Sunteți pe pagina 1din 9

PLEASE SCROLL DOWN FOR ARTICLE

This article was downloaded by: [University of Hong Kong]


On: 15 November 2010
Access details: Access Details: [subscription number 905437447]
Publisher Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-
41 Mortimer Street, London W1T 3JH, UK
Philosophical Magazine A
Publication details, including instructions for authors and subscription information:
http://www.informaworld.com/smpp/title~content=t713396797
Mechanism of the Koehler dislocation multiplication process
J. J. Gilman
a
a
Department of Materials Science and Engineering, University of California at Los Angeles, Los
Angeles, California, USA
To cite this Article Gilman, J. J.(1997) 'Mechanism of the Koehler dislocation multiplication process', Philosophical
Magazine A, 76: 2, 329 336
To link to this Article: DOI: 10.1080/01418619708209978
URL: http://dx.doi.org/10.1080/01418619708209978
Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf
This article may be used for research, teaching and private study purposes. Any substantial or
systematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply or
distribution in any form to anyone is expressly forbidden.
The publisher does not give any warranty express or implied or make any representation that the contents
will be complete or accurate or up to date. The accuracy of any instructions, formulae and drug doses
should be independently verified with primary sources. The publisher shall not be liable for any loss,
actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directly
or indirectly in connection with or arising out of the use of this material.
Mechanism of the Koehler dislocation multiplication
process
By J OHN J . GI LMAN
Department of Materials Science and Engineering, University of California at
Los Angeles. Los Angeles, California 90024, USA
[Rrccirad 6 Scptemher. 1996 and ncceprad in revised f orm 18 . !member 1994
ABSTRACT
In the Koehler dislocation multiplication process, segments of screw
dislocations glide off the primary glide plane onto secondary planes, and then
back onto a primary plane parallel to the first plane. This generates trailing
dipoles for small excursions between the two primary planes, or the pinning
points of Frank- read mills for large excursions. The concern of this paper is
the mechanism that causes the lines to cross-glide, and the distribution of dipoles
heights that results. It is proposed that self-excited oscillations of moving
dislocation lines, induced by thermally induced shear-strain fluctuations, cause
the cross-gliding. The properties of this process are described. It is similar to a
two-dimensional random walk, and this determines the dipole height distribution.
The Koehler process is dominant for dislocation multiplication in structural
materials, especially at high strain rates. The resulting dipole height distribution is
an important factor in determining various properties of the cold-worked state.
Some of these are specific heat, the thermal conductivity, fatigue degradation.
strain hardening. Bordoni internal friction and corrosion resistance.
8 1. INTRODUCTION
When a turbulent gas passes over a stretched string it is likely to cause self-
excited oscillations. This is familiar as the singing or galloping of electrical trans-
mission wires in high winds, and the flutter of decorative streamers of athletic events.
Randomly excited vibrations also appear as the flutter of aircraft wings and the
vibrations of a clarinet reed. In engineering it is known as flow-induced vibration.
Various examples have been given in texts on vibrations such as that of Den Hartog
(1956, 1985) or of Pippard (1989). For the case of dislocation lines, the present
author mentioned this phenomenon some years ago (Gilman 1968) but did not
develop the subject. In the meantime, the importance of the Koehler multiple-
cross-glide mechanism for dislocation multiplication has come to be generally recog-
nized. A consequence of the Koehler process is that it produces a high concentration
of dislocation dipoles in cold-worked crystals. Since these dipoles strongly affect
some of the physical properties, the process takes on special importance.
The Koehler (1 952) process is outlined in fig. 1. Its experimental verification has
been given elsewhere (J ohnston and Gilman 1960). Its importance has two aspects.
First, starting from only a single dislocation half-loop, a nearly arbitrary length of
dislocation line can be continuously generated. Thus, from a loop that is a few
microns in lengths in a cube of 1 cm3 volume, a continuous length (with a line
density of 1013 cm-) can be generated that equals the distance from the Earth to
the Sun. Second, the process generates a plethora of dislocation dipoles which trail
0141 8610!97 512 00 (( 1997 Taylor & Francis Ltd.
D
o
w
n
l
o
a
d
e
d

B
y
:

[
U
n
i
v
e
r
s
i
t
y

o
f

H
o
n
g

K
o
n
g
]

A
t
:

0
9
:
5
9

1
5

N
o
v
e
m
b
e
r

2
0
1
0
330 J . J . Gilman
Fig. 1
Schematic diagrams of the Koehler multiple-cross-glide process. (a) A segment Py of a screw
dislocation with Burgers vector b and velocity vector v cross-glides from the lower
primary glide plane up the (filled) secondary glide plane and then cross-glides again
onto the upper primary glide plane. Note that the jogs ap and y6 lie perpendicular to
b; so they are edge dislocations and are constrained to move only on their current
secondary glide plane. (b) As the dislocation lines of (a) continue to move, the segment
on the upper plane becomes a semicircle and then a heart-shaped configuration until
the lobes of the heart meet, and parts of the lines annihilate one another. At the same
time, on the bottom plane the lines swing forwards as well as laterally until a portion
of them becomes annihilated. Note that the upper and lower lines must be able to pass
over one another in order for the indicated motions to occur. This requires that the
separation H of the planes be greater than a value that depends on the applied stress.
(c) Further motion of the dislocation lines of (b). The configuration at the centre has
been restored to its form in (a). The original line on the lower glide plane is moving off
to the right. There is a new loop on the upper glide plane with one segment moving off
to the left, and the other to the right. (d) Continuation of the motion of the dislocation
lines in (a) when the separation h of the primary glide planes is too small for the lines
to pass over one another under the given applied stress. Two edge dislocation dipoles
are left behind as the original line and the segment that underwent double cross-glide
move forwards. Note that in a homogeneous stress field there is no net force on either
of the dipoles; so they do not move laterally.
D
o
w
n
l
o
a
d
e
d

B
y
:

[
U
n
i
v
e
r
s
i
t
y

o
f

H
o
n
g

K
o
n
g
]

A
t
:

0
9
:
5
9

1
5

N
o
v
e
m
b
e
r

2
0
1
0
Koehler dislocation mulliplication process 33 1
behind moving screw dislocations. These dipoles cause strain hardening, thermal
resistance, fatigue, strength reduction, acoustic attenuation and other property
changes (Gilman 1964). With increasing time and plastic strain, the dipoles tend
to aggregate into cell walls (Kuhlman-Wilsdorf 1985).
The concern of this paper is the initiating mechanism for multiple cross-glide. I t
is proposed that it results from self-excited oscillations (flutter) that are induced by
collisions of moving screw dislocations with thermal shear-strain fluctuations.
Previously proposed mechanisms have involved the Lorentz force on a moving
dislocation (Nabarro 1987) (since shown not to exist), and internal stress sources
such as precipitates, other dislocations (Li 1961) and surfaces (Gilman 1961).
However, the process appears to be intrinsic which eliminates the second of these
possibilities. The surface effect plays a secondary role.
From time to time it has been argued that cross-gliding results from thermal
activation. This is inconsistent, however, with the fact that the rate of multiplication
(loops per centimetre) increases with increasing dislocation velocity (fig. 2) (J ohnston
and Gilman 1960). If it resulted from thermal activation, it would be expected to
occur less frequently at higher velocities. Note that the average velocity often
increases many orders of magnitude when the applied stress increases by one
order of magnitude. Thus velocity, and not stress, appears to be the independent
variable.
Two other factors do not seem to be consistent with thermal activation. One is
that the dislocation density within a widening glide band increases with decreasing
temperature. The other is that dislocation multiplication occurs in LiF (at high
dislocation velocities) at 78 K which is well below the Debye temperature.
A condition for the generation of self-excited oscillations is that the damping in
the system be small. This condition has been verified in crystals of pure simple metals
and pure salts, by means of direct measurements of dislocation velocities and by
Fig. 2
.
.
.
.
.
.
*. . i
.
.
I I I I I I I I I ..
0.2 0.4 0.6 0.8 1 .o
0
0
Effect of stress (velocity) on the rate of dislocation multiplication in a lithium fluoride crystal.
The logarithm of the multiplication rate is given as a function of the reciprocal applied
shear stress. The full circles are experimental measurements [data from J ohnston and
Gilman (I 960)).
D
o
w
n
l
o
a
d
e
d

B
y
:

[
U
n
i
v
e
r
s
i
t
y

o
f

H
o
n
g

K
o
n
g
]

A
t
:

0
9
:
5
9

1
5

N
o
v
e
m
b
e
r

2
0
1
0
332 J . J . Gilman
ultrasonic attenuation measurements. In less pure crystals it is true for local regions
between impurities and defects. The 'Peierls force' on the primary glide planes in
these crystals is vanishingly small (Nadgornyi 1987). Therefore, the damping coeffi-
cients for dislocation motion are small, of the order of 3 x lop4 dyn cm-* which
gives drag forces on dislocations that are small compared with driving forces.
0 2. DISLOCATIONS AS SELF-EXCITED OSCILLATORS
We start with the equation of motion of a dislocation line and consider the effect
of impulsive forces on it. To minimize complications, the material that contains the
dislocation line is taken to be cubic. The line must be screw type if it is to multiply
cross-glide. Therefore, only three parameters are needed to characterize it: the
Burgers displacement b, the shear modulus G and the mass density p. The effective
mass per unit length will be taken to be rn = pb2, and the viscosity coefficient of the
medium 7. The line will support vibrational modes of length L up to L = A where A
represents a segment lying between points that are pinned by impurities, dipoles, or
other crystal defects. The tension of the line is T which is approximately equal to
Gh2/2rc.
The line tension T is not isotropic because the shear stiffness is a second-order
tensor. Thus the tension (change in energy with curvature) will be different for a
curved line lying on the primary as against the secondary glide plane. We assume
simple one-axis curvatures and, to keep the discussion simple, consider only two
tensions: one, T, on the primary plane, and the other, T,, on the secondary plane.
These correspond to two shear moduli G, and G, ( b being the same for both planes).
As a dislocation line moves, it continually collides with randomly distributed
shear-strain fluctuations which exert forces on it that are random in time and space.
These forces are distributed isotropically for a stationary dislocation but have resul-
tant directions opposed to the velocity vector of a moving line. Their magnitudes
increase as the velocity increases. Note that this process is different from the thermal
excitation. The effects of the latter would decrease with increasing velocity because
there would be less time for an excitation as the velocity increased. In contrast, the
buffeting forces increase with the velocity because the changes in quasimomentum
become larger, and the collision frequency increases.
3 3. THE FLUCTUATIONS
A standard result of fluctuation theory is that mean square shear-strain fluctua-
tions (y is the shear strain) are given by
where k is the Boltzmann constant, T is the temperature and V , is the volume of
interest (Parrinello and Rahman 1982, Weiner 1983). Thus the fluctuations are small
unless V0 is small (atomic dimensions), and they increase in proportion to the square
root of the temperature. The buffeting forces can be substantial. Taking V, = b'L,
and L = lob, G = 5 x 10" dyn cmP2 and b = 2.5 A, the stress fluctuations (strain
multiplied by modulus) given by eqn. (1) have magnitudes of about 1 GPa at
room temperature. For smaller values of L they are larger, up to the intrinsic
shear strength of the material (i.e. about G/10). The magnitudes of the fluctuating
forces can also be estimated from the rate of change in momentum when a disloca-
tion line collides with a fluctuation. The collision cross-section is approximately
D
o
w
n
l
o
a
d
e
d

B
y
:

[
U
n
i
v
e
r
s
i
t
y

o
f

H
o
n
g

K
o
n
g
]

A
t
:

0
9
:
5
9

1
5

N
o
v
e
m
b
e
r

2
0
1
0
Koehlrr dislocation multiplication procexy 333
equal to b (Hirth and Lothe 1968). Equation (1) gives the stress fluctuations
AT = G Ay; so the force fluctuation per unit length is
.f z b AT. ( 2 )
Moving at a velocity vd, a dislocation line of length L and collision cross-section
h sweeps out a volume Lbvd per second. The phonon occupation numbers np for the
atoms in this volume are
- I
np = [exp (&) - 11
( 3 )
where cp = (h/2n)w, is the phonon energy with h the Planck constant and wq the
frequency of a phonon with wavenumber q. At temperatures above the Debye tem-
perature 0, the phonon energy becomes kO, and the bracketed term in eqn. (3)
becomes approximately equal to its argument (Walton 1983); so the number of
phonons per unit volume becomes n = (9/2h3)( T/ O) . Then, the number of collisions
with the segment of length L is
L v ~ T
n, = 4.5---,
b 8
(4)
confirming that the collision rate increases with increasing temperature and disloca-
tion velocity. For L = cm, the collision rate is 4 x l o1* s- when the dislocation
velocity is 3 x lo4 cm s-, Thus the collision frequency is approximately equal to the
local atomic vibration frequency.
3 4. RESPONSE OF A DISLOCATION LI NE TO RANDOM EXCITATIONS
Using the given definitions, the equation of motion for the dislocation segment is
a2y av a2y
ax- at at 2-
-Ti -5 + v2 + m- - bT(a, t)S(x - a) ,
where j is the displacement of the line (0 <.c <L) , t is time, br(a. t ) is the driving
stress function and 6(.u - a) is the delta function which places the driving stress
function at .i = u, and i =p or s. The terms on the left-hand side are the curvature
force, the viscous drag force and the intertial force (all per unit length) (Newland
1984). This is the standard starting point for the theory of the random vibrations of a
string.
There are two resonant frequencies q = (n/L)(T,/m)2 with i = p, s. In some
cases. they will be equal; so the dislocation line will tend to walk randomly from
one plane to the other. This will tend to maximize the rate of dislocation multi-
plication, leading to high ductility (Gilman 1995). Since the dipoles left behind by the
Koehler process will strain harden the initial regions of the glide planes, dislocation
motion will be inhibited there. However, the motion will remain easy on both sides
of these central regions; so glide bands will spread sidewise into virgin regions until
the bands begin to impinge on one another (J ohnston and Gilman 1960).
The frequency spectrum of the fluctuations is broad, ranging from zero to the
atomic cut-off frequency which can be estimated from the Debye frequency which is
equal to kO/ h (if 0 = 200 K, this is 4.2 x 10 s-). Thus line segment oscillators of
various lengths can be excited by the motion.
D
o
w
n
l
o
a
d
e
d

B
y
:

[
U
n
i
v
e
r
s
i
t
y

o
f

H
o
n
g

K
o
n
g
]

A
t
:

0
9
:
5
9

1
5

N
o
v
e
m
b
e
r

2
0
1
0
334 J . J . Gilman
Note that an interesting but speculative possibility is that the fluctuations
become correlated by the strain field of the dislocation enough to form vortices.
Then a fast-moving dislocation might leave Karman vortex trails (Tietjens 1990)
in its wake (for more details see Blevins (1990)). This speculative possibility would
result in semiregular buffeting in addition to the more localized random buffeting.
Taking the size of the dislocation line to be 2b, according to hydrodynamic obser-
vations (Tietjens 1990), the spacing of vortices of the same sign would be about 8b;
so impulses in opposite directions would arrive about every 4b, or at a frequency of
about vd/4b.
Equation (5) i s difficult to solve in general because impacts on the line are
random in both x and t. Newland (1984) gave a particularly good discussion of
general solutions. In the dislocation case, a particular problem is that the nature
of Ti is not clear for dynamic conditions where time retardation effects are impor-
tant. The dependence of Ti on curvature is also not clear for dynamic conditions.
Therefore, to simplify the problem, the line tension term is linearized, becoming an
elastic restoring force proportional to y , and given by By with ,d = Gb2/4xL. This is
equivalent to associating the effective line tension with only the lines core energy.
For further simplification, let the random impacts along the length be replaced by
effective values F( t ) at the midpoint of the line at random times T and which last for
short times dT. Then eqn. ( 5 ) becomes
my + rli, + Py = F( t ) , ( 6 )
where the dots mean differentiation with respect to time, and F( t ) is a drivin force.
Dividing through by m, recognizing that the resonant frequency is w = (p/m)/, and
setting cy = 7/2mw, the solution of this can be written as the Dyhamel integral (Smith
1988)
y ( t ) = -1 F( t ) sin [w( t - T ) ] exp [-aw(t - r) ] dr.
Since 77 (and therefore a) is small, the exponential damping term can be neglected.
Furthermore, since the moving line will be subjected to constant average buffeting
forces, F( t ) = Fo = constant. Then, eqn. (7) becomes
(7)
1
mw 0
y ( t ) = 5s sin [w(t - T) ] dr.
m w o
Integrating this and substituting for the resonant frequency give
FOW
y( t ) =-[I - cos (wt)] = yo[l - cos (wt)]
P
(9)
where yo is the average displacement, and y ( t ) oscillates around the average with an
amplitude equal to 2y0.
8 5. THE DISTRIBUTION OF DIPOLE HEIGHTS
Random excitation of the vibrational modes of a moving screw dislocation will
excite all the modes of X = 2L to X = 2b where X is the mode wavelength. We are
interested in the transverse modes whose corresponding energies range from hv,/2L
to hv,/2b where h is the Planck constant and v, = ( G/ P) ~ is the transverse sound
speed. If the maximum mode length to be considered is lo3 A, the mode energy
varies from about 6 x lop6 to about 6 x lop2 eV, a reasonable range.
D
o
w
n
l
o
a
d
e
d

