J. Mech.Phys.

Solids,1968,
Vol. 16,pp. 373to 394. Pergamon
Press. Printedin GreatBritain.

EXPERIMENTAL STUDY OF THE SUBSEQUENT

YIELD SURFACE BY USING CROSS-SHAPED
SPECIMENS

By E. SHIRATORI and K. IKEGAMI

Tokyo Instituteof Technology, Tokyo, Japan

(Received 81st July 1908)

SUMMARY

THE SUBSEQUENT yield surfaces after prestraining along various loading paths or by cold rolling
were determined with cross-shaped specimens of brass plate. The shape of the subsequent yield
surface and the correlation between the change of the stress increments and that of the plastic
strain increments for a zigzag loading path were discussed with reference to the shape of stress-
strain diagram. The experimental values of the yield surfaces of the cold-rolled brass plate were
compared with the calculated values by Don’s theory.

1. INTRoD~OTI~N

IN THE existing theory of plasticity, the material is mostly assumed to be isotropic.

But the plastic behaviour depends on the strain history, as seen in the Bauschinger
effect. More precise theory including the strain history may be necessary in solving
some practical problems such as formation of ‘ ears ’ in deep drawing. To find
such a theory, the effect of the prestrain on the subsequent yield surfaces must be
studied as the hrst step.
Experimental results on the subsequent yield surfaces have been reported by
many investigators. NAGHDI, ROWLEY and BEADLE (1955), NAGHDI, ESSENBURG
and KOFF (1958), HARVEY and MCCOMB (1960),PHILLIPS and GRAY (1961), IVEY
(1961), BERTSCH and FINDLEY (1962), PAUL, CHEN and LEE (1962), MAIR and
PUGH (1963) used thin-walled tubular specimens under torsion and axial load.
Hu and BRATT (1958), PARKER and BASSETT (1964) made the experiment by applying
internal pressure and tension or torsion to a thin-walled tubular specimen,
THEOCARIS and HAZELL (1965)obtained the biaxial stress condition by bending
plates. However, in the method of using tubular specimens,

(1)testing a flat plate under biaxial stress condition is not possible,

(2) obtaining the specimen with a desired plastic anisotropy is difficult,
(3) holding the stress ratio constant is difficult on account of the partial expan-
sion or contraction of the specimen,
(4) stress state is not exactly two-dimensional because of the radial stress produced
by the internal pressure.
To diminish these defects, a new biaxial tensile testing machine with the
cross-shaped flat specimen is proposed by the authors (1967). In this paper the
373
374 E. SRIRATORIand K. IKEGAMI
experimental results by using this machine are reported in regard to the subsequent
yield surfaces after prestraining along various loading paths or by cold rolling.

2. TESTINGMACRINEAND SPECIMENS

Figure 1 shows the biaxial tensile testing machine. With four jacks in this machine
the desired biaxial stress state can be obtained in the central part of the cross-
shaped flat specimen. The capacity of the machine is 5 ton and the stroke 100 mm.
The load is produced by introducing the pressurized-oil from one pump (1) into
two jacks (2) which face each other. This load is transmitted to the crosshead
(3) and the chuck (5) for the specimen (7) through the load cell (4). To keep the
centre of the specimen at that of the machine, the equi-displacement mechanism (6)
of pantographtype is attached to the crossheads, and a slip between the specimen
and the chuck is prevented by the binding bolts.
Figure 2 shows a typical example of the biaxial tensile specimen. The specimen
consists of one cross-shaped body and eight plates to reinforce its four arms.

FIG.2. Biaxial tensile specimen.

When the authors tried to establish this testing method, the following points
came into question:

(i) method of keeping the centre of the specimen continually at that of the
machine during an experiment,
 Experimental study of the subsequent yield surface by using cross-shaped specimens 375

(ii) dimension of the specimen,

(iii) region of homogenous plastic deformation in the specimen,
(iv) method of measuring the stress.
Problem (i) was solved by using the equi-displacement mechanism of pantograph-
type as shown in Fig. 8. The problems (ii)- were solved experimentally. The
authors (1967) confumed the existence of the region of homogeneous plastic de-
formation in the cross-shaped specimen. The dimensions of the specimen shown
in Fig. 2 were so determined that the region of homogeneous plastic deformation
in the specimen became as large as possible by changing the shape of the arms and
the reinforcing plates of the specimen.

