313 Torsional Hysteresis of Mild Steel .
By J. J. Guest and F. C. Lea .
( Communicated by Sir Oliver Lodge , F.R.S. Received November 2 , 1916 .
) The stress-strain curve from no load to fracture for mild steel as usually obtained consists of three parts : ( 1 ) A straight line , followed by a part deviating only slightly from this straight portion ; ( 2 ) a sharp bend , followed by a part approximately parallel to the strain axis ; and ( 3 ) a curved rising part , leading ultimately to the breaking point .
It is generally assumed that Hooke 's Law holds throughout the part ( 1 ) , and is immediately followed by the sharply defined bend which constitutes the yield point .
For mild steel first stressed in tension and then in compression , or subjected to positive and then negative torsional stresses , the stress-strain curve within a considerable range of stress is also supposed to be a straight line .
It is further well known that if mild steel is stressed in tension beyond the yield point the elastic limit is raised , but only at the expense of lowering it in compression ; or , if it is twisted beyond the yield point in one direction , its elastic limit is raised for stresses in that direction , but lowered for those in the opposite direction .
Attempts have been made to relate the range of stress through which the stress-strain curve is a straight line with that through which a material , such as mild steel , can be stressed an infinite number of times without fracture .
This is expressed by the well known Bauschinger 's Law , which , as stated by Mr. Leonard Bairstow , * is as follows :\#151 ; " The superior limit of elasticity can be raised or lowered by cyclical variations of stress , and at the inferior limit of elasticity will be raised or lowered by a definite , but not necessarily the same , amount .
The range of stress between the two elastic limits has therefore a value which depends only on the material and the stress at the inferior limit of elasticity .
This elastic range of stress is the same in magnitude as the maximum range of stress , which can be repeatedly applied to a bar without causing fracture , no matter how great the number of repetitions .
" In the paper by Mr. Leonard Bairstow , just referred to , its author , in discussing the stress-strain curve obtained after a certain number of repetitions through a given range of stress from a mild steel specimen , remarks:\#151 ; " The behaviour of this specimen illustrates the necessity for Bauschinger 's hypothesis relating to primitive elastic limits , as the extensometer was * " Iron and Steel under Cyclical Variations of Stress , " 'Phil Trans. , ' A , vol. 210 ( 1909 ) .
Messrs. J. J. Guest and F. C. Lea .
incapable of showing the first deviations from elasticity .
At a slightly lower range , probably 13 tons per square inch , the specimen would have been really elastic , as no number of reversals would have produced a hysteresis loop .
" On the other hand , others have suggested the existence of a hysteresis loop when mild steel is stressed through ranges within the " primitive elastic limits .
" Prof. Ewing and others , from experiments on wires , have supposed its existence .
It is doubtful , however , whether the results of experiments on wires can be taken as holding for material in a more normal state .
Experiments by Lea on hard drawn steel tubes certainly show that the material of such tubes is in a very different condition to hot worked or annealed steel .
Experiments by Guest , an example of which is given in the ' Phil. Mag. , ' July , 1900 , p. 89 , showed definite hysteresis loops in Swedish iron under tension , but this does not appear to have been confirmed by other workers\#151 ; probably from lack of instruments having the necessary precision\#151 ; -and whether or not such a hysteresis loop does exist and the effect upon it of various treatments of the material , although probably a matter of very considerable practical importance , as also of scientific interest , as likely to throw light upon doubtful matters , does not appear to have been subjected to adequate investigation .
The general failure to establish clearly the existence of a hysteresis loop for mild steel stressed alternately in tension and compression , and the difficulty of reversing stresses readily from tension to compression , suggested the desirability of making an accurate and sensitive torsion machine , in .which a specimen could be subjected to pure torsion in either a clockwise or anti-dockwise direction , and in connection with which angular strains could be measured with very great precision .
Designs for such a machine were accordingly prepared by Mr. Guest in 1912 , and the machine was constructed by Prof , Lea in the Civil Engineering Department of the University of Birmingham , and was erected , after military occupation of the University Buildings in 1914 , by the kind permission of Sir Oliver Lodge in his laboratory at Edmund Street .
Description of the Apparatus .
The machine itself consists of a head A , Fig. I , carried on an axis BC , supported horizontally by a bracket , and holding the specimen DE so that it is perpendicular to BC .
The other end E of the specimen is secured to a rocking arm FJ , having turned arcs FH and JIST .
This arm is supported immediately by the equivalent of a universal joint , containing a ball bearing within the arm ; this is not shown in the diagram .
The axis of the Torsional Hysteresis of Mild Steel .
315 specimen is set to be horizontal by means of the right and left turnbuckle Q in the tackle supporting the rocking arm .
The anchorage of the suspension chain P has an adjustment parallel to the axis of the specimen , so that various lengths of specimens can be accommodated and the chain set vertical .
From the arcs FH and JN the weight-carriers K and L are suspended by the steel strips M and N ; these are loaded by hand , weights on one side givino-positive and on the other negative torques .
