for a large number, or extensive control, of the tests. Although
the device performed satisfactorily for this series, prolonged use would
require a more rugged design. The Veeder counter had to be replaced twice.
The biggest problem was to obtain a means for a positive drive system, and at
the same time, produce a constant load on the specimen. The present belt
drive permits a constant load, but the critical balance between starting
friction and vertical load sometimes causes belt slippage. This Is parti-
cularly observed In poorly-cemented, granular rocks, such as the Berea (10.1)
and Navajo (10.3) sandstones, in which the particles ultimately become lodged
between the cut and the wheel, stopping rotation completely. The prototype
apparatus consisted of a right-angle,, dental handpiece, equipped with a
cable-drive system. This soon proved to be unsatisfactory because the cable
drive did not permit the application of a constant load on the cutter, and
also the bevel gears in the dental handpiece were not constructed for such
rugged applications. Therefore, the device was modified as described in
Sectioi Four. In general, the abrasion tests as conducted were surprisingly
reproducible, and gave the overall impression that further tests on addi-
tional specimens of the same rock groups would produce essentially the same
results.
89
F ' 'w * . - - --.. ... .... . .

fbsorptlon v lvcs r'n;-cd fro, .iC% for the Pallsades diebase (2.1)
to a zxrr;uni of 2141 for th-. VTS tuf (O. 0). An average value of 3.2% was
obtained for all rocks tested. l{.cevcr, this has no particular signiflcance
as the rmajority of rocks tested were relatively dense and nonporous.
9.fint-lo~i t-.nsflIc r-nih

The avcrar.eoc:fficioent of variation for the point-load tensile strength
is com.puted to be 12%, es cc-;)rcd to 21" obtained by D'Andrea, Fischer and
Fogelson (05). l.wevcr, the latter performed an average of 17 tests for each
rock, and the results reported herein are obtained from two tests per rock group.
A series of preliminary tests was conducted on specimens of limestone

(6.1), dolomite (3.1), and siltstone (13.1) in order to determine the effect
of end (or edge) distance on the measured strength. Reichmuth (1963) Indi-
cates the use of a distance greater than D/2 from the point load to the end
of the core specimen. The preliminary tests showed quite conclusively that end
distance can be a significant factor in the magnitude of the point-load tensile
strengths. When the tensile strengths obtained for end distances of D were
compared to those of 0/2, they were found to be 15%, 33%, and 38% higher,
respectively, for the limestone, dolomite, and siltstone. As a result, for
the data reported in Table 4.2. only one test was performed on each specimen,
in order to maintain an end distance equal to the specimen diameter, D.
The primary problem in performing this test is in maintaining align-

ment between the upper and lower., hardened-steel rods. These had to be
realigned after each test. Even so, this is not a serious problem as the
test is rapidly performed, requiring approximately 10 minutes ;er test,
including set-up. In addition. because the ends require no preparation, the
tests can be performed rapidly on field core.
4. ELASTIC PROPERTIES
a. Modulus of deformation
Use of the terms, Young's modulus, elastic modulus, compression modu-

lus, and modulus of deformation is normally clearly understood (and in fact
the terms are interchangeable) for those materials which have linear stress-
strain curves. In such cases. Young's modu'us, E, is simply the slope of the
curve, tan e; and also the ratio of the stress, (',to the strain, 6. This is
shown by the relationship on the right-hand side of Figure 5.1. Thus for
linear stress-strain curves, the two methods for measuring E (tangent and
total strain or secant method) give exactly the same results.
The shape of stress-strain curves for rock in compression is somewhat

unusual among the common engineering materials, and may depart substantially
from simple proportionality between a and C. Young's modulus, or the modulus
of deformationY is a function of the applied stress as well as of the material.
Most crystalline rocks have stress-strain curves similar to that shown in
90
4?.C
w U
U)
(j)
91 (
V V!,rc I (rKryn Ir- d JuKI 1W'
lc'7; 12 racc, I_r) This fisure
Is ty-1col of the srztcsi ivotnt - in th:. prcdnt
pre t ty, At low stresses,
thu curve is strongly nonlincar end cc.;-Xnv:- L%':,'n:rd; Young's ri:,:ulus Increasing
as ths stress Is Incrcased. T;, initi tc t nodulus is 9iven by the slope
of tfic stross-straln curve at thL origin, s by line a. Gradually, a stress
bu.i
is rcac~cd whoro the curve becc;%
-- c rosu.ly lincar i.'e., betvween a1 and or2 ).
In this region, the modulus of dzforr;:1ntion is defined either as the tangent modu-
lus, el/e.. or the secant moeulus, alc, w,,hich is given by the slope of the line
OP. At this stress level, the seozint hos a lcuer value then the tangent medu-
lus bociuse it Includes the initial 'lastic" stress history of the curve.
In addition to nonlincar elastic bch-vior, most rocks exhibit an

elastic hysteresis. twevcr, unlike nonlinearity, according to W!alsh (19065),
hysterosis is restricted to uniaxial compression and does not seem to occur
during hydrostatic compression. Under uniaxial stress, the slope of the
stress-strain curve during unloading, PQ. (Figure 5.1), is initially greater
than the slope during loading for all values of stress. As the stress is
decreased to zero, a residual strain OR is often observed. Sometimes the
hysteresis loop closes at zero stress.
Reloading the rock specimen produces the curve RS which has a shape
similar to OP, but is steeper. Further cycles of unloading and reloading to
the same maximum stress produce loops similar to PQRS, but moving slightly to
the right. These effects may be associated with transient creep (involving no
permanent set) or anelasticity, a term used by Jaeger (1962). The rock can be
loaded to failure at T during any desired loading cycle.
The nonlinear elastic behavior of rocks under uniaxial compression

can be attributed to the presence of pores or minute cracks, which at low
stresses are open (Ide, 1936; Jaeger, 1962; W!alsh., 1965). As the stress is
raised, cracks are closed and the rock becomes elastically stiffer, i.e,,
Young's modulus increases with stress. Since most cracks in rock are quite
flat, nearly all are essentially closed by relatively small stresses. Above
these stresses, no change in stiffness due to crack closure occurs and the
stress-strain curve is linear. However, according to Walsh, Young's modulus
in this portion of the stress-strain curve is still less than that for an
equivalent, uncracked solid. This is because sliding of crack surfaces past
each other may take place evei after a crack has closed during the loading
cycle.
The elastic hysteresis of crystalline rocks is explained by Walsh to

be the effect of friction between crack surfaces. Because of friction, cracks
which have undergone sliding during compression do not immediately slide in
the opposite sense when the load is reduced, thus a clockwise loop is formed
as the stress is raised, then reduced to zero. Therefore, Young's modulus for
an uncracked solid can be found from the initial slope of the unloading curve
for a specimen which has been stressed to the linear portion of the stress-
strain curve. This is shown by PQ in Figure 5.1. Walsh states further that
even this modulus obtained during initial unloading is not necessarily the
value obtained for a "solid" sample, since the rock may contain cavities other
than cracks which do not close under pressure.
Typical stress-strain relationships of the rocks used in this study are

shown graphically in Appendix B. These curves show that the behavior under two
92
consecutive Ieodinngcyclcs is es~cnti:.lly a, dcecriL-d Cbovc. Loading to
feilure during the third cycle occurs along t1i, lirne rcpresented by fST in
Figure 5.1. However, since a rainimrmt period of 2 ,cks occurrcd between tho
second and third failure loading cycles, tho residual strain C1 is aSL....d
to be zero. Cecause of the s;.ll order of m cnitude of aR observed in the
present investigation this is probably a valid assuimption for all groups
except rock salt (9.11.
Stress-strain curvos for the 28 rock crc.2s In this study are typificd
by the si:e type-s sho-n in Fi.:ro 5.2. In reality, these consist of four bclsic
shapes, since Types III, IV and V are each modifications of the ropresentativ3
S-curve given in Figure 5.1. Type I is the classical, straicjht-lin3 bohavior
of brittle materials (Jaeger 1062). This is typical of the more exploslve
failures of the basalts, quartzite, one diabase, two dolomites, and the Solen-
hofen limestone. The softer limestones, siltstone, and tuff, exhibit a more
concave-downward, stress-strain Curve, s shewi.n by Type 11. These are
generally scomewhat linear in the early and central portions of the curve)
gradually yielding ',lastically" as failure is acpproached.
The Type-III curve is representative of ten rock groups. These include

the sandstones, granites, two diabases, one dolomite, and the schist cored
karallel to foliation. The three marbles and Dworshak gneiss are represented
by the steeper Type-IV curve. Only the schist, cored prndicular to folia-
tion, has the long sweeping S-shaped curve of Type V.
Each of the Types III IV and V is characterized by initial 'lastic"

crack closing, followed by a definite steeper linear portion. The upper part
of the curves exhibits varying degrees of plastic yield as failure is approached.
Obviously, the Type-III rocks have more explosive, brittle-type failures (simi-
ler to Type I) than either IV or V. The intensity, of course, is much greater
in the high-strength rocks, because of the larger amount of energy involved.
The Type-VI curve for rock salt has an initial small elastic straight-line
portion, followed by a combination of plastic deformation and continuous creep.
As demonstrated in the foregoing discussions, stated values of elastic

moduli may have very little usefulness for a given engineering project, unless
the conditions under which they were obtained are specified. Further, meaning-
ful relationships with other rock properties are impossible unless the stress
level., and the method for computing the moduli values are known, I.e., initial,
secant or tangent at a specified stress level. It may also be important to
know whether they were determined from a first-loading curve, unloading, or
reloading curve.
All of the curves shown in Figure 5.2 except the last exhibit a linear
or straight-line portion in the central part of the curve. This was found to
be consistently the case for 27 rock groups at a stress level equal to 50% of
the ultimate failure strength in compression, regardless of the magnitude of
the strength. For rock sa!t, the linear portion is in the range of 10% of
the ultimate strength.
Elastic moduli are reported for several stress levels in the summ-ary
tables of Section Four. All of these are computed by the tangent method, i.e.,
the slope, dcr/dE. of the stress-strain curve at the particular stress level
93
Typc I Ell,4,V' Z', C.
IN,
1.2 BcSwlt 13.1 SiIfone
Typo 0I Plastic-Eicsfic Typo m P11gsic-Elfctic-Plstic
1 /
10.3 So~d0,r?2 7.1 Maoble
Type R Plastic-Elastic-Plastic Type X Elastic-Plostic-Creop
-. ___ I1. ,"stf 9.1 Rock Salt
R u S FOR ROCK IN
4V.4 A4"I TO FAILUF"2
54
Indlcatc;d. ThIs reduces tlh influence of "strcss filHstory," or initial cffccts
of lar~o strcain at lower stress lcvlsl, and dc1s only with tl'at pcrticn of
the curve under study.
Two mduli of deformtion are ccnsidorod to be sinificant to the pur-

poses of this investigation; they are used exclusIvcnly In all discussions end
correlations th3t follow. These cr tc, initll ;;nt r,-;-dulus, E for tfje
first-loading cycle, and th,. trnL.,cnt nradulus, . c t . ...... at a strc.ss e:>i l to
50% of the u It f;:.te cc.,mprcss7vc strc.,; h dorirnj . .. . to fci lure,. T oe,ii.-
tHal mroulus Is reprcsentcl by line -aIn Figurc 5. ziid is given ct th..- first
stress incrcs.c-nt., 100-150 psi, in Ti'A:l1c 4.3. The tc;v-ent ri'ejulus at 50% , (ul t.)
is represented by line b in Figure 5.1, and Is a -,-d l;proxiation 0of
lus of elasticity for the uncracked rock. Since the tangent m3adulus at 50%
ultimate was measured during the third-loading cycle, at - , this lin3
would normally be shifted slightly to the right. oawever, because of the
elapsed time between the second and third cycles, C% was assumed to decrease
to zero. This shifts the curve back to the left# in the vicinity of line b and
the upper part of the first loading curve OP. It Is clearly scan that the dif-
ference between any of these values is not likely to be significant. The tan-
gent modulus at 50% aa(Ult.) is given in !able 4.3 under the right-hand section
entitled, "Static Elastic Properties (Loading to Failure)." The Initial tangent
modulus is also given for the third-Iceding cycle for co;-parison with the first
cycle in the left-hand section of Table 4.3.
The coefficient of variation for Ei (Ist cycle) is approximately 22%.

This is rcughly twice the variation com:puted for Et at 50% Cra(ult.), which is
11%. Such a relative variation is not surprising as the initial moduli are
usually some,-what difficult to determine accurately because of problcms associ-
ated with specimen seating and initial strain readings. According to Valsh
(1965), Erace has minimized this error by extending the applied stress into
the tension region.
The tangent modulus values at 50% ultimate strength, as computed from

the dial-.n strain measurements, were consistently 25-35% lower than those
obtained by the SR-4 qaqes. This difference may be explained qualitatively by
the probable presence of more numerous micro-cracks near the ends of the speci-
men and, also, by additional seating imperfections which were not removed
during application of the seating load. Both of these potential sources of
variation are obviously included in the dial-gage readings, but would not be
included in the SR-4 gage readings.
b. Poisson's ratio
The shortening of a specimen under axial compressive stress is usually

accompanied by an increase in the cross-sectional area. The ratio of lateral
unit deformation to linear unit deformation, within the elastic limitJ, is known
as Poissolts ratio.
Theoretically, Poisson's ratio may have values ranging between -1 and

0.5. These minimum anid maxirnum values are determined by the following equa-
tions" E
G E (5.2)
95
and K 2/3 G (5.3)
1-2v'
In which E - modulus of elasticity
G - modulus of rigidity (or sh-nr r;;zdulus)
K - bulk morlulus of elasticity
v - Poisson's ratio
In equation 5.2, as the valu, of v approeches th0 value, v -1, the value of
G aproclh.s infinity and Lbzccn:s ncgetiv:z as v pases the vclue, V -1.
flnce, the minlfrLm value of Poisson's ratio is v - -I. Similarly, in equation
5.3, the meaixrmm value of Poisson's ratio is v - 1/2. At this value the bulk
modulus, K, bccoeies inflnite and, for v > 1/2, K is negative (Seely and Smith,
195q).
The nonlinearity of the stress-strain curves has an influen,ce on
Poisson's ratio similar to that which it has on the modulus of deformation.
The numerous factors discussed in the previous section arc therefore appli-
cable to Poisson's ratio. Thus, Poisson's ratio. as given in Table 4.3, is the
tangent to the curve relating lateral and lintar unit deformation. It is
defined by deh/dCv. at the Indicated levels of stress.
The average, initial value of Poisson's ratio for all rocks studied in
this Investigation, is 0.125. At a stress level of 50% ca(ult.), the average
Poisson's ratio is 0.341. The average coefficient of variation for the initial
value of Poisson's ratio is 82%. Unfortunately, this reflects one disadvantage
of the coefficient of variation, in that it fails to be useful when the property
mean is close to zero. The coefficient of variation for v at 0a a 50% %a(ult.)
is 20%, which is considerably lower than the initial variation.
Negative values of Poisson's ratio were consistently observed for two

marbles (7.1 and 7.2) and one schist (11.1) at low stress levels. Although
such values are In agreement with theory, the physical concept of this phenom-
enon is not clearly understood. It should be noted here that other investi-
gators have also observed negative values for Poisson's ratio (i.e., Windes,
1950; Blair, 1955; Wuerker, 1953). Perhaps other parameters are required for
defining the elastic behavior of anisotropic materials, in addition to the
"constants" given by elastic theory.
c. Dilatational wave velocity
According to Rinehart et al. (1961) the propagation velocity, Vp, of

dilatational stress waves in rocks depend! on the following factors: the state
of stress, stress level of the wave, water content, porositv, temperature,
direction of propagation with respect to stratification of tne rock. and tex-
tiare. $m,* of thesp factors have previously been studied quite thoroughly;
others, such as thOe stress intensity of the wave, have not been investigated
es yet,
Tlht lrr"fiu:. of state of stress upon the propagation velocity of P-

itir,:,4,, ,,n v-:, kn ('rocher, 1957; Wyllie, Gregory, and Gardner,
1K1). i U,; Vc'rIc,,"v
ty Increases when pressure Increases. It has been
96
yr, .. ' -'" : '".. "" " "" ........ . ; J....' ."h. '. - -
shown in each of these references that P-,ave velocities, m2asured In the
samne directipi as the applied uniaxial pressure, ancr to be equivalent to
velocities obtained with the same speciman under uniform (hydrostatic) or
triaxial compression of the same intensity. VIhen the velocity was measured
in a direction pnrpendicular to the stress in a unlaxial test, Tocher found
values 10% lower than those measured parallel to the stress. The relative
simplicity of the experimental apparatus and the technique required for
uniaxial compression make the use of this test attractive for many purposes.
As described in Section Four, all sonic velocity tests in this

investigation were conducted under uniaxial cor.;pressicn. The measured
dilatational wave velocities, Vp, for the 28 rock groups at a seating load
of 100-150 psi, vary from about 6,000 fps to a little more than 20,000 fps.
Rinehart et al. (1961) state that for rocks with a well-defined texture the
range of propagation velocity is relatively narrow, not exceeding 15%. They
also state that for rock of the same type, but of a different origin, the
velocities may range over a six-fold interval. Generally, rocks which are
more dense and compact have higher velocities. Those which are legs dense
have lower velocities. In the present study, Solenhofen limestone (6.3) has
a velocity 1 1/2 times that of Bedford limestone (6.1). Danby marble (7.5)
has a velocity nearly 2 1/2 times the velocity of Taconic marble (7.1).
For most rocks, propagation velocities increase with an increase in

stress. When the stress increases from 115 psi to 5,000 psi (340 bars), the
P-wave velocity in Taconic marble increases 125% above the initial vIue (i.e.,
2 1/4 times). On the other hand, for Danby marble, VP increases only 13% for
the same stress increase. The largest effect of stress increase is observed
in the Luther Falls schist (01.0), for which VP increases by 132%. The
smallest effect of stress Increase on the dilatational wave velocity is
exhibited by the Solenhofen limestone, in which Vp increases by less than 1/4%.
Rinehart et al. (1961) report that propagation velocity in rocks

generally increases 10 to 13% above the initial value, when ambient pressure
increases from 1 to 500 bars. Only six rock groups in the present study fall
within this range. The majority (13 groups) fall within the range of 2 to
10% increase in Vp. Four groups, increase by less than 2%; three groups
Increase by 30 to 70%; and two, mentioned previously, by more than 100%.
The average coefficients of variation for the dilatational wave

velocity measurements are smaller than for any other property except for
unit weight. At a stress level of 100-150 psi, V0! is 2.7, and at 5,000 psi
V% Is 1.9; the relative order being as expected.
The increase in Vp with increasing axial stress is concluded as being

the result of the closing of micro-cracks. A transient stress pulse cannot be
transmitted across narrow air gaps without very large loss in amplitude (U. S.
Bureau of Reclamation, 1948). When there is no press-ire on the rock, the open
cracks impede the progress of the wave, which must then travel around the
cracks. This transit time is greater than the transit time of a wave which
could travel directly in an unfractured rock, or a rock in which pressure
closes the cracks (Zisman, 1933; Rinehart et al., 1961). Wyllie, Gregory, and
Gardner (1958) conclude that In order "to obtain simple and meaningful data
from natural, consolidated rocks it is necessary to apply a large (differential)
97
pressure so thnt a tcrr.An1,l velocity is rppr,c!d. Uc'.r t x7c circ.z,tztrcs,
the rfd`Irs tcrJcLT
dn!:.-;"t Lc con eid .rcd j:zrfcctly L 'n lc . pr surcsato
tIc fIZS&and c r ac!,s In tV wn trIx I ..!cr the r,-. 2eurcd v lc.i
oty zprercaI
The fla:arcs in f,,>pcnrdIx s'cwa the effect o,' variation In stress on V

for typical specim-:ns of all rock groups in this study. It is noted that in
general the velocitics r;,:asurcd dur!ng unloading are hi!:ir thrn those measurcd
during loding. Just Cs for the modulus of d fornotion, E, tha rcescn for this
hysteresis effect is pro'ebly due to friction bet:i:n crack surfzccs, which
prevceintts 1lin in Via c.posite scnrie (hOnce, crack c;;cninn) ir.idIatoly aftcr
the ic:di rcc~ucod. The factors affcting th eleltic pr <:ctrs, E and v.
should sirilarly influcncc Vp, as s"e::in by the fol lo,:ing quation :
p-
in which VP
5 dilatational wave velocity In an unbounded madium
E -Youngos modulus
v -Poisson 's ratio
-mass density (2)
9
This equation assumes that the static and dynamic properties are interchangeable.
The validity and irrmlications of this assumption are discussed later in the
chapter.
(2) Water content and porosity
The influence of water content upon the dilatational wave velocity was
not Included In this investigation. As noted in Section Four, all rocks started
from an Initially oven-dry condition, then remained open to the ambient air of
the laboratory for at least 2 weeks before testing. Gain in moisture content
during this period was ir'perceptible. The influence of absorption or porosity
upon Vp Is discussed in a later part of this section.
(3) Temperature
Rinehart et al. (1961) report that an increase in temperature generally

causes a decrease in velocity of from I to 5% per 100*C. at atmospheric pres-
sure. The decrease is attributed to unequal expansion of the crystals as the
temperature rises, which may cause some internal cracking. Thus when cooling
takes place, the crystals may be more loosely joined. At higher stresses, this
effect, disappears as would be expected. The effect of oven-drying on VP was
briefly studied on one specimen from each of 12 rock groups. These are listed
In Table 5.2, and show the percent change in V at 100-150 psi following oven-
drying, compared to V prior to oven-drying. fxcept for Groups 3.2 and 7.2,
the changes agree witg the information given by Rinehart et al. The 21.6%
decrease in Vp shown by 7.2 indicates that this effect might be related to the
very large grain size of this marble and, corresponding, nonuniform expansion
and contractibn. Corrections to the velocities reported for this group at the
lowest stress level may be justified. However, the 8.5% decrease shown by 3.2
is within acceptable limits.
(4) Stratification
The velocity pirallel to the layering and stratification in rocks is

usualiy greater than the velocity perpendicular to the layers. The effects of
98
TA1LE 5.2
DZLAIVIC,2L'- uJ
tPI-IY rjL
Percent
Incrces or
1.2 6-3selt (Little Coc)

01
1.3 NJS El t ( j E., )

2.2 Diabase (Co.gins) -2.2
0
2.3 Diabase (French Creck)
3.2 Dolomiite (Lockport) -8.5
3.3 Dolomite (Donne Terre) " .25
- .6
5.3 Granitc (Darre)
.1
6.3 Limestone (Solenhofen)
-21.6
7.2 Marble (Cherokee)
10.2 Sandstone (CraL Orchard) - .25
10.3 Sandstone (ravajo) -1
11.2 Schist (Luther Falls) II 0
anisotropy, which is characteristic of most rocks, were not a part of this

1 1o.v2:r) a comparison of the Luther Falls schist specimens
(nvestigation.
(11.1 and 11.2), which are from the same sermple and oriented at right angles,
illustrate this effect. The anisotropy factor is defined as the ratio of the
velocity along layers or cleavages to the velocity perpendicular to them
(Rinehart et al., 1961). For the Luther Falls schist this ratio is 3.1:1 at
a stress level of 110 psi, and 1.4:1 at a stress of 5,000 psi. The influence
of the level of stress is seen to be quite an important variable in this case.
The maximum value reported by Rinehart is 1.36:1 for a mica-schist from
Woodsville, Vermont.
5. RELATIONSHIPS AMONG VARIOUS PHYSICAL PROPERTIES
Simple linear-regression lines, relating the more significant physical

properties, are shown graphically In Figures 5.3 through 5.9. Their signifi-
cance and inclusion herein are based primarily upon their usefulness in the
development of index properties for rock. In general, a relatively high degree
of correlation, as indicated by the coefficient. r, is shown.
a. Unit weight and cc,-resslve strength
Figure 5.3 shows the relationship between unit weight and coqyressiva
strength. This figure indicates that the more compact and denser rocks
99
I~ IIVI) ~
Q U
2
C - U
__
w:
- -~17 _
OD 0
I&
-------------------------------- - -w t
CL
CD
-6 w
a0 02 o
ro x id (4gn)oo '44UC74S GAA!SO~dWO:)
100
generally hl.,v' a hlCher str.r...h thz those of leer &7risity, &s wo,1d.
exected. There is, hc,:3vcr, cc-,sidorable scatter frc7, the Icnst-squ; rcs
line. In fact, ahove 160 pcf th!is relationship is not ell-defined. The
most notable exceptions to the suggested relationship are the marbles (7.1,
7.2, and 7.5), which are grouped closely togsther sce distance belowi te
line. The reason for this bchavior is not corrpletely understood althOUC, h both
the strength of the calcite grains and the nature of the bonding bte. rains
are involved. The correlation coefficicnt, r .014 for t e two r
Is essentially identical with thet ob.tained by Judd a.d L'zLr (1l), ,ey
ob tain:,d, r *. 59, for the <rp-, r'ent speci fi c rovi ty a,, cc:Ore sivs strr:"eo-"th
froei thie U%', data. On the otLhr D0 Andrca
Mand, et al. *.(l..) o!)tair, r3 r , .7.1.
The relationship betw-en unit ,,icht and cc:epressi t strcngth is discussed in
greater detail in Section Six, in the development and formulation of index
property designations.
b. Shore hardness and cer2-ressive strenqth
The relationship between Shore hardness and compressive strength in

Figure 5.4 is comparable to the plots shown in Figures 3.5 and 3.6. However,
the relationship cra(ult.) = 300 Sh, as suggested by Figure 3.5, diverges ap-
preciably from the least-squares regression line, defined in Figure 5.4. The
slope of the latter is closer to the "500 Sh" line given in the USOM data in
Figure 3.6. Because of the difference in L/D ratio, the reported strengths
herein should be increased by approximately 12.5,10 when they are compared with
the USU4 data. Closer examination of these. as well as the referenced data,
indicates tha. a curvilinear relationship between compressive strength and
Shore hardness may be more apprcpriate.
The correlation coefficient reported by Judd and Huber (1961) between

Shore hardness and compressive strength is r = .71. This is somewhat lower
than for either the ends, r = .874, or the sides, r = .897, reported herein.
The former data, however, include a greater variety of rock types, many of
which would not generally be encountered in civil engineering projects. The
relationship between Shore hardness and compressive strength is discussed in
more detaii in Section Six, in the development of index property designations.
c. Schmidt hardness and comrressive strength
Figure 5.5 shows the arithmetic relationship between Schmidt hardness

and compressive strength. Although the correlation, in which r - .880, appears
to be quite "good," a curvilinear relationship is very strongly suggested by
the individual data points. Below R - 40, the points tend to become asymptotic
to the X-axis. Above R - 55, the points tend to become parallel with'the
Y-axis. The relationship between Schmidt hardness and compressive strength is
discussed in greater detail in Section Six, in the development and formulation
of index property designations.
d. Abrasion hardness and corm ressive strenth
The relationship between abrasion hardness and compressive strength,

shown in Figure 5.6, is remarkably similar to the USB1, data, plotted in Figure
3.4, even though completely different methods were employed to measure the
abrasion properties. If the strenoth values in Figure 5.6 were increased by
101
1n
IC77 v-.1
~.a~flr,
~a raw~
"C"-7
-el .r!~ ~ ~ st ~ ~ S~~ATf
1111 ij%% WL
0 180
30 -- - -6 -0
20V
to
014.I
0 L1 , 1 1 L I A
0 20 0 soso to Ig
- -
-nn -h Shor
- - - - -
ROC
00 --
INA~!`
- - -- -02
i 1 J
_ _ 1 I.
_ \
_ f:. I
iiD
ZI, to
OD I
r- LaJ
b ow
tx9!d %4n)~uz 227 0:]