FIG. 8. Schematic diagram of the equi-displacement mechanism.

The stress in the region of homogenous plastic deformation was able to be

calculated by dividing the load by the equivalent sectional area at the section IL
of the specimen as shown in Fig. 4.

000

Fro. 4. Equivalent cross-section of the specimen.

Besides, the strains were measured by the rosette type strain gauges (YRS-5
of T.M.L. Co.) attached to the upper and lower surfaces in the centre of the
specimen.

3. EXPERIMENTAL PROCEDURE
3.1 Loading path for the prestrain
As the basic loading paths for the prestram, the following three types may be
considered for the case of two-dimensional tensile stress space.
376 E. SHIBATORIand K. IKEGAMI

The preloading type I is the loading along the path of constant stress like OS
in Fig. 5. Type II is the loading along the path of variable stress ratios like
OSsS” in Fig. 5. Type III is the loading which is composed of the reloading along
the different path from that for the initial prestrain like 0 -+ PI + Pl’l in Fig. 5.
Resides the experiments of these three preloading types, the yield condition of cold
rolled plate is studied, too.

1 Initial yield
I , surface

iaoding type 1

--- - LoodIng type li

--- Lcadir,g i),w 4

FIG. 5. Loading paths for the prestrain.

3.2 De$nition of the yield point

In the experiments of the yield surface, the definition of the yield point may have
considerable effect on the results obtained.
The yield point is generally defined by one of the following three points in
the stress-strain diagram:
(i) the point of the proportional limit,
(ii) the point by the back extrapolation,
(iii) the point by the proof strain.
As there is obscurity in the yield point determined by the proportional limit or
by the back extrapolation, the yield point is defined by the proof strain in this
experiment. In defining the yield point with this method, a precise consideration
must be given on the amount of the proof strain.
For the experiment on the subsequent yield surfaces, if only one point on the
yield surface is determined with one specimen, many specimens of the same dimen-
sions and material will be necessary. It would be better to obtain many points on
the yield surface with one specimen. Hence the authors determined the subsequent
yield surface as follows:
Referring to Fig. 7,
(i) A specimen was prestressed to a given point Sr along the loading path
UJUz = 1.
 Experimental study of the subsequent yield surface by using cross-shaped specimens 377

(ii) Then the specimen was unloaded to any point Yt (i = 1, 2, . . . )

along the same path as (i).
(iii) After that the specimen was reloaded in the direction X or Y from Yt
to the yield point.
With these procedures, a point on the first subsequent yield surface was determined.
By varying Yg in (ii) and the reloading direction in (iii) at random, the subsequent
yield surfaces was determined. But, in this method, the hardening which occurred
in the process of determining the yield point must be omitted in comparison with
that in the prestrain. For this reason the amount of the proof strain for the definition
of the yield point must be as small as possible.
In this experiment the yield point is defined as the stress corresponding to the
definite value of the plastic strain such as O-01 or 0.02 per cent, in either one of
the gauges in the principal directions X and Y.
The yield surface was found to be substantially independent of the reloading
path to determine a yield point, except for the prestressed point in the preloading
type II, as described later.

4. EXPERIMENTALRESULTS
4.1Shape of the subsequent yield surfbee
In this experiment the specimens were made of brass plate (BsPS - &H according
to Japanese standards), 1 mm thick, and the yield point was defined as the stress
corresponding to the value of the plastic strain of 0.02 per cent. The specimens
were prestressed along the paths of three preloading types previously described,
and the change of the subsequent yield surface due to the amount and direction of
preloading was studied systematically.
In the preloading type 1, preloading paths were taken along OP (uy/u2 = 0),
OQ (uy/az = 4) and OS (gy/uz = 1) in Fig. 5. In the preloading type III, those
were OQ,Q” and OS&‘s. in the same figure, where Q. and S, were the intersections
of the initial yield surface and the paths of the constant stress ratio of uU/uz = 1,
respectively. The direction of Q,,Q” was parallel to u~/u, = co, and that of S&3
was perpendicular to us/u2 = 1. In the preloading type III, two second preloading
paths along OPw (cg/uz = co) and OPl’l ( a, /CT% = 1) were taken after the first
uniaxial prestrain along the path of ~,/a, = 0 in Fig. 5. Some experimental
results are shown in Figs. 6-12. In these figures the prestressed points are indicated
as Pe, Qe, St, Q”g, S’a, Pmi and P$‘r (i = 1, 2, 3 . . .). The yield surfaces after
prestraining are named the lst, 2nd, Srd, and 4th subsequent yield surfaces accord-
ing to the amount of the prestrain.
The initial yield surface approximately coincided with the Mises condition as
shown in Fig. 6.
In Fig. 7 and Fig. 8 of the preloading type I, and in Fig. 12 of the preloading
type III, a corner* was observed on the subsequent yield surface. But, even in
these preloading types, the corner could not be found clearly for the case of the
uniaxial prestrain as shown in Fig. 6 of the preloading type I and in Fig. 11 of the
preloading type III. In the preloading type II, the existence of the corner was
not clear as seen in Fig. 9 and Fig. 10. In these figures, only the prestressed points
*Comer
means
the convexly distorted part of the yield surface : it is used in a wider ~eme then in the Blip theory.
378 E. SHIRATORIand K. IFCEGAMI

