Journal of Research of the National Bureau of Standards

Vol. 48, No. 2, February 1952

Research Paper 2291

Film-Rupture Mechanism of Stress Corrosion

Hugh L. Logan
The atmospherically formed protective films were removed by abrasion in an argon
atmosphere from surfaces of an aluminum alloy, two brasses, a magnesium alloy, and lowcarbon and stainless steels. The resulting surfaces were 0.12 to 0.76 volt more negative with
respect to a calomel electrode than surfaces prepared and measured under normal atmospheric
conditions. Appreciable changes in electrochemical solution potentials of notched specimens,
stressed in tension, occurred at or just above stresses at which the true stress-true strain
curves deviated from the modulus lines. These changes in potential were caused by rupturing
of the protective films at the roots of the notches and were of the order of 0.16 to 0.70 volt
at failure, depending on the material. Stress corrosion is postulated to occur in corrosive
media, at stresses sufficient to rupture the protective film, by electrolytic action between
the filmed (cathodic) and film-free (anodic) areas.
>
\

1. Introduction
Stress-corrosion cracking has been defined as the
"spontaneous failure of metal by cracking under the
combined action of corrosion and stress, residual or
applied" [l]. 1 Service failures have frequently resulted from residual stresses set up in the material or
structure during fabrication. These stresses may
result from the spinning or deep drawing of sheet
brass, the work-hardening of aluminum-alloy tubing,
the pressing of oversized bushings into aluminumalloy fittings, the riveting or welding of steel structures, or even as the result of the application of
excessive torques-in the assembling of brass-pipe
structures [2]. Stress-corrosion cracking may be
either intercrystalline, as in aluminum alloys, alpha
brass, and low-alloy steels, or transcrystalline, as is
usually the case in magnesium-base alloys and the
austenitic stainless steels.
I t is agreed by most workers in the field that some
degree of stress-corrosion cracking can be produced
in many alloys and commercially pure metals by the
proper combinations of stresses and corrosive media.
The A S T M - A I M M E Symposium on Stress-Corrosion Cracking of Metals [3], published in 1944, gives
an excellent summary of work in the field to that
date. A number of papers have also appeared in
recent years [4 to 8, 10] that contain excellent
bibliographies. For the most part, the data in the
literature are from investigations of the relative
susceptibilities of various materials to stress-corrosion cracking, susceptibilities of particular materials to stress corrosion in specific corrosive media, or
susceptibilities of particular materials to stress
corrosion after various mechanical or thermal
treatments.
The purpose of this investigation was to study the
mechanism of stress corrosion and not to determine
the relative susceptibilities of various materials to
stress corrosion. Recently the various theories of
the mechanism of stress corrosion were summarized
by Harwood [10]. Of these theories the "Generalized theory of stress corrosion" proposed by Mears,
Brown, and Dix [11], the "Mechanical theory" of
1

Figures in brackets indicate the literature references at the end of this paper.

Keating [5], and the "Film rupture theory", or better,

"Film rupture mechanism," are pertinent to the
present paper. I t is considered in this paper that
the "Film rupture mechanism" and the "Mechanical
theory" are logical parts of the Generalized theory.
I n the Generalized theory, corrosion is postulated
to occur by an electrochemical mechanism along
localized paths that are anodic to the surrounding
metal. There will be stress concentrations at the
bases of localized corroded areas; the deeper the
attack and . . . "the smaller the radius at the base
of the path the greater would be the stress concentration. Such a condition would act to pull the metal
apart along more or less continuous paths. At sufficient concentrations of stress, the metal might start
to tear apart by mechanical action . . . the tearing
action described above would expose fresh metal,
unprotected by films, to the action of the corrosive
environment. Because this freshly exposed metal
is more anodic, an increase in current from the base
of the localized path to the unaffected surface would
be expected . . .". I t is seen from this quotation
that Mears and his coworkers [11] postulate a mechanism in which protective films on metal surfaces are
ruptured. Keating [5] suggested that each advancing section of the crack in time encounters an obstacle
in the form of a nonmetallic inclusion, lattice discontinuity, or unfavorably orientated grain boundary.
In each case, the result may be reduction in the stressconcentration effect to a level at which cracking
cannot proceed. Such a process would explain
branching of stress-corrosion cracks, sudden changes
in direction at grain boundaries, etc. Gilbert and
Haden [12] report the results of an investigation on
an aluminum alloy containing 7 percent of magnesium in which they found that the cracking process
was not continuous but stepwise.
There are few published data from which the electrochemical solution potential of the freshly torn
metal at the root of the crack can be estimated.
Work at the National Bureau of Standards as early
as 1940* showed that the solution potential of freshly
quenched or freshly abraded 24S-T aluminum-alloy
sheet was much more electronegative (anodic) than
that of the same material permitted to stand for
99