V I
300,
c
*l I
0 001toS30 40_0s
Co pasit S WV~.117,q (l),p x1
ROC IN -IIXA
10
12.5q (because of the diffcrence in L/D ratio),, tliese tvo plots wouli be
more similar.
The trend indicated by the rnirbles (7.1, 7.2, and 7.5), schists (11.1
and 11.2), the one granite (5.1), (i.e., divergence frcn the regression lirc )
may possibly be explained on the basis of their grain structure. Althoush
their mineral constituents have average abrasion properties co;:earable to
other rocks having higher strengths, t;he actual strength is governed by Ct--
bond (or lack of bond) betwc:n the individual grains. The coorscr-graarcld
marbles (7.5 and 7.2) dcpart frcm, thz' r'elationship nmore thln the finc>g:
one (7.1). The Caraboo quartzite (3.1), which has a very high abrasion h l-
ness., also falls below (to the left of) the least-squares regression line.
This rock, althcugh welded and made up of fine grains, typically contains
numerous bedding planes and intermittent micro-cracks related to its geologic
history, which lower the strength of the intact specimen.
e. Absorption and corIpressive strenq.th
The regression line in Figure 5.7 indicates that the compressive

strength for rock decreases with increasing absorption (or porosity). This
would be the expected relation; however, the scatter of the data shows quite
clearly that some rocks having low porosity may also have low strength. Over
one-half of the rocks have atorption values of less than 1%. For the semi-
logarithmic plot of Figure 5.7, the correlation ccefficient r is - .429 (the
minus sign indicates inverse relation). 1then plotted arithmetically (rot
shown), r is - .496. This is closer to the value obtained by Judd and Huber
(1961) in which r = - .48. (They do not record the minus sign, but their
scatter-gram indicates this to be the case,)
Griffith (1937) suggests that the averare compressive strength of sili-

cate rocks can be reasonably predicted by the following formula:
8 .000
(ulto) -- (5-5)
in which A - sensible absorption of water by weight, expressed as a rercentage
a-a(ult.) - uniaxial compressive strength in psi.
This curve is shown in Figure 5.7, and is ho~ed to be generally parallel to the
least-squares line for absor-ption values greater than 1%. For lower values of
absorption, the lines diverge. Griffith suggests that one reason for the scat-
ter usually obtained in the region of low absorption (i.e., less than I to 4%)
is that there are no very precise methods of measuring absorption near the zero
limit. For example, a sensible (or measurable) absorption of 0.25% may actually
be an insensible (or undetectable) absorption less than the parametric value 4k%.
Thus., rocks with high strengths, near A OP should be plotted In a region of
insensible absorption (O0 A. < 4%), Instead of as functions of A. Griffith's
explanation seems to be plausible but appears to leave room for personal bias,
and/or judgment, for engineering application.
f. Point-load tensile strength and comire~sive strength
Simple regression between point-load tensile strength and unlaxial com-

pressive strength, as shown in Figure 5.8, reveals a relatively high degree of
correlation. Line A includes all data, in which r - .917. D'Andrea et al.
105
I I.O
1
___ s----___
__ -- -
_ _ _ __COD_
?II
- 0
'0 LcJ-
crz
X >
___~~ _I9. ~w0_
I~~r o-.- u
N.Jb
106
5ht) 20I rT + ,2O I
40
,o . ,------ _ __
,,n I I -!
v 30
3 #0 1
20 /.-.- i *1 / .
.- o_ - _ _'
10 in~ ~ ~ QaU~t)212T3,57
0
5i6It--__
4
-
PO/
-
2.2
-L-'
____ T"i Line 8 mi t 6.3)
S
S1. 4,861 - --
I ; r :.949
0 .4 .o 2 o's 2.0 24
Point-Lood Tensile Strength, T, psi x 103
FIGURE 5.8 RELATIONSHIP BETWEEN AVERAGE VALUE4-S OF POINdT--

LOAD TENSILE STRENGTH A11;'D ULTIMIA.TE COM-PRZSCIVE
STRENGTH
10111 FOn ROCK IN U"JtAXIAL COi PiFiESSkO'Ci'
107
(l1SZ) ruport a better corrclation, in which r - .S7 and sv~ 3,744; psi.
This Is almost one-half thtl stand:rd error of estir-tc showXbj Line A. The
reason for this is prcnbely, in part, due to the n:sth-d of least squares for
"fitting" dota. This method has the disadvantae: of un~desirable weighting
being given to extreme values, because the differcnccs from the mean are
squared. For example, the Solenhofen limestone (0.3) has.a high compressive
strength, but is very brittle and has an unusually lowi tensile strength. If
6.3 is omitted from this relationship, Line B is glver, in which r - .9049 and
syx 4C51 psi. These values arc in considcrlbly better agreement with those
obtained by D'Anddce et al.
Data plots rilating point-load tensile strength with either Shore

hardness, abrasion hardness, or Schmidt hardness, produce a pattern similar
to each of these variables when related to compressive strength. This should
be expected, because of the relatively high degree of correlation between
tensile and comprqsivP strength given in Figure 5.8.
g. Shore hardness and abrasion ha-dness
Figure 5.9 is quite comparable to Figure 3.2, from USOM data, even
though two different methods were used for determining abrasion hardness.
Both sets of these data suggest a curvilinear relationship; however, this
emphasis is stronger in Figure 3.2.
6. RELATIONSHIPS AMONG PHYSICAL AND FIASTIL PROPERTIES
Figures 5.10 through 5.;9 contain plots of aver'ge data, which show
the relationships among significant physical and elastic properties, and
also among the elastic properties themselves. Simple linear-regression lines,
along with their corresponding equations, coefficients of correlation, and
standard errors of estimate are also shown, just as in the previous part.
a. Unit weight and tangent modulus of deformation
The relationship between unit weight and tangent modulus at 50% ulti-
mate strength is shown in Figure 5.10. This relationship is more highly
correlated (r = .784) than the initial modulus and unit weight, for which
r - .619 (not shown). In both cases, however. the deformation moduli for
intact rock specimens gzneclly increase propor-ionally with the unit weight.
Approximately one-half of the rock groups investigated fall within the range
of 160-170 pcf. In this zone, the relation between unit weight and modulus
is not so clearly defined. The relationship between unit weight and tangent
modulus is discussed further in Section Six, in the development of index
property designations.
b. Compressive strength and moduli of deformation
Figures 5.11 and 5.12 both show the relationship btween compressive
strength and the static moduli of deformation. Figure 5.11, on which are
plotted the initial modulus values, shows a slightly greater standard error
and lower correlation coefficient (r - .815) thar, Figure 5.12 (r = .828).
Although neither regression line passes precisely through the origin, both
108
52 .3
A'h 3.23 Sh-
Sy.l= 51.6
r :.858
400
300
t-
"-250 __
___0
6.)
32
5 3
200+ 2o10
o gto 2
.ISI
IiL
20611
c0
01
___
s
50
_
.,_ ,, 131../
2
,310
O0 2 00
Hcrdn*,s, -- id- )
Sp, (S-,
Shore
, OF SHO .RE
5.9 RELATIONSHIP EETWEEN AVERAGE VAL.".."S
FIGUFE
FOR ROCK
HARDNESS AND ABRASION iiARDNESS
109
4 I14
Fit
I 1LU
zo
z 0L
-C -- -0 ~
ow
0 w
_ _ _j
Wu)
0 c wLJ
fin wO
.- - in ".
1100
- - - ~-T
-. 7
CC,
UU
-__
___ _
-
____
_ 17
__
____
_
___ ___
__ _ _
_ _20
__PC
__ ___ C,) ;
f~~u ~L3( 0
41) s
- i
+_ _ _ _
0o
CL
Z(/
.9. w J
01 x- isd '(1n I AI~iw : 40 wo5

V -10 J-t~~w~
I i'I
2;5 u~.
I0 _ _
04
_
0
CL
0
U"C
wI
vN. 1. 5O<,
u >_ 0-~
-.
0
SiW
N LL -
'S it
(D C)
p 3 0 0
z-Jz
Co6
-~ - x
0O0
00
CD
N ;i - c4
.0
F -
0 _jI
112
arc rcsoncbly close. The ratio of th(e irnitial modulus to the compressive
strc'nith, as sho.,n in Figure 5.11, is 365 to '. The ratio of the tangent
modulus (at 5 0 '/, ultimate) to the compressive strength is shown as 300 to I,
in Figure 5.12. Both of these riots show a higher decgree of correlation than
that reported by either Judd and Huber (1961) or D'Andrea et al. (1964) for
the dynamic modulus and compres:ive st'ength. For these, r = .68 and .72,
respectively. The relationship between corpr~ssive strength and the tangent
modulus of ceformation is utilize.d in the dev oremcnt of an engineering clas-
sification system for rock. This system is presented in Section Six.
c. Shore hardness and tanrent mcdulus of dclformition
Boti. methods cf rebound hardness (Shore and Schmidt) employed in this

study demonstrate a reasonable degree of correlation with the tangent modulus
at 50% ultimate strength. Correlation with Shore hardness (which hes the
higher correlation coefficient of the two relationships) is shown in Figure
5.13. In general, the tangent modulus inceases proportionally with Shore
hardness. There is a i.oticeably distinct. vacant zone which parallels and
includes the least-squares regression line below Sh = 70. Up to this value,
the data are at a distance approximately equal to s *x from the regression
line. This relation is also more or less typical ofthe initial modulus for
all values of Shore hardnes-, but this plot is not included.
M. M. Protcdyekonov (1963) suggests an empirical formula relating

Shore hnrdness, SO with the modulus of longitudinal compression, E, in
kg/cm2 as follows:
=15-S 5h kg/cm2 56
E = 1.07 x 106 2 (5.6)
5h
This equation is plotted in Figure 5.13, and appears to represent all the data
as well as or better than the least squares line up to Sh = 60-70. Above this
value, the line diverges considerably upward, away from both the data obtained
herein and the resulting regression line. The data from which Protodyakonov
derived this equation was not presented. Therefore, it is difficult to arrive
at a reasonable explanation for the apparent differences.
d. Schmidt hardness and tangent modulus of deformation
Figure 5.14 shows the relationship of tangent modulus with Schmidt

hardness. This plot resembles the plot relating ,unit wei ht and modulus (Fig-
ure 5.10), however, the correlation coefficient (r = .731) is slightly less
for this relation, but the data are more evenly distributed. Most notable
scatter occurs in the zone between R = 28 to 40, in which schist (11.2) and
tuff (14.1) show the greatest deviations. The relationship between initial
modulus and Schmidt hardnress (not shown), for which r - .705, has a similar
pattern, but the scatter is more pronounced. The relationship between Schmidt
hardness and tangent modulus is discussed in greater detail in Section Six,
in the developmnnt and formulation of index property designations.
e. Absorption and tangent modulus of deformation
The semilogarithmic plot in Figure 5.15 shows the relationship between

absorption and tangent modl'us at one-half ultimate strength. An inverse
relation is indicated, as expected, in which the modulus decreases as the
113
0) E otlaid frcm strcifht-'ir- !
partion v-c curvm, i.e., Ot StrC.3 S
14 c %:,l0030 % 0,(ultl), etccu? fo~r

9A, '- 10%%C0 (utt) ___ __
vs
12
Et1 107,1; Sh + 1.67 x 10'/
2.56 x t0 r I8i
r : 746 / . ...- .
N I
0 " "3
3p5
I
9;
-Protodyokonov (1963)
1-.
0
O0 20 40 60 100 120
Shore Hardn3s, Sh (Sides)
FIGURE 5.i3RELATIONSHIP BETWEEN AVERAGE VALUES OF SHORE
HARDNESS AND TANGENT tkODULUS OF DEFORfrATION AT
STRESS LEVEL OF ONE--HALF ULTIMATE STRENGTH FOR
ROCK IN UNIAXIAL CO4-SZ$SION
114
III
f
IIIi
< UI
~ ~ ~*A.\~lot";
I I~iIl
L: L
Ii f
I~ 94
pjq
CzI-
__0 __ i
___
0*0
0 D 0
- i0-
- l Ion_ __ _ _ _K__
I I ~~ -- _ _ _ C)
.4ra o ~
U.
(0114s) 13lrripor 1 xisd 4uouc,
___
_ __ __ __ _ ~ _ __115J
_)
1----.- z
C
- - - - - -
0
Wi 0
CL >.WI
IonW
~wu
0 v *. =-c()
o oW 0
00
0 **-
0
- - -N W W
c- 2 7
w 300 -
00
2 I -L0
110
xx
I
__ I I~ ~ to
~, I xr I i-j <
w 27
0.
C/
-z ___ * ,:
__ ____ _ _ ____ D
LU
-0a:
01
V7 w D
-o -0
mw.-. C
V 0n
c6 _o
o z
o
z'-
_ - S _
00
il isd '(11n) 00 'Lf8uOJIS OA!SSaidwo:)

"IC
- Ld
(n)
ow t
~0.-
A 'ALL
p- '0
* ~WW
C)
0~0
-d a 0
bU *
00
~
C- w
-i 2
. N I--
CD
'(9l
ODd o'IuBz AU0d
001
ar-, re' sonc ly -Ic;,, . The rat i o.f th nit Jl rt..o ,ulus to the cc)rire!ss vec
strength, es sh>.',n in Fi jure 5.11, is 385 to L The ratio of the tangent
modulus (at 50'/ ultimate) to the ccripressiv strength is show,,n as 300 to I,
in Figure 5.12. Both of these plots shcm a hich,:-r degree of correlation than
that reported by either Judd and Huber (19l6) or D'Andrea et al. (1 9664) for
the dynaamic modulus and corpresivc strength. For these, r = .68 and .72,
respectively. The relationship between co!-.pr~ssive strength and the tang2nt
modulus of deformation is utili'f,:d in tthe deve'o,-m:2nt of an engincering clas-
sification system for rock. This system is present.d in Section Six.
c. Shore h,-ir(r'fss and tanc-rit r:,ordulus of dw:formation
eoti. methods cf rebound hardness (Shore and Schmidt) employed in this

study demonstrate a reasonable degree of correlation with the tangent modulus
at 50% ultimate strength. Correlation with Shore hardness (which hes the
higher correlation coefficient of the two relationships) is shown in Figure
5.13. In general, the tangent modulus increases proportionally with Shore
hardness. There is a i.oticeably distinct, vacant zone which parallels and
includes the least--squares regression line below Sh = 70. Up to this value,
the data are at a distance approximately equal to 5s x from the regression
line. This relation is also more or less typical ot'the initial modulus for
all values of Shore hardness, but this plot is not included.
H. H. Protcdyakonov (1963) suggests an empirical formula relating

Shore hardness, ShO with the modulus of longitudinal compression, EY in
kg/cm2 as follows:
E = 1.07 x 106 kg/cm2 (5.6)

154-S h
This equation is plotted in Figure 5.13, and appears to represent all the data
as well as or better than the least squares line up to Sh = 60-70. Above this
vatue, th," line diverges considerably upward, away from both the data obtained
herein and the resulting regression line. The data from which Protodyakonov
derived this equation was not presented. Therefore, it is difficult to arrive
at a reasonable explanation for the apparent differences.
d. Schmidt hardness and tangent modulus of deformation
Figure 5.14 shows the relationship of tangent modulus with Schmidt

hardness. This plot resembles the plot relating unit wei ht and modulus (Fig-
ure 5.10), however, the correlation coefficient (r = .731) is slightly less
for this relation, but the data are more evenly dAstributed. Most notable
scatter occurs in the zone between R = 28 to 40, in which schist (11.2) and
tuff (14.1) show the greatest deviations. The relationship between initial
modulus and Schmidt hardness (not shown), for which r - .705, has a similar
pattern, but the scatter is more pronounced. The relationship between Schmidt
hardness and tangent modulus is dis-ussed in greater detail in Section Six,
in the development and formulation of index property designations.
e. Absorption and tannent modulus of deformation
The semilogarithmic plot in Figure 5.15 shows the relationship between

absorption and tangent mod,2us at one-half ultimate strength. An inverse
relation is indicated, as expected, irn which the modulus decreases as the
113
IIF ,rc i
tr~- i
portion 7-c curv.3, i.e., ot 'Orr:
IeveI a-,-0 % %(ult), ec:;- f;r _
9.1 I00 % 014(u01)
i
12 .. . .. . . .
E:107,1
Sh 167 x 108
. t:iO7,1I5 eL.iea
I 1 yez
#h
2.56 x 10' r.al '
.
,.
"o. r.746 -
to)
00
'I- - - -
62
Ii) 6 i . . _,,,., -....
Protodyakonow (1963)
00
St141
0 /-1 -1
0 20 40 60 90 100 120
Shore Hcrdness, Sh (Sides)
FIGURE 5.13RELATIONSHIP CETWEEN AVERAGE VALUES OF SHORE
HARDNESS AND TANGENT '.;ODULUS OF DEFORMATION AT
STRESS LEVEL OF ONE-.HALF ULTIMATE STRENGTH FOR
ROCK IN U1IAXIAL COMWR.$SSION
114
J
V
400
SL4
X <
.L I
0~0
'U
__ ~K -''0X
L.. .. fi3AcF
-~
-
~;-j w-;
1
*M
N t7~J
L17 bin-. ~
0 0
6 .
1151
____
C6-
0
___
-
____ ___ __ _ __ _
-IC
J L
C~i US
'N wo
.0 2!
- S. _ _ _ _ I _ _ _ _
*0 w
CL_ _ _ -- - - _ _ _ _--
-+
-~~~~~~
-
- _ __ -w~
I no>
V 0- 1----
4b0
n'U
3) 1 snnOtDu 01xld
a ______ _ _ __
4116
absorption (or porosity) increases. This relation is better than the one for
the initial modulus values,, for which scatter is so broad below 2% absorption
as to obscure any consistent relationship.
Hamrol (1961) describes the use of absorption as a parameter for indi-

cating the degree of alteration or weathering of rocks, which is termed the
quality index i. He presents data relating the modulus of elasticity and the
quality index (or absorption) for extremely weathered granites. Because of
this correlation and its simple determinaticr, Rocha (1964) describes the use-
fulness of the quality index as a means for -o,trolling the ,depth for founda-
tion excavation in rock. Also, it can be used during the exploration of
extensive areas for accurate zoning in rock surveys in order to reduce the
number of in-situ tests.
The information given by Hamroi and Rocha impressively demonstrates

the usefulness of absorption as an index of deformation quality at a particu-
lar site, for a given rock type, subjected to the same general conditions of
alteration and weathering. The data, presented in Figure 5.15, seem to indicate
that this relationship may not be so clearly defined when several rock types
from different sources are all included on the same plot.
f. Unit weight and sonic velocity
Sonic velocity, at a stress of 5,000 psi, is shown in Figure 5.16 to

be higher for rocks with higher unit weights as expected. This is not the
case at low stress levels, in the vicinity of 100-150 psi. Whereas there is
no detectable difference in unit weight at low stresses, the sonic velocity
is strongly influenced by open micro-cracks, as previously discussed. There-
fore, rocks with unit weights in the general range of 160 to 180 pcf may have
velocities from 6,000-20.000 fps at low stress levels. As the pressure is
incrca-c to 5,000 psi the velocities increase and shift to the right, result-
ing in the relation as shown.
g. Sonic velocity and moduli of deformation
Figures 5.17 and 5.18 show the relationships between sonic velocity and
the static moduli o4 deformation for two different str-ss levels. The least-
squares regression lines defined in these plots have remarkable parallel
slopes. However, the regression line in Figure 5.18 is shifted to the right
and downward from that of Figure 5.17 because of increasing the stress level.
The combined changes in velocity and modulus, resulting from increased

stress, are more clearly shown in Figure 5.19. In general, rocks in the lower
half of this plot show relatively large increases in both modulus and sonic
velocities when the stress is raised from 100 psi to 5,000 psi. These have
stress-strain curves (Types III, IV, and V) which are initially concave upward,
and have initial modulus values which are lower than the tangent modulus at
50% ultimate. Exceptions to this are shown by 3.1, 9.1, 13.1, and 14.1.
Rocks in the upper half of the plot show very little increase in velo-
city, but, in general, show a decrease in modulus when the stress is raised.
These have stress-strain curves (Types I and II) which are initially convex
117
ow
--.- P_ t-,!- +
I 0~
Ci
- 6. x, ++
(n
C.
010
'Z" "0 U
W
<
a
o') - 0
LL~
_ '
O 3~0
CC
.0I0 , N
- --- 0>
0->
0 CD
t)I (D 1
0
4 A -----
- - (.0
jod '*A( '145ioMf i!urn Ajo
118
1-, /,
Note: 2.3
0i Doto from Ist looding cycle
14- 1
I
4_I _ _ - - - - _
/21
E1 :1,006 V1P-8.05 x 10_
s,:, 1.97 x o f u
r :.910
1 2 -... + .
"".011.2
to
03
0. 0
'i-
6 2
o .2 /a
4.'
10.2
.;.2 e3 .2
.2
1.3
16 20 24
0
4 8 12
0
Sonic Pulse Velocity, V., fps x 103
FIGURE 5.17 RELATIONSHIP BEETWEEIN AVERAGE VALUES OF INITIAL

MODULUS OF DEFORMATION AND DILATATIONAL WAVE
VELOCITY 4'.FR ROCK UNDER UNIAXIAL COMPRESSIVE
.si
STRESS OF 100-150
119
I ) VP from 1st locd~ng cycl~ for a. x 5,000 psi,
e~xccpt for 9.1 and 14.1 mo a 1,500 p~i-
2) E, oli l rm to,7 -nc portion a'-c curv:- IS
i.e., at stress :,cvc o-, :50'e' '%,(Llt), oxccpt feo
14 9.1 10 a'3 10t
-2 SY-1-- ~1.96 x 0a s
00
Cx 10
z 112
C:
10.1
ItoI
OL2 14-- 1O - 1 1
04 a 12 16 320 24
,son'ic Pulse VeIncity, Vp, f ps x 10
FIGURE 5.18 RELATIONSHIP EETWEEN AVERAGE VALUES OF TANGENT
MODULUS OF DEFORMIATIOtN. AT STRESS LEVEL OF ONE-HALF ULTIMtATE
STRENGTH AND DILATATIONALt WAVE VELOCITY FOR ROCK UNDER
UNIAXIAL COMPR~ESSIVE STRUESS OF 5,000 p~i
120
.Key :El of 0 % o(ult),
14 /.
Vp ot ct=5,000 psi J
12 Ea 8 V. from Is? looding cycle--

a?.. . -.150 psi
100 22
S,,,
W0 0
53'
6I
10?2 4 __ __
32
2 . 1.
2-
III
0 4 8 f2 le20 24
Sonic Pulse Vo~ocit) Vp, fpPS XI10
FIGURE 5.19 VELOCITY- MODULtJS VECTORS FOR FIOC~t UNCDER
INCREASING IJ9"Q1C!AI`L C071272CONE STREMS
121
uptvtard an)d t!,-.rc:forc, have h11i-.r initial tx*dulus values thcn tho tanSent
modulus at ui.r,te.
ut Exceptions to this are shown by 2.2 and 8.1.
The dividing I in-s, frr the bothaviors described above, np;ear to be

in the vicinity of.E = 9 x 10 psi and Vp 16,)000 fps. rocks having initial
moduli end vczlcities belo4i threse values are in the "lower half" of the FIlot,
and rocks for t-,Hch initial vwslucs are greater than these limits are in the
"Upper hslf. " 1,, parcntly, 3.1, 5.2 and 7.5 are bordarline ccpto-ions to these
general i"cr"s.
In accord with thoory, the data in rigures 5.17 and 5.10 indicate that
a curvilincar relationship would provide a better fit. Even so, the coeffi-
cients of correlation for the linear regression lines shown are both fairly
high, r - .910 and .8SO, respectively. The possible reasons for a lower
coefficient for sonic velocities at a stress of 5,000 psi, as plotted in
Figure 5.18, are as follows:
1) The sonic velocities for rocks of which the strengths at 50%

ultimate are less than 5J000 psi, are higher (as plotted hereon)
than they would be at a stress level equal to 50% of their
ultimate strength. Therefore, these rocks plot further to the
right of the regression line than they "should."
ii) The sonic velocities plotted for rocks, of which the strengths
at 50% ultimate are -greater than 5,000 psi, are lower than they
would be at a stress level equal to 50% of their ultimate
strength. Therefore, these rocks plot further to the left of
the regression line than they "should."
It Is apparent from Figure 5.19 and the discussion in paragraph 4.c. that the
errors, resulting from the above sources, in the procedure used herein for
comparing data at the higher stress levels (5,000 psi)., are probably less than
5%. This seems reasonable because the percentage increase in VJ, for the
majority of the rocks, is less than 10% fot the entire stress c ange from 100
to 5,000 psi.
7. COMPA.RISON OF STATIC AND DYNAMIC PROPERTIES
Application of the theory of elasticity to problems relating the propa-

gation of elastic waves through rock with the static deformation characteristics
is a subject of long and continued discussion. Elastic theory has been used
for derivation of the following well-known equations:
E V2 (l~v) (1-2y) 05..7)
M 2pV2 (1+V) (5.8)
G z2 (5.9)
122
2
1/2 ( -)- (5.10)
1)2
in which E - Youngis nmculus of a1sticity of the rn'tcrlal

Vp = dilatation:l wave velocity for an unbcindcd Gicdium
Vs W distortional or shear wave velocity
V- Poisson's ratio
p - mass density (7)
G- modulus of rigidity
Hence) E may be calculated from the above equations by measuring both VP and
Vs during dyDmic sonic-velocity tests. E may also be computed from stptic
stress-strain tests by use of the procedures described in paragraph 4.a.
Unfortunately these two methods do not always yield the same results.
During static load application to a rock specimen, the mic.o-cracks

and crevices becom- com;acted. Even though the amount of compaction is very
small, Ed, determined from wave velocity measurements, increases as the com-
pressive load is increased. Et, determined statically as the ratio of incre-
mental stress to incremental strain at various mean stress levels, also increases
with the compressive load, Both Ed and Et will approach a constant value for
the particular rock specimen (for reasons previously discussed); and this value
should be the same, regardless of w.cther or not it is determined statically
or dynamically, provided the rock is truly elastic (Onodera, 19G3).
Statically obtained values of E for rock are almost always lower than
dynamically obtained values. According to Rinehart et al. (1961), these dif-
ferences may vary from 0 to 300%. Zisman (1933) and Ide (1936) found with few
exceptions that the dynamically determined values of the modulus are fron 4 to
20% higher than those statically obtained. They conclude that the cause of
these differences is the liability of static measurements to uncertainties from
(i) the presence in the rock of minute cracks and cavities, and (ii) the rock's
nonlinear elastic response to static stress. The more conpact the rock, the
more nearly do the static and dynamic constants agree. The two methods give
the most divergent results for rocks possessing low moduli of elasticity.
Dvorak (1957) found for a medium under pressure that the differences
between dynamic and static modulus do not exceed 50% because the fractures are
closed by pressure. Data given by the U. S. Bureau of Reclamation (1953) show
that the ratio of dynamically to statically measured Young's modulus varies
between 0.85 to 2.9. Sutherland (1963) reports the percentage difference be-
tween the sonically and statically determined vzlues of E to vary from 4 to 25%,
with the sonic value greater than the static in all but two cases. lie state.,
that the large percentage difference obtained by some investigaturs may be due
to the use of total stress/total strain (secant modulus) to calculate E,
instead of incremental stress/incremental strain (tangent modulus). Thus, the
effect of the initial high strain is present throughout the test, resulting in
low values of E.
123
Gregory (Vi,3) anC .l~h (lV1) c;xplaln in a qunlitative wy tfI lack
of agreement b:t:.:z static and rw.iic values of Ycunq s roo.lulus, even tlvough
reasurcmarts are n at the s;:c! level of acpplfed stress. The pessev. of a
sound w:,ve corrcfw.-cnds to the svp:rposltion of a smalt altcrnating stress on
the existing, e;"plicd comprcssfvc stress, as shown schsrr,,-iacaily in Figure 5.20.
The static value of E Is taken as the slope of the stress-strain curve at thic
selected strest-lvel, P. On the other h3nd, the dynamic nmodulus c,'jlculatcd
from the sonic vwlocity corresponds to the average slpce: of the unloading and
rebound loo, cro-t1t,-.d by the su eri:,w.-d sound wave. pis show,,n, the static
valu Ould bs c......ctd to be ...... t lo;,:cr tlan tlhe dyro,;ic value, %whichis
corwparrble to , b in Figuru 5.1.
Ide (1936) concludes that the discrepancies found between static and
dynamic values of E are real, and that most rocks are, therefore, imperfectly
elastic media. They arc not sufficiently homogeneous and isotropic to meet
exactly the conditions for which the formulas from elastic theory were derived.
In order to compute dynamic values of Young's modulus, E, from equation