Term& stress in the direction X at yieid.[tr~~ VI

Initial yield stress in the direction X/O%)

a~. 6. InitiaI and subsequent yield surfaces for the p~IoadW path OP.

FIG. 7. Subsequent yield surface for the preloading path OS.

Experimental study of the subsequent yield surface by using cross-shaped specimens 37’9

,.‘12
-

-
in X
in Y
-

FIU. 8. Subsequent yield surfaces for the preloading path OQ.

DA tension in Y

fO;)v
@‘OS

Fro. 9, Subsequent yield surfaces for the preloading path 08.6”.

 E. and K. IKEGAMX
SHIRATORI

BIG. 10. Subsequent yield surfaces for the preloading path OS&“.

FIG. 11. Subsequent yield surfaces for the preloading path OP.
 Experimental study of the subsequent yield surface by using cross-shaped specimens 381

If

FIG. 12. Subsequent yield surfaces for the preloading path OPlk

seemed to protrude from their subsequent yield surfaces and this tendency became
clearer according to the increase of the amount of the prestrain like Q”, Qa” and
Q”s in Fig. 9, or like Sr” and Se” in Fig. 10.
Hence the existence of the corner may be considered to depend on the direction
of the path for the prestrain.
In relation to the results shown in Fig. 9 and Fig. 10, the following problems
come into question: namely, which point should be chosen as the yield point in
the close vicinity of the prestressed point, the prestressed point itself or the yield
point newly determined by reloading. Moreover, the latter point may be affected
by the reloading path.
These problems were experimentally studied in the following procedures.
The same path as that in prestraining may be considered as the most basic

Fro. la. Stress-strain diagmm in unloading snd reloading.

382 E. SHIRATORI and K. IKECAMI

one to check the first problem. In adopting this path, the difference between the
prestressed and newly determined yield point is considered to depend on the
stress-strain diagram in unloading and that in reloading. Figure 13 (a) and (b) show a
typical results of their stress-strain diagrams along the path OQOQ and OQoQ”
in Fig. 5. As shown in Fig. 13 (a), the stress-strain diagram in unloading and that
in reloading coincide with each other, and the newly determined yield point K’
and the prestressed point K have approximately the same value in the preloading
type I (and in the preloading type III, too), while there is a little difference between
these two stress-strain diagrams, namely, the newly determined yield point J’
is a little lower than the prestressed point J in the preloading type II.
The second problem is studied by comparing the yield surface or point determined
by two different reloading paths, such as the path of constant stress ratio and that
of combined uniaxial tension. Typical experiments results are shown in Fig. 10
and Fig. 14. In these figures the preloading paths are OS&?’ and OQ in Fig. 5,
respectively. In the preloading type I the two yield surfaces determined by the
paths of constant stress ratio and combined uniaxial tension approximately
coincide with each other, and the prestressed point also lies on these surfaces as
shown in Fig. 14. The same results was obtained in the preloading type III. But,
in the preloading type II, the prestressed point does not coincide with the yield
point newly determined by the different path from that prestraining as shown in
Fig. 10. In this figure the point S’s” indicates the prestressed point along the path
of OS# in Fig. 5. The points &r” and Szsn are the yield points determined with
the paths of combined uniaxial tension as shown with the arrow in this figure. The
point 6’s~” is the yield point obtained from the stress-strain diagram in reloading
along the same path as that in prestraining. It is considered from these results
that the protrusion of the prestressed point does not show the existence of the
corner in this preloading type.
Consequently, in the preloading type I and type III, the corner is found in the
subsequent yield surface, though it is not clear in the case of the uniaxial prestrain.
On the contrary, in the preloading type II, all points, except for the prestressed
point, lie on a smooth yield surface and only the prestressed point protrudes from this
yield surface.