A rod , with an adjustable weight R , projects from the rear of the head A , so that the head , together with any fraction of the specimen determined upon , can be balanced , and thus zero bending moment at any desired cross-section of the specimen can be secured .
Similarly , the balance of the rocking arm about a horizontal axis perpendicular to the specimen can be adjusted by the weight G , so that zero bending moment can be enforced at* a second cross-section of the specimen .
The rocking arm was weighted so that its centre of gravity was just below the axis of the specimen , making the equilibrium just stable .
The arrangement , see Fig. I , for measuring the twist consists of a mirror S carried on a rod held in a frame attached to the specimen near its fixed end D , and a second mirror T similarly carried near the other end E of the VOL. XCIII.\#151 ; a. .
2b Messrs. J. J. Guest and F. C. Lea .
specimen .
These are set opposite to each other with their reflecting faces parallel to the axis of the specimen .
The twist between the points of attachment of the mirror frames is read at a single reading of the scale YW , Fig. I , by a telescope at X , the course of the light being indicated by the broken line .
The scale is bent into an arc , with T as centre , to avoid the necessity for correction and to preserve the sharpness of definition of the telescope .
The machine is designed on such a scale that hollow specimens machined from the solid can be employed , and the use of drawn metal avoided .
A solid specimen , f ( 08755 + 0*00025 ) inch diameter and 16 inches long between the points of attachment of the mirror frames , has been used in this research.* Tests for sensitiveness of the rocker arm showed that it turned freely when not attached to the specimen , with an addition of 0'05 lb. on one weight-carrier , when each carrier was loaded with 100 lb. The sensitiveness of the apparatus , with the specimen in position , was also tested at various loads during the course of the experiments .
With the above-mentioned specimen in position , small load variations showed that the machine was readily sensitive to loads corresponding to the degree of accuracy with which the scale could be read\#151 ; 0-002 inch .
The weights used were all adjusted to OT oz. , and , when necessary , care was taken to employ them in the same order , thus any small inaccuracies in the determination of this mass have no appreciable effect upon the hysteresis loops presently to be described .
The diameter of the rocker was 40 inches , so that a load of 1 lb. produced a torque of 20 inch-lb .
, corresponding to a maximum stress in the bar of 152 lb. per square inch .
The scale used was 25 inches long , divided into fiftieths of an inch , and was set up at a distance of 94 inches from the mirror T. The magnification of the telescope was such that the fiftieth divisions could readily be subdivided into tenths by estimation .
Thus , a scale reading of 1 inch corresponds to a shear strain of 0-8755 h-(4 x 16 x 94 ) = 0-0001454 radians , and the readings were made to of this , or 0-000000291 radians .
Luring the experiments the bar was loaded beyond the yield point , and in order to follow the strain over the range involved in the permanent set obtained , the zero on the scale was shifted .
To do this the mirror-holder at the outer end of the bar was unclamped , and simply rotated in its carrier \#166 ; * The analysis of the steel used in the experiments , kindly supplied by Messrs. Vickers , gave the following percentages:\#151 ; Carbon 0T5 , manganese .
0-30 , phosphorus 0037 , sulphur +033 , silicon 0 055 .
Torsional Hysteresis of Mild Steel .
317 frame about an axis parallel to the specimen , and reclamped ; the apparatus was designed to permit of this being easily done .
Before inserting ; the specimen , the head and rocker were balanced so that there should be zero bending moment at two surfaces in the specimen at about a quarter the distance of the point of attachment of one mirror frame from the other .
The chain anchorage was then adjusted so that the chain was vertical with the specimen in position .
The specimen was inserted , and levelled by the use of the right- and left-hand screw buckle .
The weight-carriers were then loaded equally and sufficiently to give steadiness while passing through zero loads .
Finally , the mirror frames and telescope were set in position .
From a considerable number of tests , typical examples have been selected illustrative of the various points investigated in the course of the experiments : ( 1 ) the existence of hysteresis loops at low loads ( Fig. II ) ; ( 2 ) their relation to those over a wider ( but elastic ) range and \#163 ; o small loops at high average stress ( Fig , III ) ; ( 3 ) the effect of plastic overstrains ( Figs. IY and Y ) , of rest ( Fig. YI ) , and of mild heat treatment ( Figs. YII and YIII ) .
In Table I and Fig. II are shown the results of some tests on the bar with low loads .
The stress curve AB , which in the sequence is referred to as the natural stress-strain curve , is plotted to the scales shown on the figure .
It appears to be a straight line , the differences of the strains obtained by rising and falling loads not being visible on the scale to which AB is drawm .
To render these differences perceptible , a geometrical straight line joining the points A and B is considered , and the differences of the observed strains , from those corresponding to the geometrical line AB , are multiplied 100 times , and set off either to the right or left of this line , according as the observed strain is greater or less than that corresponding to the geometrical line .
Thus , suppose the diagram to start from the point C. On the normal scale the point C is on the line AB , coinciding with the point D. On loading with +5 lb. , taking off the load and loading with \#151 ; 5 lb. , taking off and reloading with +5 lb. , the diagram on the exaggerated scale is HG , with no width to the hysteresis loop .