(5.7), it is necessary to determine v, as given by equation (5.10). Dynamic
values of Poisson's ratio, v., require measurement of the shear wave velocity.
In the present investigation, a technique for producing and measdring shear
wave velocities under axial load conditions was not satisfactorily developed.
Some of the problems encountered in measuring S-waves are discussed by Gregory
(1963).
A comparison of static and dynamic propertiez, however, may be conducted

on the basis of the "constrained"8 modulus of deformation, which is defined as
the rate of change of vertical stress with respect to vertical strain, under
conditions of zero lateral strain (E - C 0). The constrained modulus, 19
requires only the measurement of theYdilafational wave velocity, V , and the
mass-density, p, given in equation (4.3), and restated here as fol~ows:
Mc o i r p Dynamic (4.3)
This modulus Is related to Young's modulus and other elastic properties by
the following equations:
Mc = E [ (l+ -'2V) Static (5.11)
- K+4/3 G (5.12)
in which K = bulk modulus of elasticity.
The theoretical relationship for the ratio Ml/E for all values of
Poisson's ratio is given by the curve in Figure 5.21. The average values for
Poisson's ratio, as determined by the tangent method, for all rocks in this
investigation are also given thereon. For the first and third loading cycles,
the average v = .125. At a stress level equal to 50% ultimate, the average
v = .341.
In Figures 5.22 and 5.23, data are plotted relating the constrained
modulus, computed from velocity measurements by equation (4.3), and Young's
modulus from static streas-strain measurements. These are given for the stress
124
CL
U) <
0-
C)
0:
0A~I 'ss.l
125
II -
_Lw IL
0
C
C) V
II0 _
~IIx
cr0
cD ai R
U126
-G V;.1
i In
1110i~ LL
I
~~0 _ 0t&
If I I I
-OJOS le 1L .i1i IA _______
__3 '-~nc
I 11~~127\ ,I
- - - trn- - - - - , -z
U0
C,,
-J 0
q) 0
-0 - -~ U..
*c0
a -o
-~~ -0, Z U
~~..b 0 C
0. D
~ 0 cfl.... 4- (D
W 0
"_
CL - r /,w
- -
IV . ~ - - -- -D
0.0 0 00 4 ~
0s (CLS
'Q 30 'Slfl0~ WUb~
128
ranges of 100-150 psi and 5,000 psi, respectively. The theoretical relation-
ship between M. and E for the average Poisson's ratio, v, obtained at each
stress level, is shown by the dashed line in each figure. These are taken
from the theoretical curve in Figure 5.21. It is interesting to note that the
two regression lines, defined n Figures 5.22 and 5.23, converge at an Mc-inter-
cept of approximately 1.6 x l0b psi. Since Mc is a function of VP, it would
appear that the propagation velocity of P-waves in rock tend to approach a
lower, limiting value, regardless of the stress level (within the range in-
vestigated herein). This is probably a function of the structure of rock,
the degree of cementation, grain contact, et cetera. The lowest limiting
value approximately equal to 1,130 fps should occur if there were no grain
contact or cementation; this is the velocity of sound in air at 1 atmosphere.
Logically, such rocks would have no coherence. The lower limiting velocity
from these two figures is approximately 6,570 fps, computed for the average
density of 162.3 pcf for all rocks in this investigation.
By plotting the dynamic measurements for M., computed from equation

(4.3) as abscissas, and the static measurements for M.., from equation (5.11)
as ordinates, a more "normalized" comparison between static and dynamic proper-
ties is obtained. These data are given in Figures 5.24 and 5.25. Static data
computed for initiai values, E. and, v. at stress levels of 100-150 psi are
shown in Fisure 5.24. Dynamic values of Mc also are shown for the same stress
levels. The correlation coefficient, r = .949, suggests this relation to be
quite good. It is noted that the ratio of dynamic and static values of Mc,
as defined by the regression line, decreases with incre.'ng modulus values
(this was the same observation made by Ide). The more compact rocks approach
a ratio of 1:1, whereas the lower modulus rocks diverge to a dynamic/static
ratio of 2:1 for Mc (dynamiC= 3 x 106 psi.
Onodera (1963, Fig. 3) presents data relating static, E, and dynamic,

Ed, for rock pieces under no-load conditions. A line drawn through the center
of gravity of his data points, parallel to his Line A, is defined approximately
as follows:
E S 1.0 Ed - 1.422 x 106 psi (5.13)
This equation is remarkably similar to the regression equation given in Figure

5.24 as follows:
Mc (static) = 1.11 Mc (dynamic) - 1.92 x 106 psi (5.14)
In Figure 5.25 the dynamic M was computed at the maximum stress level
of 5,000 psi for all rocks except 91 and 14.1, which were at 1,500 psi. How-
ever., since this stress level was approaching failure for rocks with strengths
less than 10,000 psi, the static Mc was computed from Et and v values at
stresses equal to 50% ultimate strengths for these rocks. For all others, Et
and v were computed at 5,000 :si. This manipulation is probably a partial
explanation of the lower correlhtion coefficient than that shown in Figure 5.24.
In Figure 5.25 the ratio betweei? dynamic and static values is more constant
for all rocks. This varies from ',5:1, for Mc (dynamic) =3 x 106 psi, to
about 1.2:1, for Mc (dynamic) = 16 :-106 psi.
For a dry rock, saturated with iir, the dilatational wave velocity may
be calculated from the static elastic constants by means of equation (5.4).
129
Note:
I) Constrained modulus, Mc (dynamic) = pYV

2) Constrained modulus, Mc (static) E,
3) Data from let loading cycle
28
SMc stat.) = 1,11 Mc (dyn.) - 1.92 x 100

foo 20 . 1.71 x Ie psi. -
g " . r,..949 J2.1
___6_ _ 2.3
______ ______
______ ~1.21____ ____ ___
1.1
2.2
"12 -"
A 6,e.2
5.3
C6,1
3.1
'7.5
6.1, /.'
4- - 1 -4 1 1 1
9.1 3.2
.;2
C1 0-_ 40. -' - 62

0 i 112
Constrained Modulus, Mc, psi. x ICtDynamic)

FIGURE 5.24 RELATIONSHIP BETWEEN AVERAGE VALUES OF
CONSTRAINED MODULI OF DEFORMATION COMPUTED
FROM STATIC AND DYNAMIC MEASUREMENTS FOR
ROCK UNDER UNIAXIAL COMPRESSIVE STRESS
OF 100-150 psi.
130
32
Note:
I) Constrained modulus, Mc (dynamic) pV
2) Constrained modulus, Mc (static) : Et.E,,tr-
3) M (dynamic) computed from I't loading cycle
for oa,: 5,000 psi., except for 9.1 and 14.1
28 oo: 1,500 psi.
4) Mc(stotic) computed from I1t loading cycle for
o'az 5,000 psi., except 6.1, 7.1, 7.2, 7.5, 9.1, 10.3,
11.1, and 14.1 which were computed from 3 rd
loading cycle for a,= 50%(r[ult)
24
L~
Cl) M (stat.) = .851Mc (dyn.)-.405 x le
S20 Sy.x : 2.49 x lIe osn.
X r = .817
o 3.3
3.2 :.2 2.1

Z .2
oa .1
5.2 102.3
C 07.1
0 2
011.2
In - -
.
4-13.1
10. 10.2
O01 8 4A 5 1 2 2
0 4 8i2 16 20 24
Constrained Modulus, Mc, psi. x I10(Dynoiic)

CONSTRAINED MODULI OF DEFORMATION COMPU'TED
FROM STATIC AND DYNAMIC MEASUREMENTS FOR
ROCK UNDER UNIAXIAL COMPRESSIVE STRESS
UP TO 5,000 psi.
131
Wyllie et 0i. (15[) cc:?{utcd tlD vwlocity for j:rca
sn.dstone from th3 values
of E and v obtained statically at diffcrent strcss levels; these were compared
to the rnmsurcd values of V at tLr san, stress IcvIs. Similarly, tho average
velocities cn-,:utod by cqua~ion (5.4), for all rocks in the present Investi-
getion, arc plotted as ordin:etc;s in FV srcs 5.25 and 5.27, The rmasu_rn
velocities are plotted as abscisszrs for co:paris on.
A high dcorcn of corralntlon (r * ) for the "static" and dynemic

velocities, obtarncd by thes tvw r:,thf!s, is show:n in Figure 5.26 at the
lower stress level of 100-150 psi. Cote that a high degree of correlation
(r approaching 1.0) does not mann that there is a 1:1 relation between static
and dynamic values. It means that the data plotted in this figure are wall-
defined by the least-squares regression line shown thereon. Again it is seen
that the more compact rocks approach a dynamlc/static ratio of 1:1 whereas the
less compact (hence lower velocity) rocks approac:i a ratio ol 2:1.
The degree of correlation in Figure 5.27 is somewhat lower, probably

for the same reasons as given for Figures 5.18 and 5.25. For the high levels
of stress, in Figure 5.27, the dynamic/static ratio for all rocks is more
constant than for low stress levels as previously noted. This ratio varies
from 1.3:1 for the lower velocity rocks (8,000 fps) to approximately 1.10: for
the higher velocity rocks (20,000 fps).
Wyllie (noted above) plotted the data for aoth measured and computed
values of Vp versus stress level up to 5,000 psi. He found that the dynamid
static ratio for the Berea sandstone, at 2,000 psi, is approximately 1.3:1.
At 5)000 psi, the ratio was about 0.9:1. Both of these values are in close
agreement with the Berea sandstone (10.1) in the present study.
A summary of the observations made in Figures 5.24 through 5.27 is

given in the following table:
TABLE 5.3
PERCENT BY WHICH DYNAVI:C PROPERTIES ARE GREATER THAN
STATIC PROPERTIES FOR ROCKS
Dynamic-Static
Static x 100%
Less compact More compact

Low stress (100 psi) 100% 0-3%
High stress (5,000 psi) 30-50% 10-20%
On the basis, of these data, the following conclusions are made:
i) The magnitudes of dynamic and static properties of rocks are

equal only for compact rocks at low stresses.
ii) Since compact rocks in general have slightly convex (Type I

or II) stress-strain curves, the initial modulus may be some-
what higher than the tangent modulus at 50% ultimate strength.
132
V 3jV I I II
4) V (cA r. C( ' f~ t
10cdn CC! fc,TC.,= 100 - 1 si

pi i
24--
dI i
el 10./
o 4 1LL2 2
Sonic Pulse Valocity, Vr,, fps~ x 103 (mnourcd)

FIGURE 512,3 r~AI~ PC~UJ JJ ZVALUZS 0, P'r-2'
WCCWVLCJ'IT C0:UTLD", NICM STATIC AN~D D~i t2a C0
L~E~K7Z7SFO~R3C LIN"C UNNAXIAL 17~:
STRESS Oil 100 -150 psi
133
OV,"
foC C, u
Al p
9 Sri= 2,145 fps

rI
'44.1
04 a~ 12 06 20 24
Sonilc Pulse Velocity, V., fps x 10,(~r~
FIGURZ 527 REAIC'?`ZXPC.'v~ I f(Ir.RtSF VALtf:30C;'MLATATI CX1.
WAVE VELC1ITY COIWUTJ MCM> STATiC A'D DYWT.`6C
STF,-:-S UP TO 5,000 vn~i
134
Therefore, at high stress levels, the magnitudes of the
dynamic properties are greater than those of the static
properties by 10-20%.
iii) The difference between the values of dynamic and static

properties of less compact rocks decreases with increasing
stress.
iv) Less compact rocks do not appear to meet the basic assumptions
for which the theoretical equations relating velocity and
Young's modulus were derived.
In this section the experimental data obtained in the current investi-

gation have been presented in graphical form. Predominrnt~y, data plots were
given for those properties of rock which indicated a relatively high degree
of correlation in simple regression analysis, and thus shc.ied a reasonable
potential for use in the development of index properties. In the section
which follows, a system is presented for classifying rock in terms of its
strength and deformation characteristics. Also, certain index properties
are proposed which are formulated from the relationships developed herein.
135
SCcT&:-N
Cru~sslffction of in-situ roc!.. fr.......

GaCCtMZ 'L% -.Visic"71 p -.. cc
cszOf I rcnck u~lfiln o
roc , , v;d
ca tlhe rn,':;or c;;d phyzicnl c .. ct.: Istics oi .......
,, tics v:hIch Lcv.md th" blocks. T... Cc
dIsc ,r_..... cc' irv:d ,".'., .'d:te .rn-s tho
enrinc-crIn; L2.evior of the rock mess in terris of strci' ,,;, f orm. II I ty,
and porr;zei 1 ity.
The present study is concerned only with the proPerties of the intact
rock material as determined by tests of representative rock cores drilled
from the joint blocks. It is advisable to devclop a classification system
based upon the physical properties of the intact rock before attempting to
consider the adverse effects of geologicol discontinuities.
In this section a system for classifying intact rock is introduced

which has as its basis the strength and deformation characteristics of the
rock. Certain index properties are also proposed which are developed from
the relationships given in Section Five and which can be used to estimate the
uniaxial cc%?rcssive strength and the modulus of deformation. The index
properties w.ay'also be used for assigning a rock to the proper engineering
classification. Statistical methods of analysis are utilized for determining
least-square regression relationships between the suggested indices and
corresponding strength or modulus properties, along with the confidence
limits for dispersion.
2. PROPOSEID EtNGINEER ING CLASSIF!CUATIOV
In classifying rock for engineering purposes it is desirable to have

a relatively simple system of classification with a small number of categories.
The classification should be based on significant physical properties so that
the system has wide application and is not limited to a specific problem.
An optimum classification system should also provide good reproducibility,
and the classification should not be a function of the background and
experience of the person making the classification.
The system of engineering classification proposed herein provides

qualitative descriptions in terms of two engineering characteristics,
strength and deformation. The rocks may be classified by using the actual
laboratory-determined values of the uniaxial compressive strength and modulus
of elasticity, or by using the approximate values obtained from the index
property correlations. A su-.mary of the proposed classificataon is given
in Table 6.1. For a complete description it is necessary to use both the
strength class (A, B, C, D, or E) and the modulus ratio (H for High Modulus
Ratio, L for Low Modulus Ratio. and no suffix for an Average Modulus Ratio,
as described below). Thus. a rock may be classified as BY BH, BL, et cetera.
136
TiA i$LE 6..1
L:.;::?7;. c. r
. .c-,
C Ic:.7:::: F:i' I:';i; n".....
rCi t
VlI S tesrCn n c ) =
ax a ;o ''.,,, r...::i ....
AVery 1 Str
111:T Ov*:r 3? ,C.-
C Kldium $ttic.ngth 8,00 0D , C1C3

D Low Strength 4,000-0.00
E Very Low Strength Less-than 4,000
11. On Pesis of rodrll.,,s Wd:tio
Class Description nodult.is rttoI
H High Itodulus Ratio Over r03

- Average Modulus Ratio 200-500
L Low Modulus Ratio Less thin 200
Classify rock as B, C!11 BL, etc.
i) Mlodulus :aio t/ca(ult.)

where Et = tangent modulus at 50! ultrmnte strength;
Oa uniaxial compressive strength
The strength classification is similar to the one given by Coates

(1964), shown as Table 2.2 in this report, except that five cate ories of
strength are proposed rather than three. The proposed Class E - Vary Low
Strength, for compressive strengths less than 4,000 psi, corresponds approxi-
mately to Coates' Weak Category (less than 5,000 psi). His Strong and Very
Strong categories (with the dividing line at 25,000 psi) are replaced in
the present system by four categories: Class D - Low Strength (4,000-8,000
psi); Class C - Medium Strength (8,000-16,000 psi); Class B - High Strength
(16 000-32,000 psi); and Class A - Very High Strength (greater than 32,000
psi 5 . It will be noted that the limiting values in the strength categories
increase in geometric progression.
The division between the High Strength rocks and Very High Strength
rocks at a compressive strength of 32,000 psi has been selected on the basis
of three considerations. First, the relationships between the com-pressive
137
t~~~~ i:Yc2, Lijre 5i :)
(r'. cV
I .dt hre~;`Cs
(i:~in wJ)s c t/. S inri, 2nC!T, c~I" t %ty I cn :c T.~s 1s rc Ia
tI nY tv L; ,s, irc,
I, t I nn t- ro c"-.s wE i cc css
'i,,In Cc~r-
src.~,:: of ~t Psi hv tin intoc.f~L
d occ
1hich c-1,"-,t L,_ :,rci by hfreiczss tests, but wh-ich ccc.nts for vc-ry
hi~h cCT:;.rscssv. strc:rvtts.
A sencccr. :; lc~~rc,tIon concerns the sh:pcs of thc stross-stroin

curves; for c1rI: ;t cll rcc's-rctcr
w!th cc.;cIress.ve s trc inC. tIn 3c
psi.* ttl:,. curvc r:Lly lin.!r (Figure 5.2,1 Tc ,nd shT,
-
brittkd~f,!u, c::rzctc~ristlcs
- &,- clvcnn in Coats ' cL.iflcetien. A
third r,,J p~rcly prcctlcal co, si.:,r-tion is that the Very NIt;h Stronnth
igneous rockl:s -- dic,.sc and don:,c bcsalt -- are si,.,;cratad by the 32,003-psi
line frcA other fine-grained igneous rocks such as dacite, rhyolite, and
andesito, as well as from most rocks of the granite fansily. In addition,
almost all foliated) mztamorphic rocks (i.e., gneisses, schists, some
phyllitos., and slates), and the ccmz.,on sedimentary rocks fall below the
3 2 ,000-psi line.
It was originally believed that the classification on the basis of

modulus could usn various sub-d visions of the actual nmodulus values, e.g.,
less than I x 106 psi, 1-2 x 10 psi, et cetera. However, as the system was
developed it became apparent that each strength category would have three or
four rnodulus categories and that fifteen to twenty classifications would
result. Of even greater concern was the fact that on the classification
plot of mnodulus versus strength the categories plotted as rectangles with
any given corner comnmon to four rectangles and therefore to four classifi-
cations. This situation resulted in as many as four classifications for a
suite of samples from a particular sample site even though the actual varia-
tions in strength and modulus values were quite minor.
In order to circumvent the above difficulty, the proposed classifica-

tion system makes use of the modulus ratio which is defined as the modulus of
elasticity (tangent modulus at 50% ultimate strength) divided by the uniaxial
compressive strength. The relationship between average values of the com-
pressive strength and tangent modulus of the rocks tested for this report is
shown in Figure 5.12 as an arithmetic plot. The plot indicates that the data
may be fitted approximately by a straight line. the equation of which is
Et - 300 cra(ult.). The slope of this line is the modulus ratio, whose value
in this case is 300.
The proposed classification chart, entitled Engineering Classification

for Intact Rock, is shown in Figure 6.1. Both the modulus and strength axes
are logarithmic, in order to provide for a wide range of value. The strength
categories A through E are indicated. Although the 300:1 modulus-ratio line
is shown in this figure, the classification system makes use of a zone (shown
by stippling) which extends on either side of this average line and which is
bounded by lines representing modulus ratios of 200 aid 500. These limiting
values were chosen so as to encompass the majority of the rocks with inter-
locking texture, e.g., diabase, granite, basalt, quartzite, rock salt, and
some limestones, dolomites, and sandstones. Rocks included within this zone
are considered to have average modulus ratios, and their classification
138
Note:
I) E = tangent modulus of 50 % ultimate strength
0 2) Classify rock as B, BH, BL, etc.
E
-
., 4 8 16 32 psi 1 10
' E D C B A
C VERY LOW LOW MEDIUM
HIGH VERY HIGH
6 LSTRENGTH STRENGTH STRENGTH STRENGTH STRENGTH
I-II -1
0_
IZ2 3.5
..
6
. . .
_
- ,---.. - - - I-- --
4 --
0 2-/
100
0 l
I ,_ ---- _-- f-
_7.O
.* - -- ---
I3 4 5 6 78 9 0 20 30 40 50 6OphiKxI0O
SII i I -I I
7'5 125 250 500 '000 2000 4000 kg/cmz
Unioaxiol Compressive trn h, ro0(ult)
FIGURE 6.1 ENGINEERING CLASSIFICATION FOR INTACT ROCK-

AVEVAGE VALUES FOR 1=3 ROCK TYPES
(28 Locations)
139
-t I
consists only of a strength category, i.e., A, B, C, et cetera, without

reference to the modulus ratio. Rocks with modulus ratios greater than
500, as noted in the figure, are of high modulus ratio and their classifica-
tion would consist of a strength category followed by the, letter H as, for
instance, BH. Similarly, rocks with modulus ratios less than 200 are
designated low modulus ratio and their classification includes the letter L
as, for instance, CL.
The plotted positions of the data points in Figure 6.1 give a strong
indication that the classification system may be a very workable one, and
one that will take into account both mineralogy and textural fabric. To
illustrate these aspects broken lines have been placed around the data points
representing given rock types in Figure 6.1. The lowermost envelope around
points 10.1, 10.2, 10.3, and 13.1 encloses all the sandstones and siltstones
(13.1) tested in the present study. Several trends are apparent. The enve-
lope is unique in position with respect to the other rock types. Although
the envelope follows the general trend of the stippled zone (average modulus
ratio), it is closer to the 200:1 limiting line. The position indicates that
the sandstones are more compressible with respect to their strength than
most rock types. This is considered to be a result of the closing of small
micro-cracks parallel to the bedding planes. The bedding was near horizontal
in most of the specimens and the axial load was applied normal to the bedding
direction.
The envelope around points 5.1, 5.2, 5.3, and 4.1 encloses all the
granites and granite gneiss (4.1) tested in the present study. Again' a
rather unique position is indicated. The granite textures are essentially
equigranular and interlocking and the plotted positions are parallel to,
and in the center of, the zone of average modulus ratio. Anisotropy of
fabric is not indicated. The position of point 5.2 in the C-strength cate-
gory is anomalous but these samples were of a slightly weathered granite.
The envelope for diabase (2.1, 2.2, and 2.3) is essentially an

extension of the granite plot, but in a higher strength category. The basalt
envelope (1.1, 1.2 and 1.3) is in the A strength category but it is located
in the lower portion of the average modulus-ratio zone. The dolomite enve-
lope (3.1, 3.2, and 3.3) plots parallel to and near the boundary between the
zones of average modulus ratio and high modulus ratio. Two of the limestones
(6.1 and 6.2) plot similarly but the third (6.3 - Solenhofen limestone) plots
well away from the trend. The marbles (7.1, 7.2. and 7.5) plot in a restricted
area in the zone of high modulus ratio. The majority of the carbonates, then,
have a trend toward a high modulus ratio. This fact is felt to be a result
of an interlocking texture and low porosity as well as the particular stress-
strain characteristics of calcite and dolomite.
Attention is also called to the schist samples (11.1 and 11.2).

Sample 11.1 was tested with the axial load applied normal to the schistosity;
it plots in the zone of average modulus ratio. Sample 11.2 was tested with
the axial load applied parallel to the schistosity; it plots in the zone of
high modulus ratio. The effect of textural anisotropy is dealt with further
in the discussion of Figure 6.3.
140
The data points in Figure 6.1 are the averages of given rock types
from specific locations. Each point is the average of from 3 to 8 individual
tests. In Figure 6.2 all test results are shown although the individual
points are not identified. Thirtcen rock types and 128 specimens are repre-
sented. The plotted points further emphasize the validity of the concept of
a zone of average modulus ratio.
To investigate further the effect of anisotropy a series of tests were

run on samples of a sandstone, which showed pronounced bedding, and on a slate.
These samples, referred to as Indiana sandstone and Virginia slate were not
part of the systematic study, and the only tests conducted on them were
strength and modulus determinations. The test results are shown in Figure
6.3. The number beside each point represents the angle between the specimen
axis (direction of loading) and the plane of anisotropy. The effect of the
anisotropy created by the bedding planes is not great for the sandstone
although both the strength and modulus are affected to some extent. The
classification designation changes from C for the 900 loading to D for all
other angles.
The anisotropy caused by the slatey cleavage has a very pronounced

effect on the classification of the slate. The designation ranges from B to
C, CH, and DH as the angle between the specimen axis and plane of anisotropy
changes from 90' to 300. At angles less than 300 both the strength and the
modulus increase; designations of CH and BH result at angles of 15' and 00,
respectively. It is apparent that samples tested perpendicular to the direc-
tion of anisotropy, or within 45 or so of that orientation, show a lower
modulus ratio than those obtained for samples in which the anisotropy is
steeply inclined (small angle between the specimen axis and plane of aniso-
tropy). The strenqth classification is also a function of the angle of
anisotropy. Minimum strengths are found in the 3V0 range although random
low strengths are often obtained whenever the anisotropic feature is steeper
than about 450
A summary plot showinS the classification of various igneous rocks

is given in Figure 6.4. Three rock groups are shown: diabase, granite
family, and basalt and other flow rocks. The granite family includes syenites,
monzonites, granodiorites, quartz diorites, et cetera. The flow rocks in-
cluded with basalt are dacite, andesite, and rhyolite. The envelopes shown
ai, those that include 75% of the points for the rock type in question. A
tot,-] 3i *76 test results were used in preparing the plot. These included
the results of the present study as well as values obtained from private
reports aod from the literature.
The individual points for the diabase plot are shown in Figure 6.5,
those for the granite in Figure 6.6, and those for the basalt and other flow
rocks in Figure 6.7. The envelope enclosing 75% of the points is shown in
each case. The envelopes have been drawn so as to omit about the si.re number
of points on each side and yet maintain a parallel or sub-parallel aTignment
with the trend of the average modulus ratio zone, wherever this could be
done without undue bias.
The tight grouping of the 26 specimens of diabase from 8 different

locations (Figure 6.5) is quite striking. The samples have modulus values
greater than 10 x 106 psi and the majority have compressive strengths greater
141
Note:
I) Et tangent modulus at 50 % ultimate strength
2) Classify rock as 8, BH, BL, etc.
0 5
4 16 32 psi O
x 10
E D C B A
VERY LOW LOW MEDIUM HIGH VERY HIGH
STRENGTH STRENGTH STRENGTH STRENGTH STRENGTH
20 .1_
t- n a W
4 -
2; 4-
2 0
0-
2 ,
bl
-.
- - i/
4-
- .-.. S""- ...
. .......
7 "
C" 2_ - --
3
K ,
. FE I
7 _25-
5678910 20 30 050 60psizx1
75 5250 500 1000 2000 4000 kq/cm2
Uni axial Compressive Strength, oa-((ut)

INDIVIDUAL TEST RESULTS (124 Specimens,
13 Rock Types)
142
a
0 r
P~Y LOV' LOW UQ J1 HL

"_
__ _ _
01
............. ! 1.... i
i3 -7
......
2 0
S- ,. . J_
.... J_ _ - -I J -4 18J
,__ -_ - , I._ 2
7 AJ 0 A U
2-
0__ I
CAt_ 2
__ - 4 Indiana Sondston
J.A/
.
___ I _ IZLI7_
>"J ..... L.. .--
4- Ii-
- 4
8~ I44 -I_
2 3 4 5 6 7 8 9 10 20 30 tO 50 60 ti a10
I I I ! ..!I I
75 125 250 500 1000 2000 4000 kg/cm
Unioxiol Compressive Strength, o0(ult)
FIGURE 6.3 ENGINEER;ING CLASSIFICATIONj FOR INTACT ROCK-

EFFECT OF ANGLE 0 MEDDlI;G AND FOLIATION
143
C'a
6- 0 "lGl r t
I I
77 LLL.~'j
EsF
3 LI 5..J.7
-J-
4 *103 0 0
-ncxc Co. r~:vo tegh,0.ut
3IWN . EG-ZERN
SUNIA
CLS 14F
7LOTIGNOS ROCK (17
O J7TOiJ
RC
NLfC
75%I [7
YA'Imns Poits
14
Cl ::i,
E2) r C;
'cc: LL,
, Ci, I tC.
10i4 4IiII 32
U K
- _
of PointsI
a iv. 75% I (2 , -
o ' o" I........I....."_
7- ---
4 . ... . L. .J . .. _i
0 2-
= . "I
I
2 -=
..- --_.I !
0 10 4 -
4 _7 00__4 _
2 3. 4 56 78a9*10 20 3 0 O5O 6OjaiuaII
75 125 250 500 1000 2000o 4000 kg/cm

Unioxiol Comprossive Strength, 0(ult)
-o
FIGUF EC.5 ENGINEERING CLASSIFICATION FOn INTACT ROCK-

DIABASE (26 Spocimang, 3 Locations, Various
Sources)
145
0 8 Clcz:'if rct-1 c 0, C1, U!,, c-1
Is IE 32,
'~~~~~"On ri! z,
J---'v ~%Wtl 4! "A
r r ~ '
"-" 75 % of Points I1I