some minutes before measurements were made [13]. cartridge brass, 61-percent-Cu-36-percent-Zn-3-perSimilar results were recently obtained at the Bureau cent-Pb leaded brass, AZ31f magnesium alloy, lowon 18-percent-Cr-8-percent-Ni austenitic stainless carbon steel, and type 302 stainless steel. The
steels. Mears and Brown [14] report a solution po- chemical compositions are given in table 1. The
tential greater than 3.0 v for commercially pure aluminum alloy, the leaded brass, and the steels
aluminum from which the film was removed by were obtained commercially. The magnesium alloy
scratching the surface with a glass stylus in the was supplied by the Dow Chemical Co., and the
electrolyte. Druet and Jacquet [15] report values cartridge brass was made in the Bureau's Experiof 0.75 v for mechanically polished specimens con- mental Foundry from electrolytic copper and 99.99taining 99.99 percent of aluminum (presumably with percent zinc. Prior to machining of the specimens,
a protective film intact) and 1.384 v for the same the materials were heat treated as follows: (a) The
material with measurements made immediately after aluminum-alloy stock was solution heat-treated a t
electrolytic polishing.
920 1 0 F , and quenched in water at approxiThe author is not aware of any data giving the mately 70 F , (b) the cartridge brass was rolled to a
magnitude of changes in solution potential produced diameter of about 0.62 in. and then annealed for 1
b y rupture of protective films on the metals or alloys hour at 1,560 F , after which the Brinell hardness
that are particularly susceptible to stress corrosion. number was 66 and the grain size approximately
Such data are necessary to establish the validity of 0.025 mm, (c) the low-carbon steel, originally in the
the so-called film rupture process in the generalized cold-rolled condition, was heated to 1,625 F and
theory of stress corrosion. I n the early phases of cooled with the furnace, (d) the stainless steel was
the investigation of the mechanism of stress corrosion, heated to 1,950 F and quenched in cold water. The
the following have been determined and are reported magnesium alloy, in the form of an extruded rod
in this paper: (a) the electrochemical solution poten- K in. in diameter, and the leaded-brass rod, % in.
tials of several alloys free from the protective effects in diameter, were used as received. The magnesium
of films,2 and (b) relationships between the applied alloy evidently had been extruded at a sufficiently
stresses and the electrochemical solution potentials high temperature that the metal had recrystallized
of these alloys.
following extrusion. The leaded brass had a Brinell
hardness number of 113 and a grain size of 0.015 to
0.025 mm. The tensile properties, as determined on
2. Materials and Methods of Test
subsize (0.355-in. diameter) ASTM round tensile
The materials used in this investigation were 24 S- specimens, are given in table 2.
When the protective film on 24S-T4 aluminum
T4 aluminum alloy, 70-percent-Cu-30-percent-Zn
alloy is broken in air, it tends to repair itself quite
2
It is recognized that contact of an electrolyte, at least with the aluminum and
rapidly. Therefore, specimens were prepared by
magnesium alloys studied, may produce a protective film on the metal surfaces.
TABLE 1.
Material
24S aluminum alloy a
Cartridge brass
Leaded brass...
AZ31 magnesium alloy a
Low-carbon steel
Type 302 stainless steel c
* Determined spectroscopically.