B
y
:

[
U
n
i
v
e
r
s
i
t
y

o
f

H
o
n
g

K
o
n
g
]

A
t
:

0
9
:
5
9

1
5

N
o
v
e
m
b
e
r

2
0
1
0
Koehler dislocation multiplication process 335
The overall Koehler process occurs in two stages: firstly, glide off the primary
plane; secondly glide off the secondary plane back onto the primary plane. For each
position held by the screw dislocation, its next move is either forward by b on the
primary plane, or forward obliquely onto the secondary glide plane. The inverse
probabilities are small. Let the probability of the former process be p , and the
probability of the latter process be s. In structural materials, p will be somewhat
greater than s except for isotropic materials where s = p = i. For strongly layered
materials, p >>s which completely suppresses the Koehler mechanism.
For the simplest (isotropic) case where p = s = i, and the shear stresses on the
primary and secondary plane (as well as any drag stresses) are equal, the probability
P, of a multiple-cross-glide event in which the excursion on the secondary plane is
one atomic height is the product of the probability of the p + s step, and the prob-
ability for the s + p step. That is, for a one-step excursion, PI = ( $ ) 2 = i ; for a two-
step excursion, P1 = (&)3 = Q; and so on to an n-step excursion, with P,, = ( 4 ) ' ' ' .
Thus the most probable dipole height is one atomic distance, and dipoles with large
heights are improbable. For example, Pl o M 0-001, or 250 times less probable than
P I . This is consistent with the fact that only a few large dipoles are seen directly,
whereas many dipoles (mostly not large) are inferred from changes in thermal con-
ductivity (ordinary electron microscopy does not see dipoles with small heights).
It should be remembered that, in specimens that are undergoing plastic flow, the
dipole population is in a state of considerable flux. Not only are dipoles being
created by the Koehler process, but also they are being dissociated by the local
shear stresses. being re-formed through trapping interactions, and being annihilated
by collisions.
Let a be the glide plane spacing. The maximum dipole spacing h* is determined
by the value of the height h at which the applied stress 7 decomposes dipoles. Then
the range of heights is (with C the shear modulus, v the Poisson ratio and b the
Burgers displacement)
Gb
a < h <
~ T C T ( 1 - V)
There seems to be no direct evidence of the distribution of heights between the
two limits. Electron microscopy detects monopoles through their long-range strain
fields, but the strain fields of dipoles decrease much more rapidly with increasing
distance than do those of monopoles. Thus, if h is small, the strains at long distances
are small. At short distances, for small h, the strains are also small because of mutual
cancellations between the two dislocations of a dipole.
Indirect evidence that there is a large population of small heights in plastically
deformed crystals is given by thermal conductivity data (Berman 1976), by energy
storage data (Bever, Holt and Titchener 1973) and by Bordoni internal friction
(Gilman 1996). Measured values of thermal conductivity are orders of magnitude
smaller than the values given by theories of phonon scattering from directly
observable dislocation populations (Moss 1965).
During plastic deformation, both conversion of plastic work into heat, and the
storage of a fraction of the work as internal energy require the presence of large
concentrations of dipoles with small heights. In pure simple metals, the drag on
moving dislocations is small. Therefore the direct generation of heat is too slow to
account for the heating that is observed (Gilman 1993). However, when two mono-
poles condense into a dipole, a large reduction in strain energy occurs, as well as a
D
o
w
n
l
o
a
d
e
d

B
y
:

[
U
n
i
v
e
r
s
i
t
y

o
f

H
o
n
g

K
o
n
g
]

A
t
:

0
9
:
5
9

1
5

N
o
v
e
m
b
e
r

2
0
1
0
336 Koehler dislocation multiplicution process
large reduction in kinetic energy in most cases. The dipole that results from the
condensation is a local thermal oscillator that helps to convert strain energy into
heat (especially since the damping is small). This path for heat generation is
especially effective for small dipole heights.
Under some conditions, especially low-temperature deformation, large fractions
of the plastic work are stored as internal energy (Erdmann and J ahoda 1964). In
order for this to be consistent with the magnitude of the flow stress, the dipole
distribution needs to be skewed in direction of small heights.
The power law distribution that results from the random-walk process discussed
earlier can be approximated by an exponential distribution. Let n(h) = n be the
concentration of dipoles at height h, n, be the concentration for h equal to the
glide spacing and n, be the concentration for the maximum height given by eqn.
(10). Then
so, when h is small, n(h) -+n, since the exponential goes to unity and, when h is
large, n(h) --f n,. Therefore, if two values of the concentration are known, the entire
distribution is determined.
n(h) = n, + (n, - n,) exp (-ph); (11)
ACKNOWLEDGEMENTS
The author is grateful to the Sutter Foundation for supporting this work.
REFERENCES
BERMAN, R., 1976, Thermal Conduction in Solids (Oxford: Clarendon).
BEVER, M. B., HOLT, D. L. and TITCHENER, A. L., 1973, Progr. Mater. Sci., 17, 1 .
BLEVINS, R. D., 1990, Flow-induced Vibrations, second edition (New York: Van Nostrand
DEN HARTOG, J . P., 1956, Mechanical Vibrations, fourth edition (New York: McGraw-Hill);
ERDMANN, J . C. and J AHODA, J . A. , 1964, Appl. Phys. Lett., 4, 204.
GILMAN, J . J ., 1961, Phil. Mag., 6, 159; 1964, Discuss. Faraday Soc., 38, 123; 1968, Dislocation
Dynamics, edited by A. Rosenfield, G. Hahn, A. Bement and R. J affee (New York:
McGraw-Hill), p. 3; 1993, Mech. Muter., 17, 83; 1995, Fracture: A Topical
Encyclopedia of Current Knowledge Dedicated to Alan Arnold GrifJi;th, edited by G.
Cherepanov (Melbourne, Florida: Krieger), p. 3 15; 1996, The Johannes Weertman
Symposium, edited by S. N. G. Chu et ul. (Warrendale, Pennsylvania: Minerals, Metals
& Materials Society), p. 41 1.
HIRTH, J ., and LOTHE, J ., 1968, Theory of Dislocations (New York: McGraw-Hill), p. 193 ff.
J OHNSTON, W. G. and GILMAN, J . J ., 1960, J. appl. Phys., 31, 632.
KOEHLER, J . S., 1952, Phys. Rev., 86, 52.
KUHLMAN-WILSDORF, D., 1985, TMS Metall. Trans. 16A, 2091.
Lr, J . C. M., 1961, J. appl. Phys., 32, 593.
Moss, M., 1965, J. appl. Phys., 36, 3308.
NABARRO, F. R. N., 1987, Theory of Crystal Dislocations (New York: Dover Publications)
NADGORNYI, E., 1987, Prog. Mater Sci., 31, 1.
NEWLAND. D. E., 1984, An Introduction to Random Vibrations and Spectral Analysis, second
PARRINELLO, M. and RAHMAN, A., 1982, J. chem. Phys., 76, 2662.
PIPPARD, A. B., 1989, The Physics of Vibration, omnibus edition (Cambridge University Press).
SMITH, J . W., 1988, Vibrations ofStructures (London: Chapman & Hall), chapter 2.
TIETJ ENS, 0. G. , 1957, Applied Hydro- and Aeromechanics, translated by J . P. Den Hartog
WALTON, A. J ., 1983, Three Phases of Matter (Oxford: Clarendon) p. 365.
WEINER, J . H., 1983, Statistical Mechanics of Elasticity (New York: Wiley) chapter 3.
Reinhold).
1985, Zbid. (New York: Dover Publications).
p. 496 ff.
edition (New York: Longman), chapter 16.
(New York: Dover Publications) p. 130.
D
o
w
n
l
o
a
d
e
d

B
y
:

[
U
n
i
v
e
r
s
i
t
y

o
f

H
o
n
g

K
o
n
g
]

A
t
:

0
9
:
5
9

1
5

N
o
v
e
m
b
e
r

2
0
1
0

S-ar putea să vă placă și