- o......- I
oo _ _it.,1j
0 a0
2J
0 -
w0 _
4 '4LjI/.h 1o '
2 - .6 1-
I I 'I II
6 ~ - - ~- All -- __
02 a I l J I' I ,I
52
Vaiu z Coasurrinc,
- -
ltr) _
12 3 4 5 6 78a910 20 30 40 50 60 Pei aIc
75 125 250 500 1000 2000 4000 kg/crf
Unioxiol Comproa2s ive Strength, co0(ult)
FIGLU~fW C.6 ENJGIN4EERING CLASSIFICATIONJ FORl IN'TACT lC-

GRANITE FAMILY (0O SpocirimLoc 1
VcrioU3 Sourcef)
146
Nl j
0. U..,
t. I
0 Andcs~itc, Dacitc~, 3~ R'l i.1If,

c I
S --- 75 % of Points
a- t
--
w 5 _.J
4 4b
* ~~~J -,E,' 3 2jI-E I 717
9 -'
.25 J -. I- _ I I' ai
75 25250 Soo 1000 2(>30 4000 tug/cm

Unioxiol Cornpressivc, Strenoth, .'G(WO~
F1GUFtZ 0.7 EllNZ~EVRING CLASSIFICATIONJ FC?! HINTACIT POCQ1.

BASALT AND OTH'~ER FLOW R=*:10 (70 Spoc-komen!
20 Locatiorn, Various Sourcca)
147
thm' 35 001 :
psi
Nrr c3 ~ '~~ rcc.-, v.Ith ern
r,:xlhd us rtto-- or cttc r: y A. f, cr:it. f;: Iy ro:. (:i sure ,6I) plot
chIefl y as ZCca tc-,,ry rocZ':s alrth!tt n ... 0d c10'rs, 'r1 or altored
sp~c -nsplot in the C c u,It F~not,:r flow rocks (Figure
6.7) show the gratcest rzare,! in val,. cs o:7:ld e' ct:,r:ctcd because of the
wide ronse in ml rneraloy, tcx:turc, a h ty. The strength classifi-
cation ranges frci A - Vrry Itish Strcri(th for the dcrn.st basalts to E - Very
Low Strength for thV w'c,,!cst s,-7plcs. "ii r-,,.jority of the flow, rocks, [::w-
ever, fall in, tha C and C strCrnc . T 72evolope coir,;icros
Tlks. 7e
rceso.:l y V2 ll with t!3 7c:.o of r J
iC tus rot io.
The SfAV,--t,-,v re.,k sur;,:7 nry plot is giwvn in Fic;ure 6.8. Thia rock
types Included are Il rir5tcn3 and dolo ito sondstoncl crnd shale. The
envelpes for sandstone and shale are not cl),sed at their lower ends
because a great number of test resultg indicated strengths less than 1,000
psi and modulus values belo:w 0.3 x 10 psi and could not be shown on the
present plot.
The positions of the envelopes of the sedimentary rocks are quite

distinct from those for the igneous rocks. The limestone-dolomite envelope
extends well Into the zone of high modulus ratio. This position seems to
be a response to both the texture (interlocking) and to the particular
mineralogy (calcite and dolomite). The detailed plot of Figure 6.9 shows
that the majority of the specimens fall in the B and C strength ranges as
BY BH C, or CH. A few points extend into' the A and D (or E) categories.
The sandstone envelope (Figures 6.8 and 6.10) extends into the low
modulus ratio range. This position is considered to be the result of a
cemented, porous texture and the anisotropy created by the bedding. The
modulus values are low because almost all the samples were tested with the
core axis perpendicular to the bedding. This orientation gives low modulus
values because of deformation caused by closure across bedding planes. As
may be noted, most of the samples plot in the C and D strength ranges
although points are present in all strength ranges.
The detailed plot for shales is shown in Figure 6.11. The shales that
plot on this graph are referred to as hard shales to distinguish them from
soft shales (clay-shales and poorly cemented silty or sandy shales) that plot
well below the limits of the present diagram. The dividing line has been6
taken as the tangent modulus value (at 50% ultimate strength) of 0.3 x 10
psi. The corresponding strength value is approximately 4,000 psi. Thus,
hard shales may plot as B, C, or D strength rocks, usually with the suffix
L to indicate their low modulus ratio, while the soft shales all plot as E -
Very Low Strength rocks, and usually as EL because of the low modulus ratio.
In the figure it may be desirable to show more of the Very Low Strength
range and to further subdivide the E-rocks. Test data, which would plot to
the lower left of the present classification, show that the soft shales extend
into the soil groups (stiff glacial clayey till, loess, residual soils, et
cetera). An interesting point to note is that the soils and soft shales
both follow the trend of low modulus ratio, i.e., from 300 to values as low
as 100 or so. There is -vidence which indicates that some of the lower values
may be the result of sample disturbance. However, the lower modulus ratio is
considered in general to be a result of anisotropy caused by the orientation
148
I fcuE t iI
0
5 2 OI xI10
VERY LOW LOV! r' 7F l
I1
6S
2-
3- Shale
Srjncstorio -
Ii1
-- -. -.- A. _dP- -
___ j - I _ ...L~ _ LJ.LJA
0 2-K~7 .j
L14 J' ~ I
4- -
55
123 4 56 7 0910 20 30 40 50 60 Pei x0d
?5 15250 500 1000 2000 4000 hg/cm
Unioxiol Compres~ive Strangth, o0(ult)
FIGUr,, C-0 ENGHNW:ENG CLASS"IFlCATION FOR IPT/%ZT (0"'OK~

SU.~Y PLOT SEDI~raENTARY OS(I
Specirkic~no, 75 % of Points)
149
2)-
VER~Y LOV LCO :wv.i
Io
cm, i2Ia
Do i g
-- 75~ of Points IIK I ~~
0- 2-
A,1
2 2 4 6 '7 91_0
_ 3 0 5 oV
ID N111E(7Soi~s
.5'ETCII
- 22 cdos Vaiu
___
Sou , I
- - - ~ z150
Noeo
'i :II In
I t
ci..~
I VEVr'Y LrG,,J CV. "r :i.
I II -1 - - - .I
4 b l.L ?!L)iiii
2 -
1L1
.__J ~ -' 11 -
2 3 4 5 6 708910 20 30 40 'i.O 60 031a 10
Is 1520 500 1000 W 4000 kq/c#M
UnicxiczI Compressive Strongth, o0,(uit)
FIGUr,, 0'.10 ENIGINIEE~t,! CLASSIFICATION MR~ INTACT ROK

SA'PDSTV2',- (02 Spschncn, 10 Location,
151
g211
"" 0
p, C I I ;. (, I
UE
".Itr ,
0-- VERY LO'i _, LOW
..... .. . . ,.,..
IC I I
I " I,.
I C.STR -I
i -- 7% f Point
___
.__,__,____'f
4- ~ . -IT
iL. -__
S... - .-........
.
-
% 1iL..:" j , I .
_I_______
2 - 2 _ / _ ----. -t- _,_,d-
3I 1
* 00
.5
pteghCo0p1,iv ut
Un0 xo
4 Z- AL --
! I 40 50 60psi
67 8910 20 30
.25L2 2 3 4 5
250 500 1000 2000 4000 kg/
75 125
Unioxial Comnprozsive Strength, o0 (Jult)
FIGURE 6.11 ENG1tlEERING CLASSIFICATION FOR INJTACT ROCK(-

HARlD SHALE (54 Spocimans, 14 Locations,
VCO~tUG Swrces)
152
of particles parallel to the bedding during deposition and diagenesis. The
bedding is normally horizontal and the samples usually have a vertical
orientation. Thus the axial loading is perpendicular to the laminar struc-
ture and a low modulus resJlts.
The results of 44 specimens of porous tuff are given in Figure 6.12.

The natural water content of the samples is in the general range of 20-30%.
The average strength is shown to be about 1,500 psi with an average modulus
ratio. Upon air-drying the tuff to approximately 5% water content, the
strength is increased to values of 2500-5500 psi. Although the tuff is
bedded, it is cemented by secondary clay minerals and zeolites and behaves
more like a uniform material of interlocking texture than an anisotropic
one and, consequently, plots in the zone of average modulus ratio.
The metamorphic rock summary plot is given in Figure 6.13. The

scatter in results is greater than noted for the other rock types because
of the great range in both mineralogy and degree of anisotropy. The rock
types summarized are quartzite, gneiss, marble, and schist. Figure 6.14
indicates the quartzite data. As would be expected, the modulus values are
high -- uniformly above 7 x 106 psi, and with more than half exceeding
10 x 106 psi. The strength values range from High to Very High and the
samples plot as A, 8B or BH. It is interesting to note that the dense,
compact rocks with equigranular and interlocking texture -- such as dense
basalt, diabase, and quartzite -- plot in the A category in essentially the
same position.
The gneiss plot is given in Figure 6.15. The gneisses are predomi-
nately granite gneisses and they plot in about the granite position but with
a somewhat lower average strength and more scatter in the modulus ratio. The
additional scattering is in response to both a greater variation in mineralogy
than is the case for granite and of anisotropy in the form of lineation or
foliation. The several points which appear above the average modulus ratio
zone as BH and CH probably represent premature failures along schistose bands
of samples with steep foliation (low angle between the specimen axis and
and plane of anisotropy).
The plot of the marble specimens is given in Figure 6.16. The

envelope includes all the samples of the 3 marbles tested for this report
(15 specimens). These samples were all of building-stone variety but they
were from three different states; the fact that they all fall in such a
tight grouping is surprising. The other 7 samples were from either construc-
tion sites or mines and the scatter in results is apparent. A greater number
of test results from several localities are needed before a meaningful plot
can be obtained. It does appear that the high modulus ratio shown by many
of the samples is consistent with the trend of limestone and dolomite which
contain the same minerals.
Perhaps the most interesting plot is that for schist, shown in

Figure 6.17. The mineralogy of schist is extremely variable with both the
kind and percentage of platey minerals such as muscovite, Liotite, chlorite,
and talc varying widely. Also, the degree of development of the highly
schistose bands may range from slight to high. When one considers the other
important variable that enters into the test results, i.e., the angle of the
153
Note:
I) E, tangent modulus at 50 % ultimate strength
o 2) Clossify rock as 8, BH, BL, etc.
4 i. 16 32 psi i 10
, E D C A
AB
VERY LOW LOW MEDIUM HIGH ERY HIGH
bTRENGTH STRENGTH STRENGTH STRENGTH
TRNTSTRENGTH
201 I
"- . Natural Water Content
a Air Dry
S -- 75 % of Points
I
7 'i
.~~ . i vlI1
4-
w 7-
) ' -- -1 ;'0 / /
3b
o 2 -4
00
00
A&01' 0
.5- , -
5 12 25 50 0020 00k/m
12 3 4 5 6 78 910 20 30 40 50OS0pui 1 1
II I
Unaxial Ci Strength,
1 (ult)

POROUS TUFF (44 Specimens, 2 Locaticn~,-
Various Sources)
154
Note
1) E, =tangent modulus a? 50 %. ultimate strength
0 ~2) Classify rock as 8, B31-,BL, etc.
4 8 16 32 psIi 10I
SE Q C BA,
Ck VERY LOW LOW MEDIUM HIGH VERY HIGH
16 -STRENGTH STRENGTH STRENGTH STRENGTH ST.RENGTH
20C - Quartzite
2 - Gneiss i
3 - Marble0 0
e4- Schist
10F_ (a) Steep Foliation - L 0
g j....... (b) Flat Foliation- - -
4- 6
2 4 -
0 2
I 2 3 4 56789~10 20 30 4500bx0
I I I 1I
75 125 250 500 1000 2000 4000 lq/cm'
Uni cxial Compressive Sirength, co

0(uflt
FIGURE 6,13 ENGINEERING CLASSIFICATION FOR INTACT ROCK-

SUMMARY PLOT METAMORPH IC ROCKS (167
Specimens, 75 % of Points)
155
I) Et --f ..c . , .,cntt = ,J + at , 0 u lt im at o tf rC r, t I
a2) C..Ify rczk c , CH, DL, Ot:.
e'a 4 0 16p-
-- U I T ?"
- -- 750/ of Points
___ _-_ .-- __/ _.[
4 40
4- -- -
0 -- /' . ......- -
.83- - ,.,-, ,,,- _ ___ _

.5 --__ o +' ___ -- -
0' 2-~~
7 -______ LLK.
en4 ... -"--- --i -- "--' - - -
"----- -i
-. T-I 4p
. ... .
c!I I _ _ ,
w155
.25 --
2 34 56 7 8910 20 30 40 50 60 psi
75 125 250 500 1Q00 2000 4000 kq/

Unioxiol ComnpresivS Strength, 0o
0 (utt)
FIGURE 6.14 ENiGINElE-R-"G CLASSIFICATION FOR INTACT ROCK-

QUARTZITE (41 Specimcne, H!Locations,
Various Sources )
156
I) Et tc r.,n2mC,. oI SO"I vItmn ot I r -rC
%
02) Clc:ify reek~ or, , CA1, C-L,c.
20
-- 75,% of, Points I II '
-- " .
In
.4
/ fj
9 _ _ . . "-w .--- -- -
.6 --. e
7i
2 5 4 5 6 7 8 9 0 20 30 40 50 60 pi aI
75 125 250 500 '000 2000 4000 kg/c.
Unioxial Comprozsive Strength, 0-o(Ult)
FIGURE G.15 ENGINEENNhG CLADSiFICATION FOR INTACT RGOCK-

GNEISS (43 Spzcir;,mns, 14 Locations, Various
Sourco)
157
2) clc'-f ro'c c3 8, CA:, CL Vc'
4 0 16 32 psi
bIVfIIIt l
C idiiStone
tT 3 Locations
Zh%'
of Pointsf
- -I
4- -
J1.4JJ
-..
-fL
CA 2
C /0-
- , 2 4 5 1 2 0 4 5 0
95
Inoxo Copesv
44:L 5_Stegh
12 20 _000 50
- -
200400_
FIUR .1 EGNERNGCLSIFCAINO INTC ROC
Li:Iz 1:1:8-
__ A BL (2 oainp -PCh1-s
__
LOY,
C) ~1 l~f DIT l L rC4~

IC,
f I 0
0 t.
.25 A.j
- -
2 3 5 67 a 10 0 30 40 5 60 i l
75, 125*--
c~S~7~ (S
Fflciaxi
2w
omprcLmiVi
50 10020
KLL~th
II IfIt 40 g
IG 1 FG? ENG,',lEEF4iolG
FURa V5COASWHA) FO INAC LOII
8C IS Pointsom
(-71 1f VaI
Sourca s
4J Y~l~~kL~..159 ..
schI*
I! iSt Vw tcA,:I r :7 t tn t 7::,is) It s r t sur; rI.i ,- tlt
th' scihit cl I f I t"i 1 .I; I'; 1 Of tL 15 catctyri s (no sci.st plotted
in the thrc.-. fl:..... o.... i t -jory) * Tho eCfcct of nnole
of schistosity On tL. t:.;t r tps Lccr: ,rc~nt -p a wv1L:n tfo s:' with
,,
stcc foliation arc s'v'r . froni those with flat foliation. Th latter
grovupIr~q cxtond's into th_- IO rirZulus ratio zone because of tre- effact of
closure of th, rmicro-cracks pzFrallel to the foliation. The sto-,p-foliation
grouping ct-n~ds up into thb zcroe of high r;iodulus ratio indicating thnt scoe
fairly hird, h"~h;ra!u~us rocks (5-10 x lOb psi) failed along a schistose
band at relativcly loc-r str:ses.
Referring a3,;in to Ficure 6.13, which surmierizes tLa mc.tcm-orp4hic rock

classificaticns, It Is sianificant to note that the gneiss cnvc1clc.e overlaps
the quartzite plot and th3 two schist plots. This transition position indi-
cates an increasing coorplcxity in both mineralogy and texture fabric as one
goes from quartzite to gneiss and then to schist. The summary plots for the
igneous rocks and the sedimentary rocks (Figures 6.4 and 6.8) show similar
responses to differences in mineralogy and textural fabric.
The proposed engineering classification for intact rock is considered

a workable and useful systeri. The classification is based on the conmpres-
sive strength and modulus of elasticity -- two important physical properties
of rock that enter into many engineering problems. The classification is
also sensitive to the mineralogy, textural fabric, and direction of anisotropy
of the rock so that specific rock types fall within certain areas on the
classification plot. These areas as shown on the present summary plots will
no doubt be subject to some modification as additional test results become
available.
3. INDEX PROPERTIES
Index properties provide a quantitative method for assigning a rock

to a particular classification, independently of the background and experience
of the person making the classification. Usefulness of index properties to
engineering classification is generally indicated by (a) their relationship
to a material property used in solving a design problem, (b) their simplicity,
low cost, and rapidity of determination. and (c) their reproducibility.
Of the ,arious properties considered, the Schmidt hammer hardness,

Shore scleroscope hardness, sonic velocity, and unit weight show greatest
promise for serving as indices of the engineering behavior for intact rock.
The relationships between Schmidt hammer hardness and the strength and
deformation characteristics of the rocks investigated herein clearly show
the Schmidt hardness, R, to be in accord with the requirements for index
properties as stated above. The Shore hardness is also presented as a pos-
sible index property for strength and deformation, but has a slightly lower
degree of correlation than that shown by the recommended Schmidt hardness;
it is also more restricted to laboratory usage. Another index of rock
deformation is the sonic pulse velocity, which has a high degree of corre-
lation with static modulus. However, the sonic velocity does not satis-
factorily meet all of the requirements specified for index properties
because of the more costly, specialized nature of the laboratory eouipment
needed for its determination. A fourth index property, unit weigh, is used
160
In conJun!trin u".. er ch of t fi rz t thrcoe i 1' rcs P In ore,,r to prcvoIc,
nu:-:ins for d! rit Ih t,,,tirg rc c-1, v: !. t-: sv; .....
tI- or sc i c vd.I c... ty,
but wh i ch hL; C1,ffc rcnt v I u s of s trcnst or r . 1
Five c'L-,rts are prcscntc-d in this section which have as their Lccs
the numericzal v:,1uos given eit",r by the Sc-.,midt hzzr, the: Shore sclcr-,
scope, or thL-; sonic pulse velocity, e.ch in conjunction with the unit tl?7;2rilht
of thc rock. Ita su stdi....x proerties are utili!ad for estlnm-tirn
the co,,'?rr;ssive strcn~th or static moculus of doforrm,-ition of Intact rcc!,
sar,iples. Ti.icsc vclues r,:.qy tLC:.n be used in obtnining an
L:Troxiet cla,; fi'
cation for t . rc 2.
Statistical rrcthdt of analysis, as dcscrilk-ed in f,7cndf, C, arc .'scd

for obtaining t:,3 functional relationships, standard errors of estii'mte, Cnd
correlation coefficients betw,,een the indices and the strength or deformrc-'tIon
characteristics of the rocks. The confidence interval (and corresponding
dispersion limits) for prediction purposes with respect to rocks is also
discussed in this section.
4. UNIT ' A Ij
6PT!
.S OF STTTUf') f.2T7
The unit weight of a material is one of the most basic properties

obtainable in engineering materials testing. W.hereas the total weisht nmay
be relatively easily determined with an accurate balance or scale, the unit
weight requires also the determination of the dirmeonsions, or volume of the
material. For a specimen which has a regular, geo:a trical shape, the
volume is easily determined. For irregular specimens, a procedure similar
to that used with the Jolly balance may be employed. This involves weighing
the specimen in air and then again while immersed in a liquid of known
density such as water. For a cylindrical core specirmon which has irregular
broken ends, but which also has a known or easily r :esured diameter, the unit
weight can be accurately estimated by measuring its total weight and closely
approximating its length. Obviously, as familiarity with weights of rock
and rock core is established, the unit weight itself may be estimated with
sufficient accuracy. Thus, it is clearly seen that the determination of the
unit weight is a simple, economical and relatively rapid test procedure.
The reproducibility of the test results in the laboratory is excellent, as
shown by tht average coefficient of variation, V% - .5, given in Table 5.1.
The reproducibility under field conditions will depend upon the methods
used for its determination.
a. Relationship between unit weight and compressive strength
Anrarithmetic plot relating dry unit weight and uniaxial compressive

strength is given in Figure 5.3. As indicated by the regression line in this
figure, the more compact and denser rocks generally have a higher strength than
those of less density. This relationship is, however. not well-defined in the
range of 160-180 pcf as these rocks are noted to vary in strength from
approximately 8,000-45,000 psi.
The data in Figure 5.3 suggest a more curvilincar relationship than

given by the least-;quares regression line. 8elow IlLO pcf, the data indicate
a flatter, and rrc. asymptotic relation to the X-axls. kbove 160 pcf, the
161
data curve upward and tend to indicate a steeper relation than shown by the
regression line. Thus, although the coefficient of correlation might be
improved for curvilinear regression, it is seen that the unit weight by
itself is not, for all values, a sensitive index of compressive strength.
However, its use in conjunction with other index properties for differenti-
ating strength values is shown to be quite significant in the following
paragraphs.
b. Relationship between unit weight and modulus of deformation
The relationship between unit weight and tangent modulus of deforma-

tion, at a stress level of 50% ultimate strength, shows a higher degree of
correlation (r - .784) in simple linear regression than unit weight and
compressive strength (r = .604). Just as for strength, the modulus increases
with increasing unit weight as shown in Figure 5.10. Some samples, however,
appear to be anomalous, i.e., 8.1, 9.1, 11.1. and 14.1. In multiple regre:-
sion., in which the unit weight is used as a coefficient for other index
properties, the correlation with modulus of deformation is considerably
higher. This is discussed in detail in the paragraphs which follow.
5. SCHMIDT HARDNESS AS AN INDEX OF STRENGTH AND DEFORMATION
The Schmidt test hammer provides a simple method for determining

relative rebound-hardness values for rock. Test data can be rapidly obtained
from either NX-size exploratory drill core, or block samples, without special
preparation. The cost of determining the Schmidt hardness is quite low
because of the rapidity and ease of test performance. The initial cost of
the instrument is approximately $135.00 (inc-luding metal sheath and leather
carrying case). Reproducibility of the test data is excellent, as indicated
by the low coefficient of variation (approximately 5%) given in Table 5.1.
The relationship of Schmidt hardness to the material properties of strength
and deformation is discussed in the following paragraphs.
a. Relationship between Schmidt hardness and compressive strength
An arithmetic plot relating Schmidt hardness and uniaxial compressive

strength is given in Figure 5.5. The correlation coefficient for these data
in linear regression is .880. Figure 5.5 indicates that a curvilinear rela-
tionship might provide a better fit of the data. Further, the higher strength
rocks tend to "stack up,)' indicating that above R = 55 the impact hammer is
somewhat insensitive to variation in compressive strength. It is also noted
that some lower-strength rocks, such as tuff (14.1) can have a relatively
high R-value, suggesting that the magnitude of rebound is influenced to a
considerable degree by some variable other than compressive strength.
In attempting to provide a better distribution of the data, the

R-values are multiplied by dry unit weight, Ya, arid plotted in Figure 6.18.
This plot clearly shows better discrimination of strength at all values of
YaR, in which the functional relationship between the X and Y variables
closely approximates a logarithmic curve. The equation of the curve is
obtained by least-square regression analysis, in which the compressive
strengths are transformed to their corresponding iogarithms.
162
w
N 0
Sn* w
N* 0
__w -Z
CL 0
39 LA CI)
00
0 Sn
b, 0
0 000 c
0
b CY
V--------1
0 Fa
x0s (I '46OI AS~WY
o W
163
Figure 6.19 shows a semi-logarithmic plot of the same data given in
Figure 6.18. Here, the correlation coefficient r = . 9 43 , which is consider-
ably higher than r = .880, as given by Figure 5.5. One standard error of
estimate for the experimental data is shown by the two short-dashed lines,
parallel to the regression line. On the basis of the characteristics of the
normal curve, Figure C.1, approximately 68% of the observations fall within
this interval, if we assume that the sampling distributions of the means
(given by the average group values shown hereon) are normal in shape, and
that the population means for all the rocks are defined by the regression line.
In order to set a confidence interval on the estimate of the means

for a fixed set of X-data (i.e., the data shown in Figure 6.19), equation
(C.18), Appendix C, is used for any desired level of significance. The
generalized confidence belt is defined by branches of a hyperbola, as shown
in Figure C.3. However, for predictions of individual events to be made on
the basis of a given set of data, a random element must be included as an
additional source of uncertainty, modifying equation (C.18) to that of
equation (C.19). Further, because of the great variability of rocks and
their properties, it is not considered to be within practical engineering
interests to expect as high a level of significance for rock data as fre-
quently used in other statistical studies (i.e., 95% confidence). Thus, for
prediction purposes, the 75% confidence limits have been selected and are
shown by the outer, curved-dashed lines in Figure 6.19. The 75% confidence
limits are branches of a hyperbola, defined by equation (C.19) modified as
follows:
CL(Y) = y + b(X-R) t t s +y-

+ R
.25 y.x n Z(xX-) 2
where t 25 is for n-2 degrees of freedom. Therefore, it is inferred that

on the average 75% of all rocks to be encountered will have properties which
lie within the confidence belt shown in Figure 6.19.
It is interesting to note that the only two rocks which have strengths
falling outside the 75% confidence belt (10.2 and 11.2) were both loaded in
a direction parallel to the direction of bedding or foliation. Thus, a higher
strength was measured than that which is indicated by the Schmidt hardness
in the relationship defined in Figure 6.19. The other 26 rocks were all
loaded in a direction perpendicular to bedding or fabric, where such was
present.
It is recognized that a requirement of reQression is that the X's,

or independent variables, be measured without error. Actually, there are
errors in X as well as in Y. However, in the present study it is reasonable
to assume that the errors in X (i.e., raR) are insignificant for two reasons:
i) the average coefficient of variation for each oF the proper-

ties in X is quite small (i.e., for 7 a, V = .5; for R,
'1%= 4.7).
ii) the unit weights given herein are in agreement with unit
weights obtained by others, for those rocks which are reported
in the literature (i.e., 2.3, 3.3, 4.1, 5.3, 6.1, 8.1,et cetera);
164
90 _ _ _ _ _- - - J__
80 - Note " ZZ2
70 - I) 7 0 = dry unit weight, pcf -i/ /i
.2) R = Schmidt hardness (L-hammer) ____
/_
sy.,= one standard error of estimate
_
60 3)
for experimental data__
50- _ -
log (o'(ult) : .00014 YoR + 3.16 .2 //2
SYK=.31 cr,(ult) .29/ 2.3/
40 r :943 - /_ O /
/. I 7Z5 2 .-
30-- /
.~//
00
-o '3 S
3 z0- - -y "" /
"0. 3't / t--"
06
- /i /
S//1 /
* Upper limit of / .
*- 75% confidence / /
10
10 _____ for prediction ______
Y /q 7.2
*- 9 7.1" /
(I ,___-_-_, ___
"a, _,____
C 6 10.3/
E
0
___
.i4.,
/ /ilt-
/ .. / / 'Lower limit of
75% confide~nce
,/"/ y.
4 olSy-1, for prediction
S~* 1401 x~
020 40 60 80 100 120

Yo'R x 10 2
FIGURE 6.19 RELATIONSHIP BETWEEN AVERAGE VALUES OF Yo' R

AND ULTIMATE COMPRESSIVE STRENGTH FOR ROCK
IN UNIAXlAL COMPRESSION (SEMI-LOG PLOT)
165
also, the R-values were regularly checked against a standard,
and were found to give identical hardness readings throughout
the test program.
Therefore, all relationships and corresponding charts, presented in this sec-

tion, are based on the assumption that there are no measurement errors in X,
or that they are so small they can be neglected. (Reasoning similar to i and
ii, above, is used in the cases of Sh and V as well.)
By means of the functional relationship presented in Figure 6.19,

uniaxial compressive strengths can be approximated for various combinations
of Schmidt hardbisss and unit weight. By use of this procedure, a rock
strength chart has been developed and is shown in Figure 6.20. By obtaining
the Schmidt hardness, R, and the unit weight, 7 a, an estimate of the strength
can be made. The statistical dispersion, or possible deviation from the cor-
rect value of strength (defined by the 75% confidence limits in Figure 6.19),
is shown by the curved line at the left side of the chart. For example, the
Dworshak gneiss (4.1) has an R-value of 48, and a unit weight of 175 pcf.
By entering the chart at R = 48, and following the short-dashed line up to
Ya - 175, the strength is estimated by then proceeding horizontally left to
the value of 0a(ult) = 21,500 psi. Statistical dispersion from the estimated
value of strength may be as much as + 8,500 psi, which is shown by the hori-
zontal distance of the dispersion line from the left border. The actual
measured strength of the Dworshak gneiss is 23,500 psi, as given in Table 4.2.
Unfortunately, a strict adherence to a purely statistical dispersion

results in an unreasonably wide variation in predicted values for the higher
strength rocks. Such a variation was not actually observed in the experi-
mental data in the present study. Therefore, in the realistic application
of this chart to engineering projec~s, use of the statistical dispersion
limits may not be warranted.
b. Relationship between Schmidt hardness and modulus of deformation
Just as for strength, the correlation between Schmidt hardness and

tangent modulus of deformation at 50% ultimate strength is considerably
improved when R is multiplied by unit weight. This relatic'nship and the
corresponding linear regression line are shown in Figure 6.21 for which the
correlation coefficient r = .847. (Figure 5.14 shows the relation between
Et and R alone, for which r = .731.) A still higher correlation is obtained
by squaring the unit weight and multiplying by R. for the independent (X)
variable. This relationship is shown in Figure 6.22 in which r = .876.
Again., one standard error of estimate for the experimental data is shown by
the short-dashed, parallel lines. The 75% confidence interval for prediction,
as determined by equation (6.1), is given by the heavier-dashed, hyperbolic
lines.
The regression equation in Figure 6.22 is used for computing the

tangent modulus of deformation for various combinations of Schmidt hardness
and unit weight. From these computations, a rock modulus chart has been
developed and is shown in Figure 6.23. Just as for strength, by determining
the Schmidt hardness, R, and the unit weight, 7a, an estimate of the tangent
modulus at 50% ultimate strength can be made. The statistical dispersion of
the data, as defined by the 75% confidence limits from Figure 6.22, is shown
166
60
III T -w-14-
500
00
K 20
I- I
4---
10Il
9.
16
C)
I uj
- ui
CLd
W4
cn- 0
0 .
- PC
Ir , - < C
-i u
WIA
ollotIS) 901 v s 113 '2lloPOnpo luo~uo.1
168
U i
.22
Z *0
- ___- 40 -!
DC
- 00'
1u- CL
C) >
0 0 N
w Q&J
fin 9.) 301 x S 3 'SlfP)d U~~
E6
iW
h. I I" I I; u C!.
i' C", -
A
3 I I -} _,_.: l--",,,
_,;
_ __i,,
. I
I- _ _ _ ___ _ i__
00
w*t 3
....."
.....
5 --
R(LHmmr
oooo ,,oo
0 5
ScmitHodn,
// 00
- - - - I - I-
ipi
1701
FIGURE C.253 RO f:ODULUS CHAIMT BASED O il SCH NItDT M -2 ,1 S3
170
by ttI: curved, so Id I i.'ie t~.tL left s I fC'ft c".rt.
OV Thus, for cx:,le,
by cntering the chlrt at R - i; (trnir, f,- i. s. .,; s) O d foflc.,ing te
short-c'a:hed i inc to Ya T 175, then hoI ::crV: lcft1 Et is estinln:t:d to be
9.1 x 10 psi (+ 2.4 x 106). This is ct:o to the
tr,,:-::surcd mJlus
value of 7.8 x TO psI, as given in Tablc 4:,3.
A strict statistical dispersion, as deterr.incd frcm the confidence

lifnits In the arithrletic plot of Ficure 0.22 incrccses with d(creIsing
modulus below the deta mcan. This is sU_>:ni by the curvinrg, di spersion l i,.
bclow-v Et - 7.5 x 106' psi. If this dS r 'r ic is used for the I;.. .
rocks, the predict;d ,.culi v:lucs ,:A:,I I,o -os t. -' r t'.;;i;A, for c;:C AI-.
at Et l x 106 psi, th ste-tisticssl disr,,rslcn is shc,:n tco be : 5 x I1O psi.
Such a dispersion is not rce,-sonable in CrTIsn,ce.ring use-e, since ncntive
moduli values do not exist. Further, thl data in the present
study are, in reality, closer to the regression line than the statistical
dispersion would jnfer. Therefore, for estimrating tangent moduli values
below Et - 9 x 100 psi, it is suggested that the dispersion be obtained from
the dashed line extending through the origin or, perhzps even more realis-
tically, ignored altogether for actual engineering usage.
6. SHORE HARDNESS AS AMl INEX OF STR.T17,' /D 07.TiD
The Shore scleroscope has proven itself over the years to be a very
useful laboratory instrument for measuring re!ative dynamic-hardness proper-
ties of metals. The scleroscope has, in recent years, been frequently
utilized for laboratory-hardness measurcrments of rock, as described in
Section Three; it has also been used in the present investigation for this
purpose.
The relationship of the Schmidt rebound-hardness values to those

determined by the Shore scleroscope is plotted in Figure 6.24. Two least-
squares regression lines are shown, tog.athe-r with their corresponding
equations. Line R is the regression line of Y on X, for which Shore hard-
ness is the independent variable; and line Sh is the regression line of X
on Y, for which Schmidt hardness is the independent variable. A reasonably
high degree of correlation between the two methods for measuring hardness
is indicated by the correlation coefficient, r = .872. The Shore hardness
measures a very fine-pointed, surface property, whereas the Schmidt hardness
appears to measure a broader, more macroscopic, mass-hardness property of
the specimen.
Because of the obvious relationship between Shore and Schmidt hard-

ness, and because the scleroscope is a standard piece of equipment in many
testing laboratories, the same procedures as presented for the Schmidt harmaer
are utilized in developing rock property charts based upon Shore hardness.
The relationships between Shore hardness and the strength and deformation
characteristics for rock, together with the corresponding charts, are
discussed only briefly in the following paragraphs because of the similarity
to the discussion presented for Schmidt hardness.
a. Relationship between Shore hardness and compressive streML
The relationship between Shore hardness and compressive strength was

previously given in Figure 5.4, in which r - .897 for linear regression. Cy
171
6o ru.72 ..
1o
4.1.
R10.1
400
35
30"
400-6 *00 -1 60-- - 00_
25 -Sh
S,h "2 15R -43 3
Sivy=- 13.1
r a .. 72
v
Shore Hordnes$, Sh (Sids)

FI GU RE .24 F(--L AT IC,", ,.! C -T N.,AV ER/G.
A .LUZ. S OF SHORZ
T) S
H^A Ri,. A 4 SCHM3IDT HARCN'ES,' FOR ROCK
172
mul t) p yinj thN: SI ra h,";- dr s b, th c ur,Lt t,: o nh o
curvif I i[ r plot in Ficre rc 6 is o Irt ....
,
A serni-lorarit!il,:,c plot of the dat3s~c in

nFigure 6.25 is given
in Figure 6.26, for which r - .921. The recression cquatiori in both ficures
is obtained from the linear regression line given in Figure 6.2-5. CnC
standard error of estimate, Sy., is shown to be .37%a(ult) in this plot, as
comparcd to .31la(ult) based on Schmidt hardh,:s. The 75%, confid tce belt
for pred. ction is, thus, somewhat wi cr tla,:n thnt for Sc.hr, idt h:rd s;
th'r.ercforc, the statistical dispersion at a;y cstiratcd strength level is
corrcepondinrjly greater.
On the basis of the functional rcletionship in Ficgvre 6.2', a rock

strength chart has been dcveloped and is so.o'.:n in Figure 6.27. This chart
is identical in formi with the chart based on Schmidt hardness in Figure
6.20. Thus, again using the Dworshak gneiss (4.1) as an example), the chart
is entered at Sh = 69. Following the short-dashed line up to 7 a = 175,
then proceeding horizontally to the left, the strength is estimated to be
26,000 psi (L 13,500). This is comparable to the measured strength of
23,500 psi, and to the strength predic'ed by the Schmidt hardness equal to
21,500 psi. Note that the statistical dispersion approaches 50% of the
estimated value, whereas the estimated value is actually only 10.5% greater
than the measured strength. Further, the strength predicted by the Schmnidt
hardness is only about 8.5% lower than the measured strength.
The rock strength chart based on Shore hardness, as shown in Figure

6.27 appears to be limited to rocks with strengths greater than 5,000 psi.
This is due to the emphasis given to regression analyses in the development
of the chart, and the consequent regression line defined in Figure 6.26 for
all 28 rock groups. The two., lo,,'est strength rocks, rock salt (9.1) and tuff
(14.1), are probably not correctly represented by the suggested functional
relationship. Indeed, the correct relationship may, more properly, curve
downward tovward the origin. Additional investigation and separate analysis
of the lower-strength rocks would likely result in better definition for
rocks having low Shore-hardness values (i.e., less than 30).
b. Relationshi betwteen Shore hardness and modulus of deformation
As in the case of Schmidt hardness, the correlation between Shore

hardness and tangent modulus of deformation at 50% ultimate strength is
conliderably improved when Sh is multiplied by 7 ap and even more improved
when 7a 2 is used instead of 7a. The re~ationship between Sh and Et has been
given in Figure 5.13, in which r = .746. Figure 6.28 shows the relation
between 7aSh and Et, for which r = .800. The best correlation is shown in
Figure 6.29 between a 2Sh and Et, for which r = .831.
Figure 6.29 shows the limits for one standard

the 75% confidence belt for prediction, as determined error of estimate, and
by equation (6.1).
When compared with the similar relationship for Schmidt hardness, Figure
6.22, the confidence interval in Figure 6.29 is seen to be somewhat broader.
A rock modulus chart based on Shore hardness and unit weight is

sho.4n i,-. Figure 6.30. This chart is identical in form and use as the chart
based (n Schmidt hardness and unit weight of Figure 6.23. The regression
173
CLl
9*4 U)
9' ~0 W
0~
- 9.
*( 0
S.. (
0 j
9.r
w Z
IM
4
cz
CL; U)u)
wa 4 U
9.C 00
- -. ~ -- V
b 0 4 w
-.- + 9
w I--
L41 x sdS41~
vm '15MI DA~soiZoo C
1704
90- -
so - Note: -/
- I) 7,= dry unit weight, pcf ____
TO - 2) Sh = Shore hardness (sides) estimat
60 3) sy.= one standard error of estimate-

for experimental dato I I /
log a 0 (ult) :.00066TYS,+3.62 .7F 4
Sy.I =.37coo(ult) l.2."'
r: .921 @1./ / *z3
40LL
6./ 2.1 i
1.20 5.2 i
30 /, so /
30 z -/s __ / ___
X
"Q 2 Uoper limit of /
41'
/
=- ,ofor prediction
,oc.o
.75% confidence -7
13.1
l
,, -
/y
-3 1/
'132 /,
0(
3..1
9%
o, ?.5Z 07.2-
in o / .v .~ L. ___,-
E -7-
/ 75 '-Y
.,-e
14. /1
91____ \-Lower limit of__
for prediction
0 40 so 120 160 200

2
a X 10
YO-Sh AND ULTIMATE COMPRESSIVE STRENGTH
FOR ROCK IN UNIAXIAL COMPRESSION
(SFMI -LOG PLOT)
175
.1 .j
20
4-
Not
Diprso limit _ defined
I I Note
o io 20 30 40 50 60 70 80 90 100 110O1;
Shore HordncS3, Sh
FIGIL'"72 6.27 ROCK STRE"1GTH CHART EASED ON SHORE- HADE-.
176
0j0
0 w~
-~~~~ -- - -
*0
0
00
0 0w
-10
a;~ zz
06 0
NS
z
0 C
- o- -.
~ 10 Ck W
ow II
-1140 1 iippy ueu Lc 401x4d
177
S0LL .
00
l* 0
w Z
000
CL 0
e~.a
w W
_A IL - -
0
E - ~ E
1~
0
0 6 N 0
w
(9tDj )l s 3'n~pf U~~ V)
C17
/1
aio
CL'
IC -t - - - z -
00
4-
36 ~
ol I _ I as shw (Se tet)I
0o 6 7 o 90_0__3'___010 11 2
17
equation, given in Figure 6.29, provides the mathematical basis oi the chart.
In estimating the tangent modulus for Dworshak gneiss (4.1), the chart is
entered at Sh = 69. By following the short-dashed line up to 7 a =6175, then
left horizontally, Et equals approximately 9.4 x 06 psi (Q 3 x 10 ). This
is comparable to the measured modulus of 7.8 g I0 psi and to he modulu,
predicted by the Schmidt hardness of 9.1 x 10 psi (+ 2.4 x 10A).
Just as for the chart based on Schmidt hardness, the statistical

dispersion is shown by the curved line on the left of the chart oor Shore
hardness; Figure 6.30. Again. for this chart, below Et = 9 x I0 psi, it is
suggested that the dispersion be estimated from the dashed line extending
down toward the origin, rather than from the stetistical, curving line.
The rock modulus chart, as shown in Figure 6.30 implies that the
tangent moJulus for rocks having low modulus values may not be evaluated by
the Shore hardness. This is apparent for tuff (14.1) and rock salt (9.1).
Again, this is due to a strict adherence to the statistical, regression line
of Figure 6.29, which apparently does not properly bapresent the lower-
modulus rocks. Such rocks may need to be separately analyzed in order to
obtain a more representative correlation among the engineering properties
of th. _-rocks and the Shore scleroscope.