TABLE 2.

Al

Cr

2.9

0.17
.072

Chemical composition of metals used

Cu

3.9
69.4
61.2
0.01

Fe

Pb

0.4

19.17

24S-T4 aluminum alloy _____

Cartridge brass
Leaded brass
__ __
AZ31f magnesium alloy
Low-carbon steel
_
Type 302 stainless steel

Mg

Ni

0.6

.4
.81
.38

1.4

8.63

0.005
.017

Si

0.1

.03
.019
.297

0.024
.013

Zn

%
t>30.6
b35.8
1.0

By difference, c Manufacturer's analysis.

Mechanical properties of metals used

Material

3.0

.003

Mn

Ultimate
tensile
strength

Yield
strength

1,000 lb/in .2 1,000 lb/in.2

69.4
42.5
48.1
16.3
54.6
43.6
41.0
33.5
54.7
31.5
91.1
33.5

Elongation
in 1.4 in.
Percent
24
54
28
11
40
77

abrading them with "400 aloxite" metallographic

polishing paper, and all measurements were made in
an argon atmosphere 3 in a dry box. I n making
measurements, drops of the desired electrolyte were
3 An analysis of the argon by the mass spectrograph method showed 0.2 percent
of N2 as the only detectable impurity.

placed on the prepared surface and brought into

electric contact with a calomel electrode of the
saturated KC1 type by means of an agar-agar bridgePotential measurements were made with a p H meter,
or for values in excess of 1.3 v, with the p H meter in
series with a L&N type K potentiometer.
I t was recognized that the preparation of the specimen surfaces by abrasion left the surfaces to be
measured in the cold-worked condition and might
produce phase changes in the stainless steel. In
some earlier work at the Bureau the author reduced
the thickness of some 24S-T4 aluminum-alloy sheet
approximately 5 percent without appreciable changing its solution potential. Druet and Jacquet [15]
showed that severely cold-worked high-purity aluminum was only about 0.04 v more negative than the
annealed material. The author has seen no data on

100

the effect of cold-work on the solution potentials of

other materials studied nor no the effect of a phase
change on the solution potential of the stainless steel.
Stress-electrochemical solution potential relationships were obtained on notched specimens made in
accordance with the drawing in figure 1. The
specimens were covered, except for the threaded
ends, with two coats of Glyptal red enamel. After
the enamel had dried it was cut through b y means of
a razor blade so as to expose an extremely narrow
width of metal around the specimen a t the root of
the notch. The specimens were placed in cells in a
tensile-testing machine, figure 2, with a calomel
electrode placed directly in the cell if potassium
chloride was used as the electrolyte or in a beaker
connected to the cell b y an agar-agar bridge if some
other electrolyte was used. The specimens were
stressed to failure in the corroding solution, and the
electrochemical solution potentials that corresponded
to given loads were determined either with a p H
meter used as a vacuum-tube voltmeter or a Miller
multimeter (used as a high-resistance potentiometer).
The free cross-head speed of the testing machine was
held between 0.05 and 0.10 in./min. I n order to
obtain true stress-true strain and load-true stress

relationships, three or more additional notched

specimens of each material were broken in the
uncorroded condition. 4 True stress-electrochemical
solution potential relationships were then obtained
by combining load-true stress and load-electrochemical solution potential data.