FIG. 14. Subsequent yield points by radial and combined uniaxial loading.
 Experimental study of the subsequent yield surface by using cross-shaped specimens 383
&2 Plastic flow under stepwise increments of biaxial tension
As described by DRUCKER and STOCKTON(1953), the change of the direction
of plastic strain increments vector corresponding to the zigzag loading will be
small in the case of a rounded yield corner, while the change will not be small
at a sharp corner.
To investigate, with this principle, whether the corner of the subsequent yield
surface was pointed or rounded, plastic strain increments were measured for some
zigzag loading paths.
The stress increments was chosen in the range of the value 0.1 kg/mm2 to
0.4 kg/mmz, and the strain was measured in about one minute after applying
each stress increment in order to exclude the effect of creep on the strain value.
The value of the plastic strain was calculated by subtracting that of the elastic
strain obtained with Hooke’s law from that of the measured strain. Some experi-
mental results are shown in Figs. 15-26.
In the preloading type I, for example, the plastic strain path from the point
10 to the point 20 in Fig. 16 was obtained by the zigzag loading from the point
10 to the point 20 in Fig. 15. This plastic strain path approximately formed a
smooth line. Hence the change of the direction of the plastic strain increment vector
corresponding to that of the stress increment vector was not observed clearly.
The same results were obtained in the other part of the plastic strain path of Fig.
16, as well as in Fig. 18 of the preloading type I and in Fig. 20 of the preloading type
II. On the contrary, in the preloading type III, the zigzag loading in Fig. 25 gave
the zigzag plastic strain path as shown in Fig. 26. This showed partly the change
of the direction of the plastic strain increment vector corresponding to that of
the stress increment vector. The change was also observed in Fig. 22. But, even
in the preloading type III, such a change was not found in Fig. 24.
From these results, it was considered that the yield surface had a rounded
corner when the specimen was loaded from the virgin state, but it had a sharp
corner occasionally when the specimen, which had been initially preloaded and

10 11 12 13 14

0; kglmm”

FIG 15. Zigzag loading path in the direction OS.

334 E. SEIRATORI and K. IKEGAMI

--__--- ----

FIG. 16. Plastic strain path for the zigzag loading path OS.

FIG. 17. Zigzag losding path in the direction OQ.

Experimental study of the subsequent yield surface by using cross-shaped specimens 385

FIG. 18. Plastic strain path for the zigzag path OQ.

Fxa. 19. Zigzag loading path in the direction for the zigzag:loading path &,S”.

Fig. 20. Plastic strain path for the zigzag loading path &Y.
386 E. and K.
L%IIRATORI IKEGUll

1
b
L

I I
8 9
0-z kg/mm2

FIG. 21. Zigzag loading path in the direction OPk

Experimental study of the subsequent yield mrface by using cross-shaped specimens 387

-7.
6

-7.8

-a.4

-8.6

-&a

-9.0
17.8

Fro. 22. Plastic strain path for the zigzag loading path OpIk
 E. SHIRATORI and K. IKEGAMI

FIG. 23. Zigzag loading path in the direction OPlil.

Xl0
-10.7

-10.8
‘L?

-10.9

-11.0

-11.1

-11.2
18.9 19.0 19.1 19.2 19.3

c:

Fro. 24. Plastic strain path for the zigzag loading path OPl/l.
Experimental study of the subsequent yield surface by usiug cross-shaped specimens 389

2
I,
I
17

FIG. 25. Zigzag loading path in the dire&ion OPlk

19.1 19.2 19.3 19.4 19.5 19.6 19.7 19.8 19.9 20.0 a.1 20.2 20.;,0_,
P
Lx

FIQ. 28. Plastic strain path for the zigzag loading path OPlk
390 E. SHIRATORI and K. IKEGAMI

completely unloaded, was reloaded along the different path from that in the initial
preloading.
The existence of the change in the direction of the plastic strain vector corres-
ponding to that of the stress increment vector may be directly due to the shape
of the stress-strain diagram in the plastic range from the foIlowing consideration.

o Loading path OQ - type I

o Loading path OPb

type m
. Loading path OP% I

FIG. 27. Comparison between the stress-strain diagram of the preloading type I and that
of the preloading type III.