7. SONIC VELOCITY AS AN INDEX OF DEFORMATION
A comparison of the static and dynamic properties for rock has been
presented in Section Five. Figures 5.22 through 5.27 show various relation-
ships between dynamic properties based on some function of sonic velocity,
and static deformation propertie. of rock. In all cases, the correlation is
quite good and the relationships are fairly well defined. Thus, it appears
to be logical to utilize the sonic pulse velocity as an index for static
modulus of deformation.
The sonic velocity has excellent reproducibility as indicated by the

small coefficient of variation in Table 5.1 (V% = 1.9). Because of the
necessary preparation of the specimen, and the specialized nature and cost
of the laboratory equipment, the sonic velocity at the present time does not
fully meet the criteria for an index property, as given previously. However,
because numerous rock laboratories are equipped to measure sonic pulse
velocity, and because of its relationship with .;tatic modulus, a rock
modulus chart based upon sonic velocity is included herein.
The relationship between sonic pulse velocity and tangent modulus at

50% ultimate strength was given in Figure 5.18, in which r = .860 for the
linear regression line shown therein. The data, in accord with theory,
indicate that a curvilinear relationship would provide a better fit. Again,
a linear relationship was shown in Figure 5.23 between th~e constrained modu-
lus (a function of 7aVp 2 ) and Et, for %ich the correlation coefficient,
r = .901. Finally, in Figure 6.31 a plot of 7aYp in relation with Et is
shown, in which a curvilinear relationship prcavdes the best fit of the data.
The data shown in Figure 6.31 are given in the Semi-logarithmic plot
of Figure 6.32, in which r = .929 for the linear regression line shown. One
standard error of estimate is equal to .31 Et7 and is shown by the lighter,
180
0
0 V
cr~~
wZ
cz:
.0 LjW
N-
>9
L:A
j
0 _ >0
_
CL_
to a) o I- r
D40
0 10
0L
b co __
U_
-
C~ o
.00 Z aJ
-1 U
w
N
'fIsnnolPd 4u9buo.j
'l !d '1
181
30
log E, .389 x 10-6y, Vt + 5.72
.31 E/
C-_r 929__
=.
20
9B ?
11,2 4 ,112
o ,o ... . :/.i/ ! :
-. I__2
*0
o Upper limit of 7r- 0
j7 ~
5 -75% conf idence
for prediction / 1 - - -
* / I~o
S-io. 7/ Lower limit of
_or /. 75 % confidence
""2
5 - 3 - /.. for prediction
01 2_ zI _ I __
/ Note:
/1) Et obtained1 from straight-line portion
o / o'- curves, i.e., at stress level
1.0 - - o-o= 50 %odult), except for 9.1
0.9 4 q0=
-"0 Wit%:(ult)"
0.80'iS i-f i 2) y =dry unit weight, pcf.
"0.7 3) Vp from IWt loading cycle for
0.6 -= 5000 psi.
0.75
/ ' It ./0estimate
4) Sy.: one standard error of
for experimental data.
0.4
1 1 1 1...
0"30
o00 200 I300 400 500
FIGURE 6.32 RELATIONSHIP BETWEEN AVERAGE VALUES

OF y.V AND TANGENT MODULUS OF
DEFORMATION AT STRESS LEVEL OF ONE-HALF
ULTIMATE STRENGTH FOR ROCK IN UNIAXIAL
COMPRESSION (SEMI-LOG PLOT)
182
short-dashed lines, parallel to the line of regression. The 75% confidence
belt for prediction, as determined by cquat ion (6.1), is also showin, and is
noted to be similar in form to the semi-logarithmic plots for strength
given in Figures 6.19 and 6.26.
A rock modulus chart based on dilatational wave ( pulse)

p:z
velocity and unit weight is shown in Figure 6.33. This chart is used in
the same manner as those for Schmidt and Shore hardness presented in
Figures 6.23 and A.30, respectively. It is based on the functional
relation given in Fioure 6.32, and requires only the dry unit weight and
the sonic pulse velocity of the rock.
In estimating the tangent modulus at 50% ultimate strength for

Dworshak gneiss (4.1), the chart is entered at V = 150350 fps (i.e., the
velocity measured at a stress level of 5,000 psiY. By following the short-
dashed l~,ne up to r0 = 75, then horizontally left, Et is estimated to be
5.8 x 106 psi ( 2 x 100). This is comparable to the measured Et of
7.8 x 106 psi at 50% ultimat9 strength. Et predicted by Schmidt hardness
is 9.1 x 106 psi (+ 2.46x 10 ), and Et predicted by Shore hardness is
9.4 x 106 psi (t 3 x i0 ). It is interesting to note that the tangent
modulus for Dworshak gneiss, when measured at the same stress l1vel as the
velocity (i.e., 5,000 psi instead of 50% ultimate), is 5.2 x I0 psi, as
given in Table 4.3.
Although the above discussion pertains to intact rock specimens tested

in the laboratory, it is pertinent to observe that the seismic (dilatational
wave) ve:ocity measured in the field is one of the most economical procedures
used in subsurface exploration. The relationships and charts, presented
herein, for intact rock cannot presently be utilized for field evaluation
in connection with seismic exploration. However, at such time as the
influence of geological discontinuities on rock-mass behavior can be
properly evaluated, the seismic velocity (which is related to the sonic
pulse velocity) will then likely prove to be an important, economical
"-' f,_'.-deformation _ractaristict8.
183
20
12.
K" j+ 4 ii/ i- /7 " 1-

S,/, t I, -, _0 ! i / ,_
'";
6__ - - -
0 - 2< '
O / <..! 7
SIC
04 . . ... "-
04
-
.. . . . .. .. . .. . . . ._.. . . . . . . . _
07 l..
05
06 ,Note
06 i Dispersion limits
O. -J ....... defined f r 7
, confidence
2 4 S0-a
6 10J 12 14 16 le 20 22 24
Sonic Pul.se Velocity, VP, fps x I0 3
184-
FIGURE 6.33 ROCK MODULUS CHART BASED ON DILATATIONAL.
WAVE VELOCITY
SECTION 7
SURMPAY AND COIUCLUSIO'IS
An experimental investigation has been conducted with the objective

of developing an engineering classification system for intact rock and a
related set of index properties. This study has not dealt with the charac-
teristics of the in-situ rock mass with its inherent discontinuities. Rocks
from 27 localities, representing the following 13 geological rock types
were studied: basalt, diabase, dolomite, gneiss, granite, limestone,
marble, quartzite, rock salt, sandstone, schist, siltstone, and tuff.
Laboratory tests were conducted on NX-size cylindrical core specimens of
these rocks in order to determiue the following ten physical and elastic
properties: unit weight, uniaxial compressive strength, Shore scleroscope
hardness, Schmidt hammer hardness, abrasion hardness, absorption, point-
load tensile strength, Young's modulus, Poisson's ratio, and sonic pulse
(dilatational wave) velocity. A total of 257 rock specimens, all having
L/D ratios of 2:1 were tested in this investigation.
All tests were conducted on specimens which were initially oven-dried,

then allowed to remain in the ambient air of the laboratory for at least two
weeks prior to testing. The following nondestructive tests were performed
on all specimens: dry unit weight, Shore hardness, and sonic pulse velocity
under a uniaxial seating load of 100-150 psi. Approximately one-half of the
specimens from each rock group were subjected to sonic velocity and stress-
strain measurements while undergoing two complete static-loading cycles up
to 5,000 psi uniaxial compression. These specimens were then loaded to
failure on the third cycle, during which stress and strain measurements only
were obtained. The other one-half of the specimens were subjected to
Schmidt-hardness tests, and then were tested for determination of abrasion
hardness, absorption, and point-load tensile strength.
The test results for each group of specimens were subjected to

statistical analyses in order to obtain average values and coefficients of
variation, and to determine the number of replications required for every
property of each rock group included in the investigation. The average data
for each rock-group property were subjected to statistical correlation and
regression analyses from which the least-squares regression lines, simple
correlation coefficients, standard errors of estimate, and functional rela-
tionships were determtned for selected pairs of variables. These relation-
ships were presented graphically for visual comparison with the corresponding
test data. All data reduction and statistical analyses were performed with
the aid of the IBM 7094 electronic computer.
Static and dynamic properties for the rocks in this investigation

were compared on the basis of the theoretical equations for the constrained
modulus of deformation. The constrained modull values were computed
dynamically as a function of the sonic pulse velocity, and statically as a
function of Young's modulus and Poisson's ratio from static stress-strain
measurements.
185
A system of engineertr.g cle.s1ficmt~on has been proposed In which
Intact rock rnay be classified in terms of its strcngth and deformation
characteristlcs. Vith this system rocks can be classified by either deter-
mining thz strength and modulus of elasticity in actual laboratory tests or
by using the approximate values from the index property charts. Five charts
have been presented from which the strength and modulus values may be
estimated. ThE charts have as their bases the numrarical indices obtained
by either the Schmidt hammar, the Shore scleroscope, or the sonic pulse
velocity, each used In conjunction with a fourth index property, the unit
weight of the rock. The charts also provide an indication of dispersion
of the strength and modulus values, based on the 75% confidence interval
determincd by statistical analysis.
2. CONCLUSION, S
The more Important conclusions and pertinent observations drawn

as a result of this research study are summarized in the following paragraphs:
1) An engineering classification for intact rock has been developed

which is based on the compressive strength and modulus of
elasticity -- two important physical properties of rock that
enter into many engineering problems. The classification is
considered a workable and useful system. It is sensitive to
the mineralogy. textural fabric, and direction of anisotropy
of the rock so that specific rock types fall within certain
areas on the classification plot. (See summrary plots for
igneous rocks, sedimentary rocks, and metamorphic rocks in
Figures 6.4, 6.8 and 6.13, respectively.) In the proposed
system a rock is classified on the basis of its unconfined
compressive strengths into one of 5 categories: Class A -
Very High Strength (greater than 32,000 psi); Class B - High
Strength (16,000-32,000 psi); Class C - Medium Strength
(8,000-16,000 psi); Class D - Low Strength (4,000-8,000 psi);
and Class E - Very Lf-i strength (Less than 4,OQU psi). ihe
rock Is further classified on the basit of its modulus ratio
(defined as the tangent modulus at 50% ultimate strength divided
by the uniaxiai compressive strength) into one of three cate-
gories: Class H - High Modulus Ratio (greater than 500);
Class L - Low Modulus Ratio (less than 200); and Average
Modulus Ratio (between 200 and 500) which carries no letter
designation. Thus, a rock may be classified as BH, BLP B,
et cetera.
ii) The results of this investigation verify that meaningful Index

properties may be determined for rocks. Usefulness of these
properties to engineering classification is indicated by (a)
their relationship to material properties which are used in
design (modulus of elasticity and unconfined compressive
strength), (b) their simplicity, low cost, and rapidity of
determination. and (c) their reproducibility. The Schmidt
hammer hardness together with the unit weight show the greatest
promise for use as Indices of the strength and modulus proper-
ties for intact rock. Again., when used together with the unit
186
weight, the Shore hardness As an index for the strength and
modulus properties, but It has a lower degree of correlation
with either of these properties and is more limited to laboratory
usage than the Schmidt hardness. The sonic velocity, also in
conjunction with unit weight, is an index of static modulus.
However. the sonic velocity apparently is not as good on index
property as Schmidt and Shore hardness because It has higher
statistical dispersion values. Further, It does not fully meet
the requirements for an index property because of the more costly,
specialized nature of the laboratory equipm,2nt needed for its
determination. It is more restricted in that it cannot be used
directly as an index of compressive strength.
Several, interesting subsidiary observations have been noted during

this investigation. Some of these observations have also been made previously
by other investigators. These observations are briefly sunmarized as follows:
i) In general, crystalline rocks have nonlinear stregs-strain curves

that are S-shaped. The curves are concave upward at low stresses,
approximately linear at 50% ultimate strength, and concave down-
ward as failure is approached. This behavior Is more pronounced
for the lower-strength, lower-modulus rocks, and is not so
apparent for the higher-strength, higher-,nodulus rocks for which
the stress-strain relation is nearly linear. The nonlinear
elastic behavior of rocks under uniaxial cori:pression is attrib-
utable to minute cracks which are open at low stresses and
become closed as the stress is increased.
if) Most rocks under unlaxial compression exhibit an elastic hystere-

sis in their stress-strain curves during unloading. Because of
friction, it has been suggested that cracks which have undergone
sliding during compression do not immrediately slide in the
opposite sense when the load 4- redt",d; thus, a clockwise loop
is formed I, the btress-strain curve as the stress Is raised and
then reduced to zero.
iII) Sonic pulse velocity for rocks increases with increasing stress.
This effect was noted to be most pronounced on those rocks which
exhibited well-defined, S-shaped stress-strain curves. Both the
increase in sonic velocity and the shape of the stress-strain
curves are apparently due to closing of minute cracks under
increasing stress. The greatest effect of stress Increase was
observed in schist (01.0). Wthen the axial stress was increased
from 115 psi to 5,000 psi) VP increased by 132%. The smallest
effect, for the same stress increase, was exhibited by Solenhofen
limestone (6.3). for which Vp increased by less than 1/4%. WIhen
there is no pressure on the rock, open cracks (which are most
predominant in schist) impede the prog-ess of the sonic wave
which must then travel around the cracks. The percentage Increase
in Vp, for the majority of the rocks in the present investigation,
was found to be less than 10% for a stress increase from 100 psi
to 5,000 psi.
187
iv) In order to obtain meaningful sonic velocity data from rocks,
a relatively large stress must be applied in the direction of
the wave prepagation so that a terminal velocity is approached.
For most rocks in this invwstigation the greatest increase in
velocity took place within the first 1,000 psi axial stress. In
nearly all cases, a terminal velocity appeared to be approached
at 5,000 psi. However, no measurements were made above 5,000 psi.
v) In general, rocks which have initial modulus values E, of less

than 9 x 106 psi and sonic velocities V of less than 16,O00 fps
have stress-strain curves which are initially concave upward.
Also, such rocks exhibit relatively large increases in both modu-
lus and sonic velocity when the stress is raised from 100 psi to
5,000 psi. In such cases the initial modulus is lower than the
tangent modulus at 501 ultimate strength. On the other hand,
rocks for which Ei is greater than 9 x 106 psi and Vp is greater
than 16,000 fps show very little increase in velocity and, in
general, show a decrease in modulus as the stress is intreased
to 5,000 psi. The latter have stress-strain curves which are
initially linear. or slightly convex upward and, therefore, may
have somewhat higher initial modulus values than tI,'. %angent
modulus at 50% ultimate strength.
vi) The values of dynamic and static properties of rocks are equal
only for compact rocks at low stresses. At high stress levels
(i.e., 5,000 psi) the values of dynamic properties are 10-20%
greater than static properties for compact rocks. For less
compact rocks, the magnitude of the dynamic properties may be
100% greater than the static properties at low stress levels,
but the difference between them decreases with increasing
stress. However, it has been suggested that these lower-modulus
rocks apparently are not sufficiently homogeneous and isotropic
to meet the cenditions for which the formulas from elastic theory
were derived.
vil) The propagation velocity of dilatational waves in rock tends to

approach a lower limiting value regardless of the stress level
to which the rock is subjected (within the range of stress
Investigated herein). This observation is probably a function
of the structure of rock, the degree of cementation, grain
contact, et cetera. The lower limiting velocity observed
herein is approximately 6,600 fps for the average density of
162 pcf for all rocks in this investigation.
viii) The influence of anisotropy on sonic velocity decreases with

increasing st.,ess level for the Luther Falls schist (1!.l and
11.2). The ratio of VP in the direction of foliation to V^ per-
pendicular to foliation is 3.1:1 at a stress level of 110 psi,
but only 1.4:1 at a stress level of 5,000 psi. Since this
effect was not studied for any of the other rocks in this
investigation, no general statement regarding the effects of
anisotropy can be made.
188
ix) In general, oven-drying rock at 100 0 C., causes a decrease in
sonic velocity of less than 2%. Two exceptions which were noted
are dolomite (3.2), for which the magnitude of the velocity
decreased by 8.5%, and marble (7.2), for which the velocity
decreased by 21,6%. The iatter is probably related to the very
large grain size, and corresponding nonuniform expansion and
contraction.
x) M!gative values of Poisson's ratio were consistently observed

for two marbles (7.1 and 7.2) and schist l1.1) at low-stress
levels (i.e., approximately 1,000 psi). The physical concept
of this is not completely understood although it appears to be
related to the closing of microcracks in the axial direction
without an attendant lateral expansion of the sample. Only
those samples with pronounced S-shaped stress-strain curves
exhibited this phenomenon.
3. SUGGESTIONS FOR FURTHER RESEARCH
Several areas of additional research are suggested. Some of these

are the natural outgrowth and continuation of the present investigation;
others are the result of miscellaneous observations made during the course
of the investigation. These are commented on in the following paragraphs:
i) Additional tests should be made on almost all rock types to

obtain more data points (tangent modulus and compressive
strength) For the rock classification charts of Figures 6.4
to 6.17. With more points it will be possible to draw more
meaningful envelopes for each rock type (i.e., the bounding
envelope which includes 75% of the poin~s).
ii) Additional testing of numerous samples of all rock types should

be carried out in which the index properties (Schmidt hammer
hardness, Shore hardness, sonic pulse velocity, and dry unit
weight) are determined as well as the unconfined compressive
strengths and tangent moduli at 50% ultimate strength. The
new data points when plotted on the corresponding index property
charts of Figures 6.19, 6.22, 6.26, 6.29, and 6.32 would aid
in verifying or modifying the regression curves and may increase
the level of significance.
iii) As more data points for a particular rock type become ava6lable
it would appear worthwhile to construct individual index property
charts for that rock type. The regression curves and signifi-
cance level should prove to be enhanced when working with a plot
of only one rock type versus a plot which contains all rock types
as is currently being presented. It is recognized that a con-
siderable amount of testing will have to be done before-enough
data become available for constructing these individual plots.
iv) Detailed investigation of the index property relationships of

the Class D and Class E rocks is desirable as the data for these
189
lower strength rocks are sparse. The shales in particular warrant
more study.
v) Tho data obtained in the present study did not permit final con-
clusions to be drawn with respect to the textural classification
suggested by Deere (0963b). As additional test data become
available, his proposed textural grouping, i.e., interlocking,
cementcd, laminated-foliated, may allow for more consistent
Interrelationships among the physical rock properties than when
the rocks are undifferentiated.
vi) Because of the qualitative use presently made of the abrasive

properties of rock in riassifying exploration drill core (i.e.,
its response to scratching with a knife), and because of the
quantitative relationships between abrasive and strength proper-
ties as shown in the present study, special investigations
directed toward the development of a portable, abrasive-hardness
indicator should prove to be worthwhile.
vii) The development of a re-usable crystal transducer which will

provide for measurement of shear-wave velocities under variable
stress conditions continues to be a requirement for better
understanding of the relationship between static and dynamic
properties of rocks.
viii) The effects of moisture and of anisotropy on the behavior of the

intact rock materials should be studied further. Both unconfined
and triaxial compression tests should be conducted at wvrious
rates of loading and different conditions of drainage.
ix) As frequently reiterated, a universal engineering classification

system for rock in-situ must include an indication of the
influence of the geological discontinuities. Thus, in order to
extend the system as developed herein to thL. in-situ rock mass,
a method of classifying the rock in accordance with the number
and type of geological discontinuities must be developed. Such
a continuing study would be a natural outgrowth of the present
investigation. Its successful completion would allow realistic
estimates to be made of the expected engineering behavior of an
in-situ rock mass solely on the basis of the rock quality classi-
fication.
190
APPENDIX A
PETROGRAPHIC DESCRIPTIONISS
AND
THIN-SECTION PllOTO;ICRCG~lJ2llHS
'9'
This rock consists of 48% plagioclase (An 4 ), 165% olivine,
18% augite, 11% glass., 6% rm1cgnotite., and 1% apatite and sphene.
Corrodcd olivine granules (0.003 to 0.02 rmi) rimm-red with dusty rnagne-
tite particles are contained within an intergrarwlar maitrix of plagio-
clase laths, (up to 0.07 rm-n.In length) Intersertal glass, and augite.
Some zoned plagioclase crystals have grown to lengths of 0.5 (Min. thus
Imparting a porphyritic texture to the rock. Euhedral to corroded.
subhedral crystals of magnetite are scattered throughout the section.
The glass., brown with upper nicols out., is in places partially devi tri-
fled to radial growths of acicular crystals.
Ae"
25X -Nicols crossed 1mm.l)
Black crystals of magnetite and laths of plagioclase

(white) with intergranular olivine and augite.
192
1.2 Basalt IUttle Goose}o Walla Vnlla._Vashin iton
This basalt shows a hyalophitic texture In which ollvina (26%),

augite (5%), laths of plagioclase (410%), and euhodral to subhiedral
grains of titeniferous magnetite (4%) lie in a matrix of dark groon',
probably iron-rich, glass (25%). The 8Vt s crystal size for augite
and plagioclase is 0.2 mn*, although several grains are as much as
2 mm. long imparting a porphyritic texture to the whole. The rock Is
essentially free of alteration products.
4*1'k "I'll IX,r r- r-7 4
4~ "e
p 'IN
25
Kwne
ioscrse
plgols
ah
nagt
rsasI marxo las(lc)
m
- 193
1.3 0rjv~,G~
This rock contains subhedral phcnocrysts of norma~l zoned

lonne) In a grc'une,ioss of olivina (33"',
plagIochase (Up to 0.0 m,.n.
glass (0,titaniferous marn-tita (77), and laths of plagioclase
AnL6 (total pTagloclase Is li%)*L The groundmass minerals are fromi
0.2 nun. to 0.3 r;-,i
long. Olivina Is Intersertal to the laths of
randomly oriented plagioclase. All minerals are free of alteration
products.
IJ
I" j ~ ~
?~
A
a4-\\ >a14
25X Nicols crossed I u.
Corroded phenocrysts of plagioclase lie in a fine

groundmass of plagioclase, olivine., and glass (black).
194
2.1 kfbje 1 York
This rock containf, .12% plagioclase (0n5 1),,30% diopside, and

6% accessory rnnarnetitle and apatite (1%). Secondary minerals are
epidote - serlcite (3%) and chlorite ().The plagioclase laths ha~ve
been altered to sericite which, In places, covers large patches of thre
section. Seconda~ry chlorite borders ma~ny of the dlopsiez grains or
replaces them along cleavage directions. Small patches of micro-
pegmatite are intersertal to the plagloclase laths.
-. 2i
25X -Nicols crossed 1 ni.
Randomly oriented laths of twinned plagioclase (heavily

sericitized - left central portion) Intersertal to diopside
anhedra. Black patches are magnetite,
195
2.2 rn1n
A very fresh rock containing 4;r/, plagloclase (An5 ) 51%

diopside, 1%i'accessory biotite anid less than I% apatite and magne-
tite. .Diopsido, crystals (0.5 to 4,0 rij. are either Interscrtal to
plag1oclase laths or cphitically enclose the~m, Exsolution of ortho-
pyroxene on the (001) planos can be seen on a few diopsida, grains.
Widely scattered irregular blobs of magnetite and biotite are closely
associated with diopside. Plagioclase exhibits multiple twinning and
normal zor~ing.
'4 1
A\
V
25X
icos crsse mm
Twine crstl ofpaicae t ipi 'esra
196
2.3 Ivni
The main constituents of this rock, are plagloclase (An59 ,,49%)

and diopsidic-augite k- ) The only mnino~r constituent Is biotite (2%).
In places altered to chlorite. Accessory manecntite (1%) Is scattered
thro~ughout the section in grain sizes ranging from 0.2 to 0.8 rmn% and
is closely associated with theo magnctitc. The pyroxene grains (0.5 mm.
to 7 mm. long) ophitically enclose plagloclase laths (up to 4 mm. long).
The plagioclase laths have a random orientation.
* - rS'~ *~4 j
42
'6e
25X
Niclscrosedmm
Twine plgicls lah yenlsdb/itc

pyroene(botomrigh). rreula pathesof agntit
(blak)
Itersitil
ae topyroenecrysals
~I. ~1(7
A rnozatc of-dolonite anhedra and rhombs ranging in size from
0.05 to 0 JS rrnm, Tho dolemilte Is clouded with opaque dust which, In
som.3 gra~ins, has tmi~rated to tlec bounedirics.
L?
iJ
25X NioscosdIM
Mozaic
inegot fdlmt ris
Smal
caliteveinat
owerrigt coner
19
3.2 Dofcwi to(Loco rl t)ia l jaFaI Is~j ~York
An equigranular,, non-porous rock in which r~'o7-is and anhedral

grains of dolomiite (0.06 mvi,) form an Interlocking notwork. Fine
opaque dust bordars soman grains.
25X -Nicols crossedM.
Interlocking rnozaic of dolomite grains low porosity.

--
(Large black areas are void spaces caused by grinding the

thin-section)
199
3.3 Dcrie!r r a n n T-r re p Mx
_P~r
cn url
An Interlocking inozaic of dolomite fragments and rhombs

ranging In size from~ 0.05 to 0.14 mm.n A few rounded grains of quartz
end microcline of equivalent size are scattered throughout. The
section shows a continuous fracture filled with calcite.
[
25 ioscrse m
Dooiewt arwvi fcliectigars

frmtpcne owrsbto et
20
4.1 Gneiss (DworshL-) Orofino d,Idho
This rock Is essentially composed of quartz, plagioclase, horn-

blende, and biotite., with minor amounts of accessory sphene., magnetite,
apatite, and secondary white mica and epidote. The sphene, h3rnb!encee
and biotite are concentrated In discontinuous elongate patches. A few
Anhedral
of the hornblende crystals have attained a length of 3 rrorn.
In size) form an interlocking
quartz and plagioclase grains (1.5 min~i.
network between the dark minerals. Wispy patches of seconidary white
mica weave and flow through the quartzo-feldspathic layers and parallel
the foliation.
X77
25X -Nicols crossed Inn
Aithedral grains of quartz (white) and twinned plagioclase

sLtrrounding biotite flakes with their cleavage planes
In the plane of foliation (top to bottom).
201
5.1 SrCrt P oro.2_-rn,,r.C
A coarso-grained alotricarphic-gr~r ular grnIte composed of

albite microclino, quartz, cnd bWotitc. Eyfiakitlc intorgrovwths of quartz
and plogioclase are cc:=cn. Aatito, sphaon (altering to leucoxone),
magnetite, and zircon arc accessory. Kast of the feldspars show patches
of secondary white mica. Sci ;:ointergranular areas are filled with fine,
angular quartzo-foldspathic m'.aterial. These areas of possible recrystal-
lization occur roughly along two different fracture planes which are
characterized by discontinuo;us veinlets of quartz and small displacer,ents
in the minerals cut by the fractures.
4 . Z ',/ 0
q411
25X - Nicols crossed I m.
Sericitized microcline (upper right), fractured quartz

(bottom center), and biotite (center). Note the two
intersecting veinlets of quartz.
202
5.2 Grani s Pok. oore Smrinn Cr Inr
This section shows laroo areas of roccrystallizzction and Is cut

by patches and voinlets of calcite. Plaicc a! f-l dspr (An9, 11%),
is much more altered to sericite than microcline (). cran
Pa~rthito
(Ir/" are essentially frce of alteration, Qua~rtz (yjIs prc~cn~t ci
large grains (up to 3 "n.), or In patchies vt-.+cre It hos boo.n recrystal-
!zed Into grains of 0.5 mmi. Dark brown biot ite is Intcrscrtal to the
irtz and feldspar. Somec flakes have bacn bent during stress and others
dre altered In part to chlorite along clcnavecs. Large crystals of
sphene (2.7 ffn. long), titaniferous rmngnetite 1) and apatite are
accessory,
~ &
I'N
4, S
4'\
~~ V.
25X -Nicols crossed ni
Twinned sericitized plagioclase crystals (left side),,

quartz,, and sericitized rmicrocline (bottom right).
203
TheJsectron shov,~1s the typical hpdoopi rnlrtxueo
VV
orholaean mcrcln (6''0cobied. othti itegrvihso
microcline (casnathd of hentralk

twinnlsemaeing51 of40)h
pusor vItio
dvlphot), bleatate (center, top) quIreuart (extrees oeft),an orthoclase.
witcesesondry (raghet)ine
whaite zirconan mak
hypdessorhcganularo grnthe.
tot204
6.1 tLor
Limecstone_(Cedford)_df d Indionn
A fossiliferous limastone in which foram tests (up to 0.66 rame

in diam.), brachiopod shells, fragments of bryozoans, ostracods., and
oolites of carbonate material (I m-m. in diam.) lie In a ccm-,nt of
optically continuous calcite.
J,
P
A.*
-~%
25X -Upper Nicol out 1mm.
Fossiliferous limestone. Clear patches represent areas

where calcite cement was worn away during grinding. Small dark
circles are bubbles in the mounting media.
205
ki oolitic limastone with oolites, up to 0.6 rrm. in dicmeter, of
microcrystalIIna calcite with smil su nvrutlar cdctritat qut,,rtz grain,
or s h11l! frvn.,nts at tha ccnt ers. Othor saLl frn2rts cd fossil-
Iferous motter, varyIng In size frc-m 0.14 to 1.2 cmn, are firmly ccm-;ntdd
by a mrtrix of fina-Sraincd calcite having a sc~i,:-!wzt v arile graia size.
25X - Upper Nicol out I Mm.
Oolites (center) and shell fragments in

a fine calcite cement.
206
6.3 Soltcffrv.n Cvar13
a
An extrcmaly fine-grained rock consisting of minute carbonate

crystals which ir,-y.art a turbid appea~rance to the section. The only
discernable variations are lighter, r~ire coarse-grained., slightly
elongate patches of calcite (up to 0.4 nz% long). Faint b--dding tracos
are evidz~nt (sec photo).
25 Upe ioIu m
Extrmel
fie-grine
cacitegrondmss
Iwhch
ie
Extrgemeyfn-rre
calcite graisadoauemnerals. Fintwhichdieg
planes (top to bottom) L.-e evident.
207
7.1 f ~ ,Tct , iprtL Pn
An extremcly pure calcite rmarble having a tightly Inter-

lockinorizaic texture. Grain size varies from 0.02 tym to 0.1 rai1.
k/ 4
2X Nicols crossed I rmm.
Clieanhedra of variable grain size in a tightly

Intelocingmozaic. Cleavage and twinning (bottom
-- center) eieto soegan.Dark circles are
air bubbles trapped in mounting media.
208
(7-0
A coarse-grained calcite ma~rble with Interlocking calcita

grains (2 rn-n.), svome containing round quartz crystals.
Owl~
4 4
/25 Nioscose1m
Inelokn twne act ran.Salqat
Inlu ion bak fotwt h act uprlf)
I209
7.5 r1(
An In: uIgrc~nulr ozoaic of calcite anhedra In which "flo,3t'

s ll (0 07 rsz,.) rcnd;d rains of quartz and plagloclase.
I;i
25X Nicols crossed I mm.
Calcite grains showing rhombohedral

cleavage and twinning.
210
A highly pure cjuartzose rock In which quartz grains ranging In
size from 0.15 to 0.85 rrrn* show parallel elongation. The grains have
been pressure w,:lcdcd toocthor In a tightly interlocking mozoic pattern.
M~ost grains shclii unduilatory extinction. Spmall wisps and patchcs of
white mica lie bct,ein somi Srain boundaries. Small particles of iron
oxide are distributed throu~hout.
2X Nicols crossed 1mnu.
Elongate grains of quartz welded together to form a

mozaic pattern. A few shreds of mica are seen in the top left
section of the photograph. Grains are elongated in a direction
from top to bottom.
211
F4 ..It
.. ..S.
... It . ........ Cr ., ,, ,:.fTr o sliind. txo;,.Irj n.na
A very ccorse-Sralned mnozoic Intergrowth of halite crystals

varying frcom 1/14 In. to 1/2 In. long. All grains contain Inclusions
of anhydrite rannfng In size frcm 0.05 to 0.8 im. These are the only
two minnrals prcsont in this section.
3X Irm.
Ab-ove Is close-up photography of core specimen.

Photomicrograph of thin section was essentially
"blank" because of very large crysols.
212
10.1 Sand-.t-ne (rr%- 10~~~rs~Tj
This rock consists of tightly packed subangular grains of quartz,

and small amrounts of plagioclase and microcline all having a well sorted
average grain size of 0.15 to 0.2 rnim.Secondory quartz growth serves
as the predominant ccmenting material ; ho-wovcr, in places., a fine -grainecd
calcite comcnt holds the detrital quartz grains in place.
iJ,
fJ
A"'
25X -Nicols crossed1 u.
Eveni-grained quartz fragments In

fine-grained calcite cement.
213
0,.2 Snndstornj(Crn! rcrlrd). Crorsville.j Tnmzssn
This section shon,,s a tightly Interlocking, welded mozaic of

detrital quartz grains (0.1 rma). Thorp Is no authigenic cemenrt., but
a concentration of carbonate or opaq~ue material along some-, grain
boundaries. The cgrains tend to lie with their long axis parallel
to the bedding planos. Flakes of muscovite are scattered throughout
the section.
AL; A
40S
\NJ
25X Ncolscrossd mm
Mozicinerroth f uatzgrin
In ado ooiysnsoe
2149
10.3 Sendston nuorJrizona
C!avajlrAen
This rock is cornposod of well-sorted, subrounded quartz grains

and minor amnunts of microcline. and plgioclase. The grains vary In
size from 0.07 to 0.26 rm. There is very little authigenic cct1,nt
except for sc;.ne hematite and silica and, as a consequence, the rock
has much open pore space.
V. 4
t'N.
"7 .ipp " ,,
T . I,.' t+" N
"!77
25X - Nicols crossed !mm.

I
Quartz grains of varying size., some with inclusions of

opaque iron oxides, loosely packed together in a porous aggregate
215
11.1 11.2 Schst (I ut-"4r V- )O i
M~ extrc"-1y cremulakod qurtz-mica-garnot schist with minor

amounts of tourrmfline, stcurolitc, sillirncnnite, zircon, and mna~nctite.
The major constittwcnt, quartz,, Is present as an evon-grained (0.4 roi.)
Interlockingj matrix in w!ihfolia of biotite and muscovite are st~attcrcd.
The biotite co~nt.-LIns s-r1.l crystals of zircon.
Tourmalince, present as
small euhedral grains (0.07 r),has a preforred orientation within the
plane of schistosity. Smiall almounts of StaUrolite and sillimanite arc
scattered thrcu~hout the section. Euhedra! garnet crystals, Some up to
0.7 mm. In diamicter, enclose small quartz grains., or have been partially
replaced by rna~matite along Irregular fractures.
-~ u~~--~ ~>* 7,711
rp
4 '? / '4
JLJ
25X -Nicols crossed IMm.
Quartz (gray, bl~rk, and white) enclosing blades of mica

and garnet (large nearly black crystals., upper left).
216
13.1 SI ics tone akn sae.k)I0hcens V:
k,Jcrney
Angular detrital quartz and plagioclase grains (0.o0 to 0.1 mm,.)

lie in a red, hcmatite-rich matrix, speckled with tiny flakes of clay
minerals (?). Thin cleavage flakes of muscovite lie wit!1 their c -
axis roughly perpendicular to the bedding planrs. Irregular narrow
veinleLs of calcite travrersc the section.
25X -Upper Nicol out Imm.
Dark to black areas are hematite which mark Irregular

bedding planes. Light patches are quartz. Elongate
shred at top center of photo is muscovite.
217
14. 1 ('TffS
This tuff contains num~rcus ftansofwle pyroclastic

material (up to 4 nmi. in length) wilth elongate vesicle's and partially
devitrified sphorulitces. Somn frac,,ants of very fine-grairaed recrystal-
lized quartz are laced with secon~dary vwhite mica. Etshodrol crystals and
crystdi chips of zoned plaggloclase and quartz are scattered throughout.
All are Inclosod In a fine-grained dark., dusty, red matrix of devitri-
fled glass and shards which are rimmcd with black Iron dust and small
amounts of chlorite.
25-ios rse mm.
Chp of qurz(ht) akobclrmtras n

shrseepiytevtolsi hrce fti ok
218 ~ Y K
APPEtDIX B
TYPICAL TEST RESULTS
219
0- 0
- 0
0 of
Li $ i
iow
'Sd~is eceaij ___ __
CL
V -o
4. 0 0
220
___ ___.2
~ ~ r ft4N
b. L
oil -. - 1
I JI
CDY
"Ic tsiSoix
____ __ ___221
10a
0, .
~34
C j I
ldG1;S o%
(ssi) b led iu
222.~
I 'iI r 2
- - U) j A
L L iJJ
in5-5-,I -
I II I I
I I
---- ' - - -
I I I ti.
g
L
----- 5
- - - - - -
0
1 LL
Se
i3
0
- . c
(I)
0
- . w
[
'-5
1*
/1
a
- -- -
0 >1'
Og
-@0--
QP
gil
223
* *
(Al it If
IU
V - ~Z C
-
.40
I4%I -i%
[le vekaj
~. U *
r 8
-- 0
isIseA 'czoaIS
!oxVld oixV
--
224-
FA I 12
CC
0 o
-- led me 4sC)ry
.22
'00 0.
cy~
-A a
0 c-
__UP-- 1of
I '
it'd ItcU& -otA
!Zd - CZJ CL-
1 0
Cc
!sd~~~~~~~ ~4
oiV!dlso4 .- o
* 0226
IL L
cr FL
WC
!s s--i I_ - f
227
IC -13
.i.,i Is S piV
b--
C! ~~Q ~
~'iR~
O __~
isd lssj4 oixi Is I~%V ... 40;
228 C
-~~ - I i..ii.J .
CY o
C6)
--
a- -
III ____ __~ j~j_

md
__ sto IS m0o _
_ _ _ _ .;229
C-1)
FO I t
!td IsoI ID
vf VV
!t ImoI.cq
166
- N - - - - -
c ___ui~J S c W -
1~4 21111
-4------to
!sdlssj;S lo!V id IsoiS j~x LL
* 230
I ~~ilf,, nm Vj ob -
Iii10 %b I
I I \I .iiI I I~,-0
~~e7
- - yL ~ Qf
1
N ID
soI
nU) n, ?snf
LcLL.J.r
-i -so l - - II
-- - - -- - _231
7 Ii
I EL
!t Icot Ivij
-I AJ
-0 le C
I- Pb1
LAC ~ '-~ -
ty - - to
q i
- t:- - - -
ied_ -tol *0v o sciSlp
232
C I [k
- -
-5.
0-N ,
CL)
Sopn
--1C
led IsojISjoi%0
_ _ 233
e34
t,-s ng.zr'
CIOI
c. Z.,
ic it go
- I l l
C) Cluj v!
~ ~ ~O*~~e5
onj
* fy
41-~
cJ
N 0
It q-
-t Tsol, "Iix Is ox Isol

234.I~
~arr~-r-~rto o rrr- . .,f r, -- v- -------- .'---- ,-v- ~ r ft~
-6 0
le imls10JA
- - _~ f~~t~I I35
(I)
6
t 0iiOv CYM
"0
cri V) qSi
0 uI
*0 06
9 -YA~
~~~sd~~~~
jox' 'ssooiS oix____
___
~~~~~0236
zizzi Il U TILI 7'iLI:
-l2 "
-- - - -,,,i.~
_ _ I
_ _ , _ _
0-,,____
2>
I-Z
- -_ _-7
MI
11J~L 04
loci
-,rw~m
-a3
tz viI
JsICDIC4 0%
I L I
1 mu
o tJ -
-- .~Oft
led~
1238
~~~ jo% tsoj s t~j o%
L II'
_
_ -_...............
.
'.-
t.
"- I V)o o
........Scr.
239
ae W
ODd
Vsl
As!V
00
a3 cliii (F
CIA 4
4;!
CO W
mdI csil I-IX VAL
isdiItoli4 ov4
oo ox -esl t osl ot I ull
--- 4: C ~O W
-F- 2400
-,rr--,.-n.. ~ - .a.p -- -.. ,ert- .-. 'r
too
CLLv
ci:
II 9GI -o'/-
241>
o ~
so it It I
IKI
~Jo
10 -. 06
4. r.
o cu
!s- Issil ssiS o cx- I-
!Cd ID!X~~C~A~ 242