3. Results a n d Discussion
3.1. Electrochemical Solution Potentials of the Metals
The different electrolytes used and the electrochemical solution potentials of the various materials,
determined in air and in argon, are given in table 3.
All values are in terms of a calomel electrode of the
saturated KC1 type and were determined at approximately 25 C. The values reported for atmospherically formed filmed surfaces were equilibrium values
obtained by immersion of the prepared surfaces in
their specific electrolytes; the values reported for the
specimens prepared in argon were instantaneous
maxima and not necessarily equilibrium values.
These values are given inasmuch as they represent
the conditions under which changes in electrochemical solution potential values reported in a
later section of this paper were obtained.
T A B L E 3. Electrochemical
solution potentials (with respect to
calomel electrode of saturated KCl type) of materials ijoith
surfaces prepared and measurements
made in air or argon

Material

L
L-n
p a n "_* 00 001 0
"0-280

Electrolyte

Electrochemical
solution
potentials in
Air

F I G U R E 1.

Details of notched

Argon

specimens.
24S-T4 aluminum alloy
70%-Cu-30%-Zn brass
61%-Cu-36%-Zn-3%-Pb
brass
AZ31 magnesium alloy
Type 302 stainless steel

5% NH4OH+
5% (NH 4 ) 2 C0 3
do....

Volt
Volt
-0.67 -1.43
- . 4 7 -0.96
-.39

-.98

3.5% NaCl+2% K2CTO4 -1.51 -1.645

-0.60 -0.72
5% NH4NO3
-.28 -.78

The data indicate that, under normal atmospheric

conditions, protective films are formed on the surfaces
of these alloys. These filmed materials are in some
cases as much as 0.76 v more electropositive than the
same materials with surfaces prepared under argon.
The largest differences in potential are found on the
aluminum alloy, the brasses, and the stainless steel,
all of which are relatively corrosion-resistant materials. Those materials that corrode more readily,
such as low-carbon steel and the magnesium alloy,
had very much smaller differences in potential
between filmed and unfilmed surfaces.
In order to determine whether the difference in
solution potential of the cartridge brass was due to
the formation of an oxide film or to some other cause,
the following investigation was carried out. A
F I G U R E 2.

Notched specimen (A) in place in cell,

(B) in testing
machine.
Electrode (C) was placed directly in cell if potassium chloride was the corrodent,
otherwise an agar-agar bridge was used.

4 '1 he use of notched specimens, of course, introduces stress concentrations of

undetermined magnitude at the root of the notch. The maximum stress at the
root of the notch is a function not only of the geometry of the specimen, but also
of the aprlied stress. Although the radii of the notches used were small (approximately 0.005 in.), they were nevertheless large compared to those at the roots of
stress-corrosion cracks in many materials.

101

specimen was placed in the dry box, and a surface

was prepared and measured under argon. Approximately 20 percent of the argon was then
replaced by dry oxygen, and additional measurements were made on the same surface. The experiment was repeated with wet oxygen, wet carbon
dioxide, and dry ammonia gas. The results are
given in table 4.
T A B L E 4.

Electrochemical

TRUE STRAIN , LOG e A</A

.010
110

100 I

90

and
s

Electrolyte was 5% NH 4 OH+5%(NH4) 2 C0 3

Argon
80% argon+20%
80% argon+20%
80% argon+20%
80% argon+20%
Air
.

dry
wet
wet
dry

Electrochemical
solution
potential

Atmosphere

.030

.040

~I

.050

.060

.070

1 /
\

80 I

60

50

I t will be noted that in every case the solution

potential was more positive after the addition of
the second gas, the change being most marked in the
case of oxygen, either dry or wet. In the atmosphere containing ammonia, the surface of the alloy
adjacent to drops of liquid used in making the measurements became, tarnished; its final potential was
0.31 v compared to 0.47 v for brass measured in
air. Measurements made on the surfaces of brass
specimens covered with a blue corrosion product
after exposure to an atmosphere containing ammonia,
carbon dioxide, air, and water vapor were 0.75 v more
electropositive than film-free surfaces of the same
alloy.

y^
/ A V RANGE

^^^
/^"^

\\ 1

70

Volt
-0.96
- . 4 5 to - . 5 0
-.45 to-.50
- . 7 2 to - . 7 8
- . 5 3 to - . 6 1
-.47

02-.
0 2 -_.
C0 2 NH 3 .