Figure 27 shows an example of the comparison between the stress-strain diagram

for the loading path OQ of the preloading type I and that for the loading paths
OPrjs and OpZ’l of the preloading type III. In this figure the origin of the stress-strain
diagram in the preloading type III was shifted so as to coincide with that in the
preloading type I, and the stress and strain were represented by the following
equivalent values.
0e* = d(d - % ql + %2),
Eeq = d[g (d + EzzEy 4 4
 Experimental study of the subsequent yield surface by using cross-shaped specimens 391
Corresponding to a small stress increment, a large strain increment was observed
in the plastic range for the preloading type I, while such a large strain increment was
not found for the preloading type III. This change of the stress-strain diagram
in the plastic range might have an effect on the existence of the change in the
directions of the plastic strain increment vector. Such a change of the stress-strain
diagram was not found for the loading path OS of the preloading type I and OPl'l
of the preloading type III. The correlation between the change of the stress incre-
ment vector was a little clearer in the hard plastic behaviour like the curve A than
in the soft plastic behaviour like the curve B in Fig. 27.

4.1) Yield surfaces of cold rolled brass plate

Brass plate (BsPl-H according to Japanese standards) 400 mm wide was cold
rolled to 1 mm thick from 8.5 mm thick under three stages. Biaxial tensile specimens
as shown in Fig. 2 were made with this plate. Their principal axis made various
angles (0 = O’, 15”, 30”, 45’) with the rolling direction of the plate. The equivalent
cross section of these specimens were newly determined in the each direction.
The yield surfaces were obtained by radial loading paths of Us/U2 = 0, l/4, l/2,
l/l, 2, 4, co as shown in Fig. 28, where the yield point was defined as the stress
corresponding to the value of the plastic strain of 0.01 or 0.02 per cent.
According to the theory of DORN (1949), the yield condition of cold rolled
plate, in the plane stress condition, is given as follows:

(ail cos4 0 + ass sin4 B + (a12 + a21 + ~44)toss 8 sin2 6) Uz2

+ (ali sin4 0 + a22 cos48 + (a12+ a21+ a44) coS2 0 sin2
S}uy"
+ {2 (a11 + a22 - 444) cos2 B sin2 8 + (a12 + a21) (cos4 B + sin4 O)} UZ Uy

+ (2 (2au + 2a22- 2al2 - 2a21 - a44) Cos2 8 sin2 8

+ tr44(~054e + sin4 0)) 7zy2

+ { 2 (2aii - ais - asi - ~44)sin e cos3e

+ 2 ( - 2a22 + a12 + a21 + ~44)sin3 8 cos e> Ux7zy

+ { 2 (2au - ai2 - a21 - ~44)sin3 8 cose

+ 2 (- f&X22+ CC12
+ a21 + a44)Sin 8 COS3 o} Us Txy = Uk2 (4.1)

where ukl is the constant of anisotropy, and 0 is the angle between the direction of
rolling and that of the principal axis of the specimen.
The coefficients of the equation (4.1), (ass/au), (a&u) + (azl/an) + (a&all),
(a12/all) -t (aZl/all),b44/ald9 (uk2/a11) = K, were determined as follows by using yield
stresses (I+)) of the tensile specimens cut at various angles (a) to the rolling direction
and some sets of stress components in the directions X and Y [(Q)Y and (us)r] at
yield under the radial loading of uy/uz = p for the case of the biaxial specimens
which direction of the principal axis X coincided with the rolling direction, namely,
the angle B = 0.
From the equation (4.1), the following equation can be obtained: for the unaxial
yield stress ufaj,
392 E. SHIRATORI and K. IKF,GAMI
-t
%A a22 sin2 u cos2 a + coS4u
-= - sin4 u + E+E+E (4.2)
=m { a11 ( 1

and, for the yield condition in the case of the angle 8 = 0,

(4.3)