- n
LL 4
in_ -D - t - i 0
CD 0/
w ID
0--
-i *(fl m
0k
isd
pu sceiS
243>
7-; 1
13,I I
wdp Icai~I I
1cd Ic-jo I 1 *- -
p w f
e esls01V le jor -toj
244
Ek I II
"o" 4[ \
___ ___
...................
-- -- - ~ -245
o ii
!Itd 'lssoIS oIx!V

_ - L
6)i~ -, -
F-
I,
rI LWI rum
I'., tn IS~%,
06
0*
-y - -- - a , - -
IN)
- - - - s~'aj
0'
-
CdL!j .
!s i iov s twS3~x os
-. -. _-- _ _ _ _246 ;
C4 .4
-6 -' _ _ _
LMI
D_
-t is _ _iu s i
"~~,247
2 VII
o in
- a
000
___ 0
~~i2
led 'ssj4S oIxIV - t -s ; - o v - -A
- ----
248- ~
C, IC
0 in
_
'I
00 LL-
00 C)
LO~ Li.
o>
I I-M
cx.
I-
100,
249
Io >
0-- u C
0 K-~u
0 a-0
0 1--. 0.
(I,0 U.
Q v I-
isd ;ssoa4s loiw - -. N
isdIso4 ix
- 0
is ~siSjVisd IssoA4S IonAV
o .4
In ,..250
________V ____
~
I ____
J
-,.
-
Lo_
-i
Jt~~1_____
__ C
c
-
-~ b
Ic
___
__ _ _ I .L __
___
_ _ __ ______0
000
0)401
E 0 0
(no~
;;p
cr-o
!cd~~. a..iSjo%
251
*- 0
U. o u 0!
cyi
Uo
~ 0
0 (flU
j it 'a'
(Vo o I
*3u!lioJ;S ItD!%V
0
_.
0
0
0
tid~~~~~
0i .i ox
o (/
IL
sIm~I!V!S
252L
rNr
CL
Q_ (n
o no
34 .
0C1 C-
- C---
-- -
In
ied -toqlm
253~
J4i )
E cu S,
c0 c0
0 F-
iof It' d' V
-0
N -
if~d ctsoils Ic)i ,s

-- -J
isd 5IstaiS jo~x
- ~~ ~ ~
~---------------- : -
0,
254
COO '
cQ ,,.
IvL
-
CLs
hi --
= Q. (i,
c-C I
____
___ I - - - ___
06II
inI_ _ _
il- --
U >
CL~CJ____ _ __
690 _ _
is jm IsojS U-
__ ~-255
0 0 c
0- _ !
E CCstr 4 b"
Co
J- C
- o
0
0c (A b~
Fo
___ %.. C$
___
q C*
____________ 0 (
w u tC
0~40
.C I'
Nr - 0
-o7
0 c
ird
CJ-56
10ejr C,
s~; ox
ofi
;.ix &s
-Yc
w
C
0a
1L
00
0)
2572
CO 0
.. 0
is' 01 1,' to
no
0 I0 IC)
isd *8204S Ic~iV - 0