"

solution potentials of 7 0 % - C u - 3 0 % -

Zn brass with surfaces prepared in an argon atmosphere

measurements made in a mixed atmosphere

.020

rr //

\\

\\
W

i
i

L
f1

/
//

40

100

-i

\
\

/TRUE STRESS- TRUE STRAIN

A
J

J
A
~\

200

300

., 1 ... ,.. _i
400
500

,.

600

700

AV* V i - V o , Millivolts

F I G U R E 3.

True stress-electrochemical

tionships for 84S-T4

aluminum

solution

alloy

potential

rela-

specimens.

Corroding medium, saturated KC1. Data are from seven specimens.

.100

TRUE STRAIN , L0G e A 0 /A

.200
.300

3.2. Stress-Electrochemical Solution Potential

Relationships
The relationships between the true stresses and
electrochemical solution potentials of the materials
are shown in figures 3 to 8. The maximum values
of the potentials were usually obtained just at the
failure of the specimen as the result of the exposure
of large areas of unfilmed material. These values
are of course not shown on the figures but are given
in table 5. I n any measurement of electrochemical
solution potentials the measured potentials are
T A B L E 5.

Maximum

change in electrochemical
tial with stress.

Material

24 S-T4 al uminum alloy

70%-Cu-30%-Zn Brass
61%-Cu-36%-Zn-3%-Pb
brass.
AZ31 magnesium alloy
Low-carbon
steeL
302
stainless steel.
_.

solution

poten-

Electrolyte

Change in
electrochemical solution
potential

Saturated KC1
5% NEUOH+5% (NH 4 ) 2 C0 3
do

VoU
0.610
.700
.510

3.5% NaCl+2%K 2 Cr0 4

5% N H i N 0 3 - - - ~ Saturated KC1

.240
.160
.315

300
4 0 0 500
AV=Vi-Vo, Millivolts

F I G U R E 4.

True stress-electrochemical

tionships for 70 percent OM-30

800

solution

potential

900

rela-

percent Ttii cartridge brass.

Corroding medium, 5 percent NH4OH + 5 percent (NHO2CO3. Data are from

seven specimens.

102

necessarily the integrated potentials over the entire

area in contact with the electrolyte. I t is therefore
to be expected that potential measurements made
on notched specimens of the same material may
vary somewhat from specimen to specimen, hence
the ranges of potential at given true stresses rather
than average curves are shown in the graphs.
There was an appreciable change in the solution
potential of the aluminum alloy (fig. 3) at stresses at
which the true stress-true strain curve deviated
from the modulus line, that is, at stresses at which
plastic deformation had occurred. Marked changes
in the potentials of the brass specimens (figs. 4 and 5)
occurred at stresses that were 10,000 to 25,000 lb/in. 2
above those at which the true stress-true strain
curves deviated from the modulus line. The solution
potential of the magnesium alloy changed appreciable
at stresses only about one-third of those necessary to
produce deviation from the modulus line (fig. 6). The
yield point in the low-carbon steel was accompanied
by a marked change in potential (fig. 7). Stresses
sufficient to produce a deviation of the true stresstrue strain curve from the modulus line for stainless
steel also produced appreciable changes in the
solution potential (fig. 8). The electrochemical
solution potentials (except for the low-carbon steel)
became increasingly more negative as the stresses
were increased. For every material (except lowcarbon steel) stresses were reached, as is seen in
figures 3, 4, 5, 6 and 8, at which the slope of the true
stress-potential curves decreased markedly, and the

TRUE

.050
I

STRAIN, L 0 G e A 0 /A

.100

.150

100

150

AV = V i - V o , Millivolts

F I G U R E 6.

True stress-electrochemical

tionships for AZ31 magnesium

100

90 Y

TRUE STRAIN, LOG e A 0 / A

.20
.40
.60
r
i ' i
T
T"

AV RANGE ^

70

\
g60
o
$50
UJ
QL
y-

</)

u 40

/ /

TRUE STRESS-TRUE STRAIN

\ 1
A

30

20

10
1

F I G U R E 5.