The values of the coefficients, (as4&11) and (.212/un)+ (0121/cm)+ (a44/cm), were
calculated from the equation (4.2) with the least square method by using the uniaxkd
yield stress (Q) in the various orientations (a = O", 15",30",45', 60°,75", 90').
The values of K and (a12/o(~1)+ (a2~/c4 were calculated by the following relations
obtained from the equation (4.3),

kL)v kq/md

30

20

IO

0
(4 rk/mms (4 Ykg/mm’
FIG. 28. Comparison between the experimental values of the yield surfaces of cold rolled brass
plate and the calculated ones by Darn’s theory.
(a) e = o” (b) 8 = 15’ (c) B = 30° (d) 19 = 45’.
 Experimental study of the subsequent yield surface by using cross-shaped specimens 393

0(9]2 = K 9

E+zp K _‘_Ep
NdY (4Yl P

with values of a(s), (o& and (uy)r along the radial loading path of uy/oZ = p. The
authors adopted the mean value of (c&rr) + (asr/arr) calculated for five stress
ratios (p = l/4, l/2, l/l, 2, 4). Values of these coefficients are shown in Table 1.

TABLE 1. Darn’s constants of anisotropy

Value of
the proof strain 0.01% 0.02 %

aar
- I.075 0.898
(111

s+z+z 2.029 2.070

g + ;:: - I.081 - 0.948

a44
- 3.110 8.013
@I1

K w* 1.075X10' l.~Oxlo'
a11

Darn’s yield surfaces calculated by these coefficients were compared with the
experimental values as shown in Fig. 28. A qualitative coincidence was observed
between the calculated and experimental results, but the former was considered to
be quantitatively a little smaller than the latter.

5. CONCLUSIONS
The subsequent yield surfaces after prestraining along various loading paths
or after cold rolling were experimentally investigated with the cross-shaped speci-
mens of brass plate.
The following results were obtained.
The corner was found for the prestrain along the path of constant stress ratio,
except the case of the uniaxial prestrain. But the corner was not clear for the
prestrain along the path of variable stress ratios.
The yield point was independent of the path to determine it, except the pre-
stressed point for the prestrain along the path of variable stress ratios.
The correlation between the change of the stress increment vector and that of
the plastic strain increment vector for the zigzag loading was a little clearer
in the hard plastic behaviour than in the soft plastic behaviour.
The observed yield surface of the cold rolled brass plate quantitatively coincided
with the calculated values by Dorm’s theory, but the latter were a little smaller
than the former.
394 E. ,%IIRATOlU and K. IXEOAMI

REFERENCES
BEBTSCH,D.K.~~
FINDLEY, W.N. 1962 Appl. Mech. 2, 898.
DORN, J. E. 1949 J. Appl. Phys. 20, 15.
DUJCKER, D. C. and
STOCKTON, F.D. 1958 Proc. Sot. Ezp. Stress Analysis 10, pt. 2, 127.
ELULVEY, G. and McCowa, Jr. 1969 NASA TN D-896.
Hu, L. W. and BRATT, J. F. 1968 J. Appl. Mech. 25,411.
IVEY, J. H. 1961 J. Mech. Engng. Sci. 3, 15.
MAIR, W. N. and PUGH, H. Ll. D. 1983 NEL Report, No. 90.
NAGEDI, P. M., ROWLEY, J. C.
and BEADLE, C. W. 1965 J. Appl. Mech. 22, 416.
NACFIDI,P. M., ESSENBERO,F. 1958 Tram ASME 89,201.
PABKEB, J. and Bassmr, M. B. Trans. ASME, Ser. E, 31, 676.
PAUL, B., CEEN, W. and LEE, L. Appl. Mech. 2, 108’1.
PHILLIPS, A. and GUY, G. A. Trans. ASME, Ser. D, 83,275.
SEIRATOBI,E. and IKEOUI, K. Bull. Tokyo Inst. Technol. No. 80, 105.
TEIE~CARIS,P. S. and
HAZELL, C.R. 1965 J. Mech. Phys. Solti, 13, 281.
1”~. 4. I~efornmation of hard-drawn copper wedge. /I = GO”, l/lo = 165. The obscrvcd
deformation should be compared with the theoretical field shown in Fig. 1.
FIG+ 5. Profiles of a plasticine wedge at an advanced stage of’ defor~Il~~i~~.
(a) T/N z iM$Y, (b) T/iv = 1.0, (e) T/iv = 1.3, (d) T/N = I.5