N 9
isd IssoelS oi-x'V
I- F~- CI
C4 )
ItZ
+7~
Y J, w
070
65,
258
z
I z
0
C,)
0.
.) z
49..
r-or
O~ _ C
isd 'ssaJlS oiO!V U
259
oo - -
- __
30
CAA
P U.
At _ __ _ _ _ o c- - -
st.
2 Lv)0 4. 4
to it I00
cn w
'7 75
NV
I -1-
,1 I )
- C I
r
_
_ V I
- -d itiis7ssis -- so IDII
260
S - in~ n~f ,~ ,,,,rwCl~i
10
0
0-
tI I U-)-:
0l
V~.a .
___ ox 0sol
261-
I M CY-1
e IY
10 0
0 CL
-
_ 0
CC
'~ V 0
___~I ___ __
N._ 8 I
m26
I~~4I~dz
IZZI_ L
VLi...iH
rc.
C. - fn
u
acto 2 CQ v
2C
Iox
___ es*I
_263
__
~ I~ nI
Ii3
S.,
fil
0 0
~~.i Q '
j~ I/) '~ IP
v'~-. .tam
~. 1 0
cLii
0 --
cc~
0 LA.
* 2o'4 i~d s~oji
i I J I I
e I -J
j IL
- -III
- .- III
-~c U 'I Ii!
V - I
Cl-l
An
CIL- 10
I w
__
__5 (>
is V saj -oA
___ __ 0 st
___ _ _ -L-
-sd - Iso4 !s sal Io

ou
QW
266
*00
cc
UU
. z
0
.0-
C-
0
Ir-
D
0~0
100.
.ui
J-4
isd
joxV ssaiS 0
*__ _ __ -. _ _ _
267 _ i- _ _
E cc
g '0
aa
in~ ? 0 Lo-
!sd~~~~ le~jSjiVC
0 C9 3
VT It -v
isds
gIfXI-A 'ssoj4S -
0 0 0
U'
C,
o*o.
~
!sd ~ ~ ~ ~ ~ lso& ~ ~
DIV!dIso43Ioiq N
~I~I4 268 4
-- J 4
100
00
0 0
o 0_ CL
cr.~ w OD v
-ci% io losjI
- -60
V~ a
0 Iv0
c us 0
___ to CLcc
.2- 0
se I*Q- ,
4 TN
0
27
rz.. - -I-Aft 0
10
E. z
E .2
0 0-
0
z
CLi
N N w
IDoX8 !sd 0,j~

00
271
C,"
10
~C. M V
cli C.
-t IsI;sj! iciV
-
led~
-
o~vIs
~tw
- - n 272
CC
~. ~ wwd
11s~
soI o%
I' I
U.
-- J\
* -
I I 0
1~b? _ __CA
- -- an
0 80
8 8 _0. 8 _ o
0
cyb rw
LL
273:
VA.
a
CO
vi u 6l
- r~~
V
I;
II >
~
V 8
U8C8
0
C,
-- --- -
_ 88 _
Y o - CL__
- 0
on i
__ a a
!sd essiS joxV Id IsoilS loix
27
S~tJ'Wt.~t rwrrt -- A3
liiC,
IjE
cu0in
z__ It
- - __ _ _ - - 0t
i)
isd IseilSjoL%
_ _ _27_
APPENDIX C
STATISTICAL MIETHODS OF ANALYSIS
276
APPENDIX C
STATISTICAL METIDDS OF ANtALYSIS
1. GE14FRAL
"Statistics is concerned with scientific methods for collecting,
organizing, sumarizing, presenting and analyzing data, as well as drawing
valid conclusions and r.king reasonable decisions on the bosis of such
analysis." (Spiegel, l961)
In a narrower sense the term is used to denote the data themselves or

numbers derived from the data as, for example, the averages. In physical
tests on rock the average of a series of readings is said to be representa-
tive of the whole. This average characteristic is Influenced by three Impor-
tant factors which introduce uncertainties in the result:
i) Instrumental errors.
ii) Variations in the sample being tested.
iii) Variations between the sample and the other samples that
might be drawn from the same source.
In collecting data on the properties of rocks it is impossible to

observe even one type from all its numerous sources, let alone all rock types.
Instead of examining all sources called the universe or population only a
small part of one source, called a sample, is examined. A sample may consist
of one single observation, or a million observations. A population can be
finite or infinite.
If a sample is representative of a population, important conclusions

about the population can often be inferred from analysis of the sample. The
phase of statistics dealing with conditions under which such inference is
valid is called statistical inference. Because such inference cannot be
absolutely certain, a statement of probability is often used in stating
conclusions.
If a number of identical specimens from a homogeneous sample were

available for testing, or if the test were nondestructive and could be
repeated a number of times on the same specimen, determination of instru-
mental errors (or errors of observation) would be relatively simple because
the variations could be observed independently of the specimen. Since
277
'r~~c~ct s~ccirwns are not avnil&ble, tVi variatlons encountered in testinj
inclv:: vnriations irhrcrnt within and cm-onn t,%, spccifmns and those caused
by the test eaparatus. Sc;ma of the tests used In deterr,iriing the physical
pro;.ertles of rock are dostructive (cc:rpletely or in part). For these, it is
particularly difficult to distin-uish bet,,ccn instrumental variations and
those caused by chaen rosltion
in c or structure of the rock because the
test crnot be recactad with tl14 IddontIcal specirien (or part thereof). This
prob1c- is further cc.,plIcatcd In the co,,prIson of the results of two or
more different tests, all of which are destructive.
The usual method of recording the highest, lowest, and average readings
fails to give an adequate conception of the rock property under study. For
example, in determining the average Shore (scieroscope) hardness, one sample
may be essentially uniform with occasional stray inclusions or impurities that
account for the low and high readings. Another, with the same average hard-
ness, may be a heterogeneous mixture in which the hardness ranges gradually
from the same low to high value. It is Important to know not only the size of
the deviations from the average, but also their frequency of occurrence
(Obert, et al.) 1946).
2. 11ASURES OF CENTRAL TENIMIDCY
An average is a value which is typical or representative of a set of

data. Since such typical values tend to lie centrally within a set of data
arranged according to magnitude, averages are also called measures of central
tendency. One of the most common averages is the arithmetic mean or mean
which is denoted by the symbol R and computed as follows:
n
Xi + X2 + X +--"'+X Z I
1 2 3, n = i' = _U (c.!)
n ~ n n
This is the sum of all the observations in the sample divided by the number of
observations.
3. MEASURES OF DISPERSION
The degree to which numerical data tend to spread about an ave-age

value is called the variation or dispersion of the data. A commonly used
measure of dispersion is the variance.
278
Vi rI I
a. V~r__en.
Tho variance is the suM of the squares of do&vhtions of Indavldual

observations from the mean, divided by one less than the total number of
observations. The sariple variance, s2 is computed thus:
n(x - 2
2 in1l (C.2)
n-I
The divisor is known as the number of deres of freedom In the

sample, which has the same meaning in statistics as in geometry and mechanics
(Liu and Thornburn, 1965). The result obtained by use of (n-1) in the
denominator instead of (n) is said to represent a better and unbiased esti-
mate of the variance of a pE.ulation from which the s.an.rf was taken. For
large values of (n), say (n) > 30, there is practically no difference between
values of s2 determined by using (n) or (n-1).
b. Standard deviation
The dispersion of the observations about the mean is usually measured
by the standard deviation., which is defined as the positive square root of the
2
variance, s . It is thus denoted by the equation:
n2
z (x -) R
I n-i (c.3)
The standard deviation can also be calculated without first calculating the
sample mean by using the equivalent equation of the form:
n 22
E72 (i-I 1
STm n- I "(C.4)
The sample mean and deviations therefrom are related to the mathematical
statistical principle of least squares in the following manner: If deviations
of the observations are measured from the sample mean, then the sum of their
squares is a minimum (i.e., the sum of the squares of the deviations of a set
of numbers, X,, from any number, a, is a minimum if and only if a R),
279
c~ nfflclcnt of vnr1ntir7,n
The actuit variation as detcrmined fro- tho standard deviation or
;:
othcr rmeaisure of dispersion Is called the abtryc eiffrsiOn ,1fo n-;v ar, for
the purpose of comparing the dogree of variation within the sarnle or between
samples with respect to different propertics, it is more ronvenient to
express the standard dcviation, s, as a percentage of the mean R. This
provide; a measure of the relative dispersion, called the coefficient of
vorIatio ,,V . This coefficient is usually expressed as a percenta(l, and
is given by the equation:
(100) (c.5)
Since it Is a ratio of two quantities having the same unit ot measurement,
the coefficient of variation is independent of the units used. A disadvantage
of the coefficient of variation is that It fails to be useful when R is close
to zero, This is particularly exemplified by the initial values of Poisson's
ratio as reported in Table 4.3.
4. NORMAL DISTRIBUTION, SAPLING AND CONFIDENCE INTERVALS
a. Normal Distribution
Observations that differ by little from the mean occur more frequently
than observations which differ considerably from the mean. Experimental evi-
dence in many fields of science, in conjunction with theory, has led to a
general distribution relating the frequency of occurrence of an observation
to the amount by which it differs from the population mean. This is known as
the normal distribution and is one of the most important examples of a con-
tinuous probability distribution in statistics. The curve representing this
distribution has the physical appearance of a symmetrical bell-shape, extend-
ing infinitely far at both ends as shown in Figure C.I. The equation of the
normal curve is:
2
Y= e (C.6)
in which X = abscissa
Y = ordinate
W=3.14159
e - 2.71Mi8
280
f(x)
2.5%a of area [, c ,+I9G2.5% of area-
At- 3 L.
-2o
L.,
2 j - ' Ir-- 68.27%/ "- -A.- Io t+,, 2oa ++3a
95.45% ... . ...

99-73%0
FIGURE C.I CHARACTERfISTICS OF NORMAL DISTREBUTION,
y
Y ~
(Xt., Yn)
Y=Oa+ bX -.
c
D,
Iu,
Vorioble X
FIGURE C.2 RELATIONSHIP BETWEEN VARIABLES X AND Y
. ..( 2 Y2) , I I I
281
= mean of the distribution or population
0= standard deviation of the distribution or population
If the total area between the curve and X-axis is one (100%), then the por-
tion of the area under the curve between any two vaiues of X is completely
determined by p and o'. By integration, it is found that 68.27% of the total
area lies in the interval L + 1o' while 95.45% of the observations are in
the interval p + 20. Only 0.27% of the total frequency lie beyond .+ 3cr.
Figure C.l also indicates that 5% of the total area lies beyond p. t 1.96a
(Liu and Thornburn, 1965).
If large samples of a specified size. n, from any distribution with

finite variance are taken and the mean values calculated, these sample means
would also form a distribution which is called sampling distribution of the
means and would approximate a normal distribution. In most experimental work
it is safe to assume a normal distribution. In a sampling disLribution of
the means, the standard deviation of the means, s., is related to the popu-
lation standard deviation, a, as follows:
x _
s- = 0 (C-7)
This is referred to statistically as the standard error. On the basis of the

characteristics of Che sampling distributions
22 of the means and the variances,
it is possible to conclude that s can be used as an estimate of a-2 and A of
p. (Liu and Thornburn, 1965).
b. Sampling and confidence intervals
In order that conclusions of sampling theory and statistical inference

be valid, samples must be chosen so as to be representative of the population.
One way in which a representative sample may be obtained is by a process
called random sampling according to which each member of a population has an
equal chance of being included in the sample.
Since the sampling distribution of the means has a mean of p.and a

standard deviation of L and is normal in shape, it is possible to deter-
mine the probability that a sample mean, , might be found within a specified
distance of the population mean, pj based on the characteristics of the
normal curve.
282
Assuming that p and orare known, the chances are 68.27 in 100 that
the sample mean, R, will be found within the distance + - fro
from
Similarly, the probability is 95.45 in 100 that R will be found within
+ 20_.. This distance can be designated to cover any specified probability
as . If x is added to any of these distances, an interval is formed
which includes p at a specified probability, depending on what value is
chosen for Z. (The Z-distribution is a standard normal distribution obtained
from the equation Z = X U-.) If the probability is chosen at 95 in 100,
the Z-value will be 1.96. In other words, L will fall within the range of
S_.96 to (Z + !_&96o- This can be expressed geometrically as follows:
n-6c \r96-n
An interval is set up around R, and then the probability is 95 in 100 that

will be found someplace in this interval. This is called the confidence
interval. That is, there is a 95% confidence of finding the population mean,
p within this interval.
Usually, both the population parameters and a,are unknown; however,

the sample statistics R and s are know. As previously noted, R is an esti-
mate of p, and s of o. When statistics are used instead of parameters, the
confidence interval inequality is:
Sts ts (c.8)
The t-distribution is similar to a normal Z-distribution and closely

approaches it, provided that the sample size is large., Generally, t is
larger than Z for a specified probability. The t-values may be found in
most statistical texts.
The values of R + or x + ts-

- are referred to statistically as
ts4
-irn x
confidence limits and or ts- as the limit of accuracy. The larger the
value of s, the wider the confidence interval will be because the confidence
interval is directly proportional to the size of the standard deviation.
The shared areas in Figure C.l are each composed of 2.5% of the total
area. These are the areas of the distribution of the statistics in which a
283
hypothesis will be rejected, and are known as the critical recilons or regions
of slonificnnc.. The chance of rejecting the hypothesis that p falls within
these critical regions (when In reality it does) is 5% because the critical
regions are determined by the t-value at the 95% punt. Thus, there are about
five chances in 100 that we would reject the hypothesis when it should be
accepted, I.e., we are 95% confident that we have made the right decision.
This chance Is called the level of significance and is denoted by the Greek
letter, (X . It is usually stated tOat a result is significant if the hypo-
h
thesis is rejected with a= .05 and hIL1 significant if It ;s rejected with
CC- .01. These are designated conventionally as iifica-: t the % level
and sicniflcant at the 1% level respectively. It can be seen that when the
level of significance is raised, the confidence interval will be wideneJ,
because the chances of rejecting a true hypothesis are decreased (Liu and
Thornburn, 1965).
5. CURVE FITTINJG AND THE METHOD OF LEAST SQUARES
a. Cu-ve fitting
It is frequently desirable to express a relationship existing between

two (or more) variables in mathematical form by determining an equation con-
taining the variables. When a series of concomitant observations (X,. Y ),
(X2 ) Y2 )p --- (Xny Yn) is plotted on a rectangular ccc'rdinate system, the
resulting set of points is sometimes called a scatter diagram. If the data
appear to be well-approximated by a straight line, a linear relationship is
said to exist between the variables. If a curved line is a better approxi-
mation, a non-linear relationship is said to exist. The general problem of
finding equations for approximating curves which fit given sets of data is
called curve fitting.
b. The method of least squares
In order to avoid individual judgment in constructing lines, parabolas,

or other approximating curves to fit sets of data, it is necessary to agree
on a definition of a "%est-fitting line," %est-fitting parabola," et cetera.
Figure C.2 shows a scatter diagram obtained from a series of observations.
For a given value of X, say XI, there is a difference between the value Y
and the corresponding value as determined from the "curve" or line C. As
284
indicated in the figure, this difference is denoted by D1, which is referred
to as a deviation, error, or residual and may be positive, negative, or zero,
Similarly, corresponding to the valucs X2, -----. XnW we obtain D2 P -----.. Dn
Of all curves approximating a given set of data points, the curve having the
ert tha
pro
property that D2 .. .. + nI is a minimum,.9 is called a best-fitrii!'
curve. A curve having this property is said to fit the data in the lcast-
square sense and is called a least-square curve. Thus, the line C in Figure
C.2, is a least-square line if it has this property.
The above definition applies when X is the independent variable, and

Y the dependent variable. If X is dependent, the definition is modified by
considering horizontal instead of vertical deviations, which amounts to an
interchange of the X and Y axes. These two definitions, in general, lead to
different least-square lines. Thus. when these two lines approach coincidence,
there is an indication that the data are very well described by a linear
relationship.
Non-linear relationships can usually be reduced to linear relation-

ships by appropriate transformation of variables, such as logarithmic func-
tions. In such cases, the same methods for curve fitting may be applied as
used for the linear relationships.
6. CORRELATION AND REGRESSION
The degree of relationship between variables, or correlation, seeks

to determine how well a linear or other equation describes the relationship.
If all values of the variables satisfy an equation exactly, we say that the
variables are perfectly correlated, or that there is a perfect correlation
between them.
On the basis of sample data, we can estimate the value of a variable

Y. corresponding to a given value of a variable X, from a least-square line
(or curve) which fits the sample data. The resulting line is called a
regression line of Y on X, since Y is estimated from X. Conversely, the
value of X may be estimated from the regression line of X on Y in which case
X is the dependent and Y the independent variable. In general, the regres-
sion line or curve of Y on X is not the same as the regression line or curve
of X on Y. When only two variables are involved, we are concerned with
285
sm~pe corrcinticn and slr.ile rcqression. -lcn more tt;3n two vzric-bles are
involvcd, v:e are concerned with mult.olel correlation and ri-ult ipe rcfiressi.n.
a. _I.nenr correlation
If all points in a scatter diDorzrn eppcar to lie near a line, as In

Figure C.2, the correlation is called line,. If Y tends to increase as X
Increases, the correlation is a p2ositive or a direct crrrelot~on. If Y tends
to decrease as X increases,, the correlation is a nef tijv, or an inverse
correlation. If there is no relationship indicated between the variables,
they are uncorrelated. The correlation is non-linear if all points appear
to lie near some curve.
b. Least-souare reqression
The least-square regression line of Y on X has the equation:
Y V a + bX (C.9)
In which Y = dependent variable

X - independent variable
a - Y-intercept of regression line
b - slope of regression line, also
called the coefficient of recression
By means of the calculus, the so-called normal equations are obtained,

and expressed as follows:
-Y -= an
_XXY + +b bXZX2
a Z( (C.0O)
These two equations are solved simultaneously to obtain a and b:
a =EY
(ZY)(x ) - (ax)(EXy)
nX2 _ ( X) 2
(c.ll)
b nZXY - (X) (ZY)
nZX2 (DO 2
The regression line of X on Y is similarly obtained. However, the

resulting least-square line is, in general, not the same as noted previously.
The two least-square regression lines thus obtained both pass

through the point (R,), which is the centroid or center of gravity of the
data. These regression equations are identical only if all points of the
286
scatter diagram lie on a line. In ;uch case there Is a rfeet corrnlation
between X and Y (Spiegel, 1961).
c. Crenfficient of correlation
A correlation coefficient is a mnasure of Intensity of association or

relationship between variables. A regression coefficient describes the chansc
in the dependent variable as a result of chnnging the Independent variable.
Unlike a variance or a regression coefficient, the correlrtion coefficient is
independent of the units of measurement; it is an absolute or dimensionless
quantity. The use of X and Y is no longer intended to Imply an Independent
and a dependent variable.
The total variation of Y is defined as T(Y-9) 2 i.e., the sum of the

squares of the deviations of the values of Y from the mean 9. This can be
written by the equation:
-(y.9)2 = (yY-Yest.)2 + gYest. 2 (C.12)
in which Yest. = value of Y for given values of X as estimated from (C.9)

)2
Z(YYest.) = unexplained variation
Z(est.-) xlie
~~ =explinedvariation
The deviations Yest.- have a definite pattern while the deviations Y-Yest.
behave in an unpredictable manner.
The ratio of the explained variation to the total variation is called

2
the coefficient of determination denoted by r 2 hence is always positive.
The square root of the coefficient of determination is called the coefficient
of correlation) r, which is given by:
explained variation g* est. -

Z(Y. )2
(c.13)
total variation
The quantity, r, varies between -1 and +1, which are the values for a per-
fect correlation. The signs + denote positive or negative linear correlation,
respectively. The practical range of r varies with opinion from one field of
research to another. In some studies. investigators will look upon r - 0.90
as small, while in others r = 0.20 may be considered as unusually large.
287
As previously indicated, no relation:hip Is perfect; therefore, the
actual values will not coincido vlith the theoretical values estimrted from
the regresslcn )in--. If the scatter Is definitely measured, the variation
is datermined and a ranoe is estebl ished within which a given percentage of
values will fall The standard crror of nsti"rnt s is a mneasure of the
y.X'
scatter about the regression line of Y on X and is given by the equation:
i ~Yes
y.x s stZ(C.14)
2 'Y
Similarly, the standard error of estimate of X on Y is obtained by inter-

changing Y with X in this equation.
The standard error of estimate has properties analogous to those of

the standard deviation. If lines are constructed parallel to the regression
line of Y on X at respective vertical distances + syx , 2s yxand 3
from it, approximately 680%, 95% and 99.7% of the sanmple points would be
included between thesc lines.
e, Confidence limits
An unbiased estimate of the true variance about regression is given

by the residual mean square with n-2 degrees of freedom. It is defined by
the square of equation (C.14) as follows:
2 Z(Y-Yest.) 2
yx = n - 2
Figure C.3 shows that a single standard deviation of Y for a fixed

X (i.e., standard error of estimate) does not apply to all Yest. 's but must
depend upon the X-value that determines the Y-population. For a fixed set
of X-values, R is a constant, while - and b are variables. Variation in
raises or lowers the regression line parallel to itself (thus increasing or
decreasing all estimates of means by a fixed valun). Variation in b rotates
the regression line about the point .y,,but has no effect upon th-. estimate
of the mean if X=R. Otherwise, it increases the estimate of the mean in
proportion to the size of (X-R). This is readily seen from the equation
that estimates the population mean, Yest. + b(X-R).
288
Y
Y '0 + bX
Effect of voriaoion
in olone
"Effect of voriation ir. b olone
A -- Confidence Belt
FIGURE C.3 EFFECT OF SAMPLING VARIATION ON

REGRESSION ESTIMATES OF POPULATION
MEANS (After Stool Q Torrie, 1960)
289
A stan d,,rd ,:viatIon I Iic, le to an cs tImn.te of a n:n m.ust a IIo
for variation in both 9 and b0 and foi- thc distr.,c-3 (X-R). The variance of
lI 0,2 S2
is an estimate of .. . namely yj . An estirmte of the variance of b is
n n
gqven by the equation:
2 s2
2 . b.(c.16)
The required variance of an estimate, Yest b(X-R), of a population mean

is given by the sum of the variances of 9 and b(X-R) as follows:
2 2
Sy.x _i (C.17)
This variance increases as (X-;) increases. If t times the standard deviation

were plotted in conjunction with the regression line, the confidence belt
shown in Figure C.3 would be formed.
To set a confidence interval on the estimate of the mean, the variance

in equation (C.17) is added to the regression equation. The 95% confidence
limits are then defined as follows:
CL(Y~
+(-):tt s+ _ (C.18)
est. 05 y.x Fn _(XR))2
where t is for n-2 degrees of freedom.
If a prediction is to be made, it is usually a prediction of individual

events rather than for the population. Thus, the random element is an addi-
tional source of uncertainty and equation (C.18) is modified as follows:
CL(Y) - + b(X-R) t. t.05 Sy.x It (C.19)
Using this procedure, a confidence belt for Y may be plotted which is some-
what broader than that defined by (C.18). In general, about 5% of the
sample points are expected to fall outside the belt given by (C.19).,
(Snedecor, 1956).
290
tlerican Geological Institute, (W2), f and Ri atcd
Sclenc
a,, The Anarican Geological Institute, Wshington, D. C.
American Society for Testing and Materials, (lc9C), "Resistance to Abrasion

of Small Size Coarse Aggregate by use of the Los An-nles Machine"
Tentative Mothod ASTIM- Designation: C 131-6?4TO tEi: StandaresJ part 10,
p. eo.
American Society for Testing and Vi3terials, (KSIc), 'VArasion Resistance of

Concrete" Tentative Method A[.Zfl Designation: C 418-VkT, ASi Standards,
part 10, p. 309.
American Society for Testing and Materials, (1964), '"elative Resistance to

Wear of Unglazed Ceramic Tile by the Taber Abraser," Tentative Method
ASTM Designation: C 501-62T, ASTI Standards, part 13, p. 455.
American Society for Testing and Materials, (1964), '71esistance to Abrasion

of Large Size Coarse Aggregate by use of the Los Angeles Machine,"
Tentative Method ASTM Designation: C 535-64T. ASTH Standards, part 10,
p. 361.
American Society for Testing and Materials, (1964), "Standard Method of Test
for Abrasion of Rock by use of the Deval Machine," Standard Method ASTM
Designation: D 2-33, ASTM Standa1rds, part 10, p. 363.
American Society for Testing and Mcterizls, (1964), "Standard Method of Test
for Toughness of Rock," Standard Method ASTM Designation: D 3-18, ASTM
Standards, part 10, p. 365.
American Society for Testing and Materials, (1964), "Resistance of Transparent

Plastics to Surface Abrasion," Standard Method ASTM Designation: D 1044-
56, ASTM Standards, part 27, p. 442.
American Society for Testing and Materials, (1964), 'itesistance to Abrasion

of Plastic Materials," Standard Method ASTM Designation: D 1242-56$
ASTM Standards, part 27, p. 472.
American Society for Testing and Materials, (1964), "Bierbaum Scratch Hardness
of Plastic Materials," Tentative Method ASTM Designation: D 1526-58T9
American Society for Testing and Materials, (1964), "Identification Hardness

of Plastics by Means of a Durometer," Standard Method ASTM Designation:
D 1706-61, ASTM Standards, part 27, p. 570.
American Society for Testing and Materials, (1965), "Conpressive Strength of

Natural Building Stone," Standard Method ASTM Designation: C 170-50,
ASTM Standards part 12, p. 144.
291
American Society for Testing and Materials, (1965), "Brinnell Hardness of
Metallic Materials," Standard Method ASTM Designation: E 10-65,
American Society for Testing and Materials, (1965), "Rockwell Hardness and
Rockwell Superficial Hardness of Metallic Materials," Standard Methods
ASTM Designation: E 18-65, ASTM Standards, part 30, p. 187.
American Society for Testing and Materials, (1965). '"iamond Pyramid Hardness
of Metallic Materials," Standard Method ASTM Designation: E 92-65,
ASTM Standards part 30, p. 317.
Appli, F. C., and Gatley, W. S., (1961), "Rate of Loading Effects in Chisel
Impact," Proceedings of the Fourth Symposium on Rock Mechanics, Bulletin
of the Mineral Industries Experiment Station, The Pennsylvania State
University, p. 229.
Blair, B. E., (1955), Physical Properties of Mine Rock, Part III, U. S.

Bureau of Mines, R. I. 5130.
Blair, B. E., (1956), Physical Properties of Mine Rock. Part IV, U. S.