1
1
1
1
100
200
A V = V i - V o , Millivolts

True stress-electrochemical

1
300

solution

-L

1
400

potential

solution

alloy

potential

rela-

tionships for leaded brass.

Corroding medium ,5 percent NH4OH+5 percent (NH4)2C03. Data are from
five specimens.

rela-

specimens.

Corroding medium, 3.5 percent NaCl+2.0 percent K^CrOi.

eight specimens.

.80
1

^ ^ ^~~

80

.200

Data are from

potential increased rapidly to the failure of the

specimen.
I t is postulated that during the loading of the specimens in tension the protective film is broken over
minute areas of the surface. If the potential of the
unprotected area alone could be measured, it would
presumably be of the same order of magnitude as
that determined on a film-free surface. The area
where the film has been ruptured is very small compared with that of the filmed surface of a specimen,
and consequently the measured potential differs by
only a small amount from that obtained on an unbroken filmed surface. As the stresses are increased
the ratio of the film free to the filmed surface of a
specimen increases, and the solution potential becomes more negative, until at the instant of fracture
the area of unfilmed surface may be very large compared to that of the filmed surface, and the potential
may closely approach that of completely unfilmed
material.
At a constant stress the protective film tends to
repair itself. This was demonstrated by results
obtained on aluminum-alloy specimens. After an
appreciable change had been obtained in the solution
potential during loading, the applied load was held
constant, and within 1 minute the solution potential
was the same as that of the unstressed specimen.
After the specimens broke (exposing large unfilmed
areas) the electrochemical solution potentials rapidly
103

assumed the same values as those of the filmed and

unstressed materials.
Examination of the fractured surfaces of the
notched specimens and the marked changes in the
potentials of all of the materials except low-carbon
steel at fracture indicated t h a t the final failure had
been almost instantaneous. I n the low-carbon steel
specimens, however, both the appearance of the fracture and the small change in potentials at failure
indicated t h a t prior to the final fracture, cracks had
penetrated into the material from the roots of the
notches and that the progression of fracture was not
rapid and was probably discontinuous.
Consequently, at failure the amount of freshly exposed
unfilmed metal was too small to produce any marked
change in the potential.
The data indicate that when the stresses are high
enough to cause plastic deformation the atmospherically formed protective film will be broken and the
exposed metal will become electronegative with respect to the filmed material. These film-free areas
will, in general, be very small compared to the filmed
areas. If such areas are connected to filmed areas by
a conducting liquid, for example, by a drop of liquid
containing dissolved gases or salts, a short-circuited
electrolytic cell will be set up, the circuit being com-

TRUE STRAIN, L0G e A 0 /A

.200
.400
.600

.800

225

80
120
160
200
A V = V i - V o , Millivolts

280

F I G U R E 8. True stress-electrochemical solution potential

tionships for type 302 stainless steel specimens.

320

rela-

Corroding medium, saturated KC1. Data are from six specimens.

TRUE STRAIN, LOGe A 0 /A
.100
.200
.300

.400

110

100

90

8 eo
o

70

60

50

40

F I G U R E 7.

50

100
A V = V i - Vo, Millivolts

True stress-electrochemical solution

tionships for low-carbon steel.

Corroding medium, 5-percent NH4NO3.

200

150

potential

rela-

Data are from nine specimens.