Bureau of Mines, R. I. 5244.
Bond, F. C., (1946), "Crushing Tests by Pressure and Impact," Trans. AIME,
Mining Technology, Vol. 169, p. 58.
Bond, F. C., (1961), Crushing and Grinding Calculations, Allis-Chalmers

Manufacturing Company, Milwaukee, Wisconsin.
Bowker, A. H., and Lieberman, G. J., (1959), Engineering Statistics,

Prentice-Hall, Inc., Englewood Cliffs, N. J.
Brace, W. F., (1960), "Behavior of Rock Salt, Limestone and Anhydrite During
Indentation," Journal of Geophysical Research, Vol. 65, No. 6, p. 1773.
Brace, W. F., (1963), "Brittle Fracture of Rocks," International Conference

on the State of Stress in the Earth's Crust, Santa Monica, June.
Brown, P. D., and Robereshaw, J., (1953), "The In Situ Measurement of Young's
Modulus for Rock by a Dynamic Method," Geotechnique, Vol. III, No. 7,
p 283.
Burbank, B. B., (1955), '"easuring the Relative Abrasiveness of Rocks,

Minerals and Ores," Pit and Quarry August, p. 114.
Cheatham, J. B., Jr., (1958), "An Analytical Study of Rock Penetration b a

Single Bit Tooth," Proceedings of the Eighth Drilling and Blasting
Symposium, University of Minnesota, October, p. IA-25A.
Christensen, C. J., (1947), "Diamond Coring in Rangely," Presented at

Regional Meeting of AIME, Denver, Colorado, September.
Coates, D. F., (1964), "Classification of Rocks for Rock Mechanics,'' Interna-

tional Journal of Rock Mechanics and Mining Sciences, Vol. 1, No. 3,
p. 421.
292
Colback, P. S. B., and Wiid, B. L., (1965), "The Influence of Moisture Con-
tent on the Compressive Strength of Rock," Third Canadian Symposium on
Rock Mechanics University of Toronto, January.
Dana, E. S., (1932), A Textbook of Mineralogy, John Wiley & Sons, New York.
D'Andrea, D. V., Fischer, R. L., and Fogelson, D. E., (1964), "Prediction of

Compressive Strength from Other Rock Properties," Quarterly of the
Colorado School of Mines, Vol. 59, No. 4-Part B. p. 623.
Deere, D. U., (1963a), Class Notes from Advanced Graduate courses in Rock
Mechanics.
Deere, D. U., (1963b), "Technical Description of Rock Cores for Engineering

Purposes," Felsmechanik und Ingenieurgeologie/Rock Mechanics and Engi-
neering Geology, Journal of the International Society of Rock Mechanics,
Vol. I, No. 1, p. 16.
Deere, D. U., (1964), Private communication.
Devries, R. P., (1911), "Hardness in its Relation to Other Physical Proper-

ties, Proceedings ASTM, Vol. 11, p. 726.
Dvorak, A., (1957), '"Feld Test of Rocks on Dam Sites * Proceedings of the
Fourth International Conference on Soil Mechanics and Foundation Engi-
neering, Butterworths Scientific Publications, London, Vol. 1, p. 221.
Eshback, 0. W., (1952), Handbook of Engineering Fundamentals, John Wiley &

Sons, Inc., New York.
Evison, F. F., (1956), "The Seismic Determination of Young's Modulus and

Poisson's Ratio for Rocks In Situ," Geotechnique, Vol. VI, No. 3, p. 118.
Fairhurst, C., and Lacabanne, W. D., (1957), "Hard Rock Drilling Techniques,"
Mine & Quarry Engineering, April, p. 157, May, p. 194.
Fairhurst, C., (1961a), '.ave Mechanics of Percussive Drilling," Mine & Quarry
Engineering, March, p. 122, April, p. 169.
Fairhurst, C. (1961b), "Laboratory Meas:urement of Some Physical Properties

of Rock,' Proceedinqs of the Fourth Symposium on Rock Mechanics,
Bulletin of the Mir.nral Industries Experiment Station, Pennsylvania
State University, p. 105.
Gilbert, B. W., (1954), Shore Scleroscope Hardness Tests Made on Mohs' Scale
Minerals from Talc to Quartz, Inclusive, Department of Mining and
Metallurgical Engineering, University of Illinois.
Greaves, R. H., (1909), "The Physical Interpretation of Hardness as Measured

by the Shore Scleroscope," Proceedings Institute of Civil Engineering,
London, Vol. 181, p. 478.
293
Gregory, A. R., (1963), "Shear Wave Velocity Measurements of Sedimentary Rock
Samples under Compression," Proceedings of the Fifth Symposium on Rock
Mechanics The Macmillan Company, New York, p. 439.
Griffith, J. H., (1937), Physical Properties of Typical American Rocks,- Iowa

Engineering Experiment Station Bulletin 131.
Grout, F. F., (1940), Kemp's Handbook of Rocks, D. Van Nostrand Company,

Inc., New York.
Gyss, E. E., and Davis, H. E., (1927), '"The Hardness and Toughness of Rocks,"
Mining and Metallurgy' Vol. 8, p. 261.
Hamrol, A., (1961), "A Quantitative Classification of the Weathering and

Weatherability of Rocks," Proceedings of the Fifth International Con-
ference on Soil Mechanics and Foundation Engineering, Paris, Vol. II,
p. 771.
Hardy, H. R., Jr., (1959), Standardized Procedures for the Determination of

the Physical Properties of Mine Rock under Short-Period Uniaxial Com-
pression Department of Mines and Technology Surveys, Ottawa, Canada,
Mines Branch Technical Bulletin TB 8.
Harley, G. T., (1926), "Proposed Ground Classification for Mining Purposes--

I and II." Engineering and Mining ournal Vol. 122, No. 10 and 11,
pp. 368 and 413.
Hartman, H. L., (1959), "Basic Studies of Percussion Drilling," Mining

Engineering. Vol. 11 p. 6 8 .
Harvey, R. D., (1963), Impact Resistance of Illinois Limestones and Dolomites,

Illinois State Geological Survey Circular 345.
Head, A. L Jr (1951),. "A Drillability Classification of Geological

Formations., World Oil, October, p. 125.
Huang, W. T., (1962), Petrology McGraw-Hill Book Company, Inc., New York.
Hucka, V., (1965),"' Rapid Method of Determining the Strength of Rocks In

Si*u," International Journal of Rock Mechanics and Mining Sciences,
Vol. 2, No. 2, p. 127.
Ide, J. M., (1936), "Comparison of Statically and Dynamically Determined

Young's Modulus of Rocks," Proceedings of the National Academy of
Science Vol. 22, No. 2, p. 81.
Jackson, F. H., (1916), Methods for the Determination of the Physical

Properties of Road-Building R3.L,, U. S. Department of Agriculture,
Department Bulletin 347.
Jaeger, J. C., (1962), Elasticity, Fracture and Flow, John Wiley & Sons,
Inc., New York.
294
Jastrzebski, Z. D., (1959), Nature and Properties of Engineering Materials,
John Wiley & Sons, Inc., New York.
John, K. W., (1962), "An Approach to Rock Mechanics," Journal of the Soil
Mechanics and Foundations Division, Proceedings of the American Society
of Civil Engineers, Vol. 88, r SM4, Part 1, p. 1.
Judd, W. R., and Huber, C.;, (1961), "Correlation of Rock Properties by

Statistical Methods," International Symposium on Mining Research,
Rolla, Missouri, February.
Kapadia, A. H., (1951), Correlation of Scleroscope Hardness with Physical

and Elastic Properties of Rock, M. S. Thesis, Colorado School of Pines,
King, R. F., and Tabor, D., (1954), 'The Strength Properties and Frictional
Behavior of Brittle Solids," Proceedings Royal Society Ser. A, 223,
p. 225.
Klapka, K. J., (1948), "Drilling with Industrial Diamonds," World Oil,

October, p. 101.
Knill, J. L., and Jones, K. S., (1965), 'The Recording and Interpretation
of Geological Conditions in the Foundations of the Roseires, Kariba,
and Latiyan Dams," Geotechnguee, Vol. XV, No. 1, p. 94.
Knoop, F., Peters, C. G., and Emerson, W. B., (1939), "A Sensitive Pyramidal-
Diamond Tool for Indentation Measurements," RP 1220, Journal of Research
of the National Bureau of Standards, Vol. 23, No. I, p. 39.
Kolsky, H., (1963), Stress Waves in Solids, Dover Publications, Inc., New
York.
Kraatz, P., (1964), Rockwell Hardness as an Index Property of Rocks

M. S. Thesis, University of Illinois.
Krynine, D. P., anl Judd, W. R., (1957), Principles of Engineering Geology.
and Geotechnics McGraw-Hill Book Company, Inc., New York.
Lea, F. C., (1936), Hardness of Metals, Charles Griffin & Co., Ltd., London.
Liu., T. K., and Thornburn, T. H., (1965), Engineering Index Properties of

Some Surficial Soils in Illinois, University of Illinois Engineering
Experiment Station Bulletin 477.
Mahin, E. G., and Foss, G. J., (1939), "Absolute Hardness," Trans. American
Society for Metals, Vol. 27, p. 337.
Mather, W. B., (1951), '"ock Hardness as a Factor in Drilling Problems,"'

Trans. AIME, Vol. 190, p. 173.
McCarthy, J. H., (1964), Elec.tronic Methods for Determining Young's Modulus

and Poisson's Ratio of Portland Cement Technical Report R 295,
U. S. Naval Civil Engineering Laboratory, Port Hueneme, California.
295
McCray, A. W., (1948) "Diamond Coring of Chert, Limestone and Dolonite of
West Texs Permian Basin," Petroleum Vol. 19, No. 6, p. 76.
Mielenz, R. C., (1961) Ind FnninP,

,rinnr ,,
Pro-certies of
, !ous
Rocs U. S. Department of the Xnterior, Bureau of Reclamation, Denver,
Engineering Monographs, No. 1.
Rohs, F., (1824), Grundriss dpr Min-ranloqi, English translation by W.

Haidinger, (12), Tratse on iineralo_gy Constable and Co., Ltd.,
Edinburgh.
Nicholis, H. R.) (1961), In Situ Determination of thej Oniic Elastic Con-

stants ofRock, U. S. Bureau of hines R. I. 5b88.
Obert, L., Windes, S. L., and Duvall, W. 1., (1946), Standardized Tests for
Determiinin the. Physical Properties of Mine Rock, U. S. Bureau of
Mines, R. I. 3891.
Onodera, T. F., (1963), "Dynamic Investigation of Foundation Rocks In Situ,"

Proceedings of the Fifth Symposium on Rock Mechanics, The Macmillan
Company, New York, p. 517.
Panama Canal Company, (1947), Isthmian Canal Studies Memo 210, Department
of Operation and Maintenance, Special Engineering Division, April.
Panama Canal Company, (1959), Geologqcal Logs of Drill Holes for Construction
of Balboa Bridge, Substructure, Canal Zone, Balboa Heights, April.
Peck, R. B., Hanson, W. E., and Thornburn, T. H., (1953), Foundation

Engineering, John Wiley & Sons, Inc., New York.
Pennington, J. V., (1954), '"ock Failure in Percussion," The Petroleum

Engineer, May, p. B-76.
Pomeroy, C. D., (1957), ;A Simple Method for the Assessment of Coal Strength,'
Journal of the Institute of Fuels, February, p. 50.
Protodyakonov, M. MI, (1960), "New Methods of Determining Mechanical Proper-

ties of Rocks," Proceedings Third International Conference on Strata
Control, Paris, p. 187.
Protodyakonov, M. M., (1963), '4echanical Properties and Drillability of

Rocks," Proceedings of the Fifth Symposium on Rock Mechanics, The
Macmillan Company, New York, p. 103.
Reich, H., (1930), 'Geoligische Unterlagen der Angewandten Geophysik,"

Handbuch der Experimentalphysik, Vol. 25, part 3.
Reichmuth, D. R., (1963), "Correlations of Force-Displacemern Data with

Physical Properties of Rock for Percussive Drilling Systems,"
Proceedings of the Fifth Symposium on Rock Mechanics, The Macmillan
296
Richards, C. W., (1961), .n n tfrlals SctEnr, Wadsworth Publishing
Company, Inc., Belmont, California.
Ridgway, R. R., Ballard) A. H., and Bailey, B. L., (1933), '"-lardness Values
for Electrochcmical ProduIcts "Trans The Electrochemical Society,
p. 369.
Vol. 63,
Rinehart, J. S., Fortin, J. P., and Burgin, 1., (191), 'ropaqation Velocity
of Longitudinal Waves in Rocks" of the Fourthly.
...........
Rock Hechanics Bulletin of the Mineral Industries Experiment Station,
Pennsylvania State University, p. 119.
Rocha, M., (1964), "Hechanical Behaviour of Rock Foundations in Concrete

Dams " qi International Conference on Lar. Dams Edinburgh, R. 44,
Q. 26) p. 785.
Rollow, A, G., (1963), "Estimating Drillability in the Laboratory," Proceed-

Ings of the Fifth Symposium on Rock Mechanics The Macmillan Company,
New York, p. 93.
Roxborough F. F., and Whittaker, B. N., (1964), "Roof Control and Coal Hard-
ness,' Colliery Engineering December, p. 511.
Roxborough. F. F., and Whittaker, B. N., (1965), "Roof Control and Coal Hard-
ness, Colliery Engineerinq, January, p. 19.
Scott, F. L., (1946), '"ard-Rock Rotary Drilling," The Oil and Gas Journal.
October, p. 74.
Seely, F. B., and Smith, J. 0., (1956), Resistance of Materials. John Wiley
& Sons, Inc., New York.
Seely, F. B., and Smith, J. 0., (1959), Advanced Mechanics of Materials,

John Wiley & Sons, Inc., New York.
Shepherd, R., (1950), 'Vhysical Properties and Drillability of Mine Rock,"

Colliery Engineering, December, p. 48.
Shepherd, R., (1951), '"hysical Properties and Drillability of Mine Rock,"

Colliery Engineering. January, p. 28, February, p. 61, March, p. 121.
Shepherd, R., (1953), '"rillability of Rock in Rotary Drilling," Manual of

Rock Blasting, AB Atlas Diesel, Stockholm.
Singh) M. M., and Hartman, H, L., (1961), 'Hypothesis for the Mechanism of
Rock Failure Under Impact," Proceedings of the Fourth Symposium on Rock
Mechanics, Bulletin of the Mineral Industries Experiment Station,
Pennsylvania State University, p. 221.
Snedecor, G. W., (1956), Statistical Methods, The Iowa State University

Press., Ames, Iowa.
Sosman, R. B., (1927), Properties of Silica, American Chemical Society,

Monograph No. 37, Chemical Catalog Co., Inc., New York.
297
Spiegel, M. R., (1961), '.or and of Sttistics, Schaum Publishing
Co-pzny, .w York.
Spock 1. E., (1953)., Givf:,' to thn Study.v of P.cr... irper F.Brothers,

1.
PublishKrs, ljzw York.
Statistiral Sarvice Unit, (1.4), ....... lorrr
Stat stItl,, ! ysjs Uniersity of Illinois, Urbana) Illinois.
Steel, R. G. 0D, and Torrie, J. H., (1960)j, rincinles and Procedures of

Statistics kcGraw-Hill Cook Coimpany, 11c.
e York.
t~w
Stokes, I. L., and Varnes, D. J., (1955), 'Vlossary of Selected Geologic

Terms, Colorado Scientific Society Procedinn,, Vol. 16, Denver.
Sutherland, R. B., (1963)., "Some Dynamic and Static Properties of Rock,"

Proccedin s of the Fifth Symnosium on Rock ltlechanics The Macmillan
Tabor, D., (1947), "A Simple Theory of Static and Dynamic Hardness," Proceed-
ings Roya] Society of London Ser. A., Vol. 192, p. 247.
Tabor, D., (1951)0, The H3rdness of Metals, Oxford at the Clarendon Press.
Tabor, D., (1954), '14ohs' Hardness Scale -- A Physical Interpretation,,"

Proceedinqs Physical Society of London, Sect. B: Vol. 67, p. 249.
Tabor., D., (1956), '"he Physical Meaning of Indentation and Scratch Hard-
ness;, British Journal of Applied Physics. Vol. 7, P. 159.
Talobre, J., (1957), La Mecanique des Roches, Dunod, Paris.
Terzaghi, K.J (1946), '"ock Defects and Loads on Tunnel Supports," Section
I, Rock Tunneling with Steel Supports, The Commercial Shearing &
Stamping Co., Youngstown, Ohio.
Terzaghi, K.Y and Peck, R. B.: (1948), Soil Mechanics in Engineering Practice,
John Wiley & Sons, New York.
Tocher, D., (1957), "Anisotropy in Rocks Under Simple Compression," Trans.

American Geophysical Union, Vol. 38, No. 1, p. 89.
Travis, R. B., (1955), "Classification of Rocks,,"Quarterly of the Colorado

School of Mines Vol. 50, No. I.
U. S. Army Corps of Engineers, (1961), '"resenting Subsurface Information in

Contract Plans and Specifications," Engineering and Design Manual, EM
1110-1-1806, 24 January.
U. S. Bureau of Mines, (0962), A Study of Mine Examination Technioues for

Detecting and Identifying Underground Nuclear Ex.losions Information
Circular 8091.
298
U. S. Bureau of Reclamation, (15w, Effcct of Cr's In rnc-.rete U!'rqt
PyniaMijc J131SU rcm--,nts of EInvt I V)rl I (sU Llujcau of Rac~cm.-tion
Materials Lab. Rcport N(o. C-33, kesearch and Geology Division.
U. S. Cureau of Reclamation, (1953), hysic .r rtis of SogmaTypical

Foundation Rocks U. S. Bureau of Recla;)ttion Concrete Lab tPR'ort
No. SP-39.
Von Moos, A., and De (juervain, F., (l4), Tchnische Cinstunrd,

Verlag Birkheuser, Case1.
Walsh, J. B., (1965), "The Effect of Cracks on the Uniaxial Elastic Cor,,,pres-
sion of Rocks," Journal of Geophysical Research Vol. 70 No. 2, p. 399.
Watstein, D., (1953), "Effect of Straining Rate on the Comprehensive Strength

and Elastic Properties of Concrete)" fmerican Concrete Institute Journal
Vol. 24, No. 8, P. 729.
White, R. Z., (1963), '"retest Tiebacks and Drilled-In Caissons ,"Civil-

Engineering, April, P. 36.
Williams, S. R., (1942), Hardness and Hardness 1,easurcmnts, American

Society for Metals, Cleveland, Ohio.
Windes, S. L.) (1949), Physical Properties of Mine Rock P U. S.

Bureau of Mines R. I. 4459.
Windes, S. L., (1950), PLhjsical Proprties of Vine Rock Part 11 U. S.

Bureau of Mines R. I. 4727.
Wolansky, D., (1949), "Zur Frage der Harteprufung der Karbongesteine nach
dem Ruckprallverfahren," GIUctjaui, Vol. 85, January.
Wooddell, C. E., (1935), '%iethods of Comparing the Hardness of Electric

Furnace Products and Natural Abrasives," Trans. The Electrochemical
Society Vo!. 68, p. Ill.
Woolf, D. 0., (1930), The Results of Physical Tests of Road-Building Rock.

U. S. Department of Agriculture Misc. Pub]. 76.
Wuerker, R. G., (1953), '"he Status of Testing Strength of Rocks," Trans.

AIME, Mining Engineering, p. 1108.
Wuerker, R. G., (1956), Annotated Tables of Strength and Elastic Properties

of Rocks Published by Petroleum Branch, AINE, December, Paper No. 663-G,
Wuerker, R. G., (1959), "Influence of Stress Rate and Other Factors on

Strength and Elastic Properties of Rocks," Quarterly of the Colorado
School of Mines, Vol. 54, No. 3, p. 3.
Wyllie, M. R. J., Gregory, A. R., and Gardner, G. H. F., (1958), "An

Experimental Investigation of Factors Affecting Elastic Wave Velocities
in Porous Media," Geophysics, Vol.. 23, No. 3, p. 459.
299
ZisnW. A,, (19,13), "Cr-parlson of th2 Statically and Seismologically
Detorrmined Elastic Constants of Rocks r ........... of t
n Vol. 19, No. 7, P. 680.
Zwikker, C., (19510), LinLrLL erties of Solid Kriterfals, Interscience

Publishers, Inc., lroa, York.
300
Unclarsi fled
10 t(;INATIN ACTII'IIV (C v 20 I'lr@

Cjofu
P2*. T SECUPITY
Patno) C LAZSIlCATION
University of Illinois UnclassIfled
Department of Civil Engineering 2b "nupI
Urbana, Illinois _
3 nrPORlT TITLE
ENGINEERING CLASSIFICATION AND INDEX PIROPERTIES FOR INTACT ROCK
1 February 1964 through I April 1966

5 4 (S) (Levi ncaln, 1ItW rmt, Iinitial)
AUTHO
Deere, D. U.; Miller, R. P.
6. RZPORT DATE )7a. OAG'S 7b. NO. o0 Rt
December 1966 3241 140

8S CONTRACT Oft GRANT NO. SM. OnIGINATORVP REPORT NUtMDEP(S)
AF29(601)-6319
NO.5710
b. PnROJEC AFWL-TR-65-116
c.Subtask 13.144 3b. o THIF R5PORT NCo ) (Any othr numbadintmay be av.i&ied
d
10 AVA ILADILITY/LIMITATION NOTICES
Distribution of this document is unlimited.
I1. SUPPL. EV.NTARf NOTZI 12. SPONdSORINJG MILITARY ACTIVITY

Air Force Weapons Laboratory (WILDC)
Kirtland Air Force Base, New Mexico 87117
13 ABSTRACT
Physical and elastic properties of NX-size rock core from 27 localities were
investigated in order to develop an engineering classification 6ystem for intact
rock, and also to develop index properties related to important physical and
engineering characteristics. Thirteen rock types are represented. Laboratory
tests were conducted on these rocks as follows: 'unit weight, Shore scleroscope
hardness, Schmidt hammer hardness, abrasion hardness, absorption, sonic-velocity
stress-strain under cyclic loading to 5,000 psi, uniaxial stress-strain to
failure, and point-load tensile strength. A total of 257 specimens with L/D
ratios of 2:1 were tested. Statistical studies were conducted with the IBM 7094
computer to determine correlation and regression relationships for selected pairs
of variables. A system of engineering classification is proposed in which rocks
are classified on the basis of their strength and modulus properties either
obtained directly from laboratory tests, or approximately from index properties
recommended herein. Application of the proposed engineering classification
system to data obtained by others is shown by individual charts for each of
several different rock types. Five charts are presented for estimating the
strength or modulus properties for intact rock from the numerical indices ob-
tained by either t*-: Schmidt hammer, the Shore scleroscope, or the sonic pulse
velocity, all usc?
1 n conjunction with the unit weight of the rock.
DD 1
I JAN
Rm.
64
I rAs
Unclassified
Security Classification
' , I , I I I I I- I I I II I I I I i I
-irsei cfI.d----------i::;e~
LINK
A LINK 0 IN
KEYL j WT ROL Wi' P~oL.W
Rock Mahrvl-icre
Intact Rok Cores
Various Rock Types
Engineering Classification System
Index Propertics
Elastic Properti~es
Physical Properties
Staltic Tests
Dynamic Tests
* Strength
Defornnt ion
Statistical Analysis
INSTRUCTIONS
1. ORIGINATING ACTIVITY: Enter the name and address imposed by security classificatior, using standard statem
of the contractor, tubcontrrctor, grantee, Departmnent of Dc- such as:
fense activity or other organization (corporate author) issuing (I) "Qualifitd requesters may obtain copies of this
the report. report from DDC."
2s. REPORT SECUEITY CLASSIFICATION: Enter the over- (2) "Foreign announcement and dissemination of thit
all security classification of the report. Indicate whether report by DDC is not authorized,
"Restricted Data" is included. Marking is to be in accord-
anct with appropriate security regulations. (3) "U. S. Government agencies may obtain copies o
this report directly from DDC. Other qualified D'
2b. GROUP: Automatic downgrading is specified in DoD Di- users shall request through
rective 5200. 10 and Armed Forces Industrial Manual. Euter
the group number. Also, when applicable, show that optional
markings have been used for Group 3 and Group 4 as author- (4) "1U. S. military agencies may obtain copies of thi
ized. report directly from DDC. Other qualified users
3. REPORT TITLE: Enter the complete report title in all shall request through
capital letters. Titles in all cases should be unclassified.
If a meaninCiul title cannot be selected without classifica-
tion, show title classification in all capitals in parenthesis (5) "All distribution of this report is controlled. Qui
immediately following the title. ified DDC users shall request through
4. DESCRIPTIVE NOTES: If appropriate, enter the type of *_
report, e.gv, interim, progress, summary, annu il, or final. If the report has been furnished to the Office of Terhn
Give the inclusive dates when a specific reporting period is Services, Department of Commerce, for sale to the public,
covered. cate this fact and enter the price, if known.
S. AUTHOR(S): Enter the name(s) of author(s) as shown on IL SUPPLEMENTARY NOTES: Use for additional expli
or in the report. Entet last name, first name, middle initial, tory notes.
If military, show rank and branch of service. The name of
the principal 'Athor is an absolute minimum requirement. 12. SPONSORING MILITARY ACTIVITY: Enter the nami
the departmental project office or laboratory sponsoring (p
6. REPORT DAT'- Enter the date of the report as day, ing for) the research and development. Include address.
month, year; or month, year. If more than one date appears
on the report, use date of publication. 13. ABSTRACT: Enter an abstract giving a brief and fc
summary of the document indicative of the report, even the
7s. TOTAL NUMBER OF PAGES: The total page count it may also appear elsewhere in the body of the technical
should follow normal pagination procedures, ie., enter the .port. If additional space is required, a continuation sheet
number of pages containing information. be attacked
7b. NUMtIER OF REFERENCES: Enier the total number of It is highly desirable that the abstract of classified rt
references cited in the report. be unclassified. Each paragraph of the abstract shall end
So. CON-RACT OR GRANT NUMBER: If appropriate, enter an indication of the military security classification of the
the applicable number of the contract or grant under which formation in the paragraph, represented as (TS). (S), (C), o
the report was written. There is no limitation on the length of the abstract. I
8b, &-, & Sd. PP.t.IECT NUMBER: Enter the approoriate ever, the suggested length is irom 150 to 225 words.
military depart'-ent idcntification, such as project number,
subproject number, system numbers, task number, etc. 14. KEY WORDS: Key words are technically meaningful t
or short phrases that characterize a report and may be use4
9a. ORIGINATOR'S REPORT NUMBER(S): Enter the offi- Index entries for cataloging the report. Key words must be
cial report number by which the document will be identified selected so that no security classification is required. ld4
and controlled by the originating activity. Thls number must fiers. such as equipment model designation. trade name, in
be unique to th-. report. project code name. geographic location, may be used as ke
9b. OTIIER REPORT NUMIBER(S): If the report has been words but will be followed by an indication of technical co
assigned any other rcpert numbers (either by the originator text. The assignment of links, rules, and weights is optio:
or by the spolnsor), also enter thus number(s).
10. AVAILABILITY/LIMITATION NOTICES: Enter any llm-
itetions on further d~ssemanation of the report, other than those
Unclassified
AFSC-HOLLOMAN IFS, NMEX Security Classification

1966 Deere D.U. y Miller R.P. - Eng Class and I Prop For Intact Rock

Uploaded by

Copyright:

Available Formats

1966 Deere D.U. y Miller R.P. - Eng Class and I Prop For Intact Rock

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

1966 Deere D.U. y Miller R.P. - Eng Class and I Prop For Intact Rock

Uploaded by

Copyright:

Available Formats

What is the purpose of the document?

What is the purpose of the document?

What information does it contain?

What information does it contain?

Best Available Copy

TECHNICAL REPORT NO. AFWL-TR-65-116

AIR FORCE WEAPONS LABORATORY

Research andTechnology Division

When U. S. Government drawings, specifications, or other data are used for

Distribution of this document is unlimited.

rUIUTOA, AYAFIAILITY C400

ENGINEERING CLASSIFICATION AID INDEX

TECHNICAL REPORT NO. AETL-TR-65-116

Distribution of this document

This report was preparcd by the Dci;:rtn7nt of Civil ELgincering at the

Inclusive dates of research were 1 February 1964 through I April 1966.

This report has been reviewed and is approved.

ALLEN F. DILL GEORGEC. DBY, JR.

Physical and elastic properties of NX-size rock core from 27

6. ENGINEERING CLASSIFICATION AMD INDEX PROPERTIES FOR IITACT ROCK 136

Table No. Page

2.1 SCALE OF ROCK HARDIN.ESS 9

Figure Vo. Page

3.1 RELATION:SHr? &:T1:7EN1 RLATi,2 VALUES C7 li?-ACT

4.18 TYPICAL CC;:PRESSIVE FAILURES 66

5.14 RELATIC::ZNIP iETWEEN AVZGE VALUES OF SCHU7IDT

5.27 RELATICNSHIP CET',.'EN AVERAGE 0; DILAT.ATIOC.L

6.15 WGENICERING CLASSIFICATICO FOR INTACT ROCK -

6.31 RELATIONSHIP CETWEEN AVERAE VALUES OF 7a.Vp AND

B.0IA STRESS-STRAIN BEHA. IOR AND SCNIC PULSE VELOCITY

a * Y-intercept of rcgrcssion line

Et - tangent modulus of deformation

P2 = weight of saturated specimen

Sy.x - sample standard deviation from regression, or standard error

Ya = unit weight (air-dry)

aa(Y) - yield stress

This page intentionally left blank.

1. NATURE OF TIVE PROILFM

Traditionally, civil and military engineers have been involved with

Many early projects were successfully completed by applying procedures

Physical discontinuities are present in all rock masse. in the form

i) the physical properties of the intact blocks of rock, and

The problem is, therefore, two-fold. In order to develop, ultimately, a

The present investigation is not concerned with the behavior and

i) the physical properties of the constituents, and

The identification and classification of the products of nature con-

i) It must be an index of a material property which is used

The objective of this research was to develop an engineering classi-

In the following section, a -eview of existinq systems of rock classi-

In Section Four, a description of the 28 rock groups studied in this

The test results are discussed and interpreted in Sect ic Five.

In tlkz lest section, the sumciary and conclusions are presented.

EXISTIrWG SYSTEMIS OF RCIC CLA$SIFICATION

One of the primary purposes for rock classification in subsurface engi-

For the engineer, the appropriate classification of both the rock

This section is concerned with a description of some of the systems

The following part of this section deals with geological classifica-

In its broadest geological sense, a rock is any naturally-formed