pleted through the metal itself. As the film-free

(anodic) area is generally very small compared with
the filmed area, the current density over the anodic
area will be high, and corrosion will proceed at a
rapid rate until a readjustment of stresses permits
reforming of the protective surface film. Corrosion
will then become more general (or stress corrosion
will proceed along different paths) until the stress
concentration at the particular path under consideration again becomes sufficient to rupture the protective film. The author believes that successive
breaks in the time-potential curves reported by
Gilbert and Haden [12] indicate that such processes
were occurring, probably at several points, on their
test specimens.
The experiments reported in this paper confirm
the general idea of the film-rupture mechanism
suggested by Mears, Brown, and Dix. I n the theory
as first presented, it was postulated that tearing of
the metal was necessary to expose film-free material.
However, the present data indicate that film rupture
occurs with comparatively small amounts of plastic
deformation of the underlying material. When the
film ruptures, stress corrosion may start in the filmfree areas at stresses much below those normally
required to tear the metal apart. Unless conditions
permit film repair, stress corrosion may progress to
cause complete failure in a relatively short time. If
the film is restored and the stresses are not increased
or readjusted, stress corrosion may cease. If, however, the stresses are readjusted by any means whatsoever, for example, by a change of external load or
by corrosion, the film may again be broken and

104

stress corrosion proceed discontinuously to failure.

The last process may be t h a t which occurs in the
transcrystalline type of stress corrosion.

The author gratefully acknowledges the help of

Thomas P. Royston, Jr., in obtaining the data
presented in this paper.

4. Summary

5. References

As a part of a detailed study of the mechanism of

stress corrosion, the electrochemical solution poten*~ tials of filmed and film-free surfaces and the true
stress-electrochemical solution potential relationships
for several alloys were determined.
The electrochemical solution potentials of surfaces,
from which the atmospherically formed films had
been removed by abrasion in an inert atmosphere,
were 0.10 to 0.76 v more electronegative (with
'* respect to a calomel electrode) than filmed surfaces
of the same materials.
Appreciable changes in electrochemical solution
potentials accompanied plastic deformation, and just
prior to fracture the solution potentials were 0.16
to 0.70 v (depending on the alloy) more electronegative than the unstressed material.
^
The results of the investigation indicate that at
stresses sufficiently high to produce failure of the
protective film on metals, stress corrosion occurs in
a corrosive medium by electrolytic action between
film-free (anodic) and filmed (cathodic) areas. The
process proceeds continuously or discontinously to
failure, depending on whether or not the subsequent
^ adjustment of stresses permits reforming of the
protective film.

[1] Metals Handbook, p. 14, 1948 ed. (Am. Soc. Metals,

Cleveland, Ohio).
[2] W. B. Price, Am. Inst. Mining Engrs. 89, 268 (1930).
[3] Symposium on Stress-Corrosion Cracking of Metals,
Am. Soc. Testing Materials-Am. Inst. Mining Engrs.,
1944 (Am. Soc. Testing Materials, Philadelphia, Pa.).
[4] U. R. Evans, Symposium on Internal Stresses in Metals
and Alloys, Inst. Metals, p. 309 (1948).
[5] F. H. Keating, Symposium on Internal Stresses in Metals
and Alloys, Inst. Metals, p. 329 (1948).
[6] E. Hugony, Alluminio 17, 225 (1948).
[7] C. A. Zapffe, Trans. Am. Soc. Mech. Eng. 66, 81 (1944).
[8] H. P. Croft and G. Sachs, Iron Age 151, 47 (March 11,
1943); 151, 62 (March 18, 1943).
[9] H. P. Croft, Proc. Am. Soc. Testing Materials 41,
905 (1941).
[10] J. J. Harwood, Corrosion 6, 249 and 290 (1950).
[11] R. B. Mears, R. H. Brown, and E. H. Dix, Jr., Symposium on Stress-Corrosion Cracking of Metals, Am.
Soc. Testing Materials-Am. Inst. Mining Engrs., p.
329, 1944 (Am. Soc. Testing Materials, Philadelphia,
Pa.).
[12] P. T. Gilbert and S. E. Haden, J. Inst. Metals 77, 237
(May 1950).
[13] H. L. Logan, J. Research NBS 25, 315 (1940) RP1328.
[14] R. B. Mears and R. H. Brown, Ind. Eng. Chem. 33,
1001 (1941).
T15] Y. Druet and P. A. Jacquet, Metaux & Corrosion 32,
139 (1947).

105

WASHINGTON, October 1, 1951.