On the natural scale , the part corresponding to HG is indistinguishable from the portion of the line AB , obtained by projecting HG horizontally on to it .
On now loading to \#151 ; 10 lb. , and then unloading by intervals of 5 lb. to zero , and loading to +10 lb. and back to zero , the small loop CLDKC on the secondary scale is obtained .
Actually , in this experiment this loop CLDKC was taken before HCG ( see Table I ) .
By loading to 25 lb. , unloading to zero , loading to \#151 ; 25 lb. , and unloading to zero , the much larger loop CMAFBN is obtained .
All the parts of these loops , when 2 b 2 Table I ( from which Fig. II has been plotted).\#151 ; Scale Headings in Inches .
The arrows indicate the sequence in loadi 318 .
Messrs. J. J. Guest and F. C. Lea .
bb c \#163 ; .3 \#163 ; Negative .
25 .
00 I 00 1 \#163 ; 1 CO 00 CM CM a\gt ; 0\gt ; id rH 00 CO. IP o CO o\gt ; oi o r-H 10 -087 10 -088 10 -090 i 10 -076 id 1 \gt ; 1 I 10 *520 10 *523 10 507 i cm \gt ; 1 \gt ; g 1 A 1 S4 M o mm ii i s p OOOOOOv0 o o o oo rHrHrHrHrHrHTrH rH rH rH HrH / d 1 i CM S f !
f !
s I 1 I id \gt ; / \#163 ; CO CO CO rH rH rH rH rH rH o rH 11*806 11 *806 11*806 11 *805 11 *818 id rH S 8 .
12 *236 12 *246 cm cm S rH CO " 1 Torsional Hysteresis of Mild Steel .
319 plotted on the natural scale , appear to be within the width of the physical straight line AB .
In the secondary curves all observation points are joined by straight .
\p\j ; ts ?
qi2SI .
-xrfaoL sqi -\p^|Q2 = 911 ?
q-| -Muy vo prwq o !
fl ' O IT ) CM \#151 ; lines , no attempt having been made to ease off any errors by drawing smooth curves .
This description will make clear the procedure in connection with the succeeding curves .
In all cases a geometrical line similar to AB is Messrs. J. J. Guest and F. C. Lea .
taken as an axis of stress , from which the secondary curve strains are plotted .
The slope of the line AB gives the modulus of rigidity " C " for the range of loading of \#177 ; 25 lb. , or range of stress \#177 ; 3800 lb. per square inch , and , if plotted only on the natural scale , it would appear to be constant .
The exaggerated secondary curves show , however , that it varies slightly with the range of loading .
The value of the modulus for the bar is about 5390 tons per square inch .
In Fig. Ill is shown in a striking manner the increase in the size of the hysteresis loop for larger ranges of stress , and also the small hysteresis loops obtained by increasing and decreasing the load at any stage in the circuit .
Again , the secondary curves are plotted from the line AB , but they are exaggerated 50 instead of 100 times .
The sequence of the tests is shown by arrowheads on the diagram and by the figures in Table II .
Starting at the point C on the zero stress line , a 50-lb .
cycle was taken , and is shown marked No. 1 ; a 50-lb .
half-cycle , marked No. 2 , is then drawn .
It terminates at D , and is followed by cycles 3 ( through E and terminating at F ) , 4 ( to G ) , 5 ( to M ) , and a half-cycle 6 ( to L ) taken to loads of +100 lb. , with a number of small loops ( such as GNHG ) , having a range of 20 lb. , taken at various stages of the loading .
For clearness of the figures , the loops 5 and 6 have been shifted vertically , the line A'B ' being the new position of the line AB .
Different types of broken lines are used to aid lucidity .
The course of the observations is indicated by the arrowheads on the secondary curves and by the arrows in Table II .
The width of the hysteresis loops increased with the range of loading .
It will be noticed , however , that , wherever on the range of stress the small included loops were taken , the widths for a definite range were nearly constant , but were rather greater near the zero load line than at the loaded parts of the main curve .
The value of C , which has previously been decreasing , shows a very slight rise in the + 100-lb .
cycles .
The central pair of small loops are inclined to the main curve at the point where they branch from it , and , compared with AB , they give a slightly higher value for the rigidity .
The unloading side of the other pair of small loops coincides with the unloading line of the main hysteresis loop .
In these tests there was no perceptible permanent creeping at any definite ' point , but a small amount apparently occurred towards the \#151 ; 100 lb. loading , as shown by the curve 4 not coinciding exactly with 3 .
Except for this , within all these hysteresis loops the strain arrived at was definite and unaffected by time .
The strain , however , at any or no load , depends upon the previous history .
Thus at no load the specimen has been successively left with the strains indicated by the points CDEFGHKLM , and the 'Torsional Hysteresis oj Steel .
cqisgi = anbuoispunod^i 02 - 911--- stH P*'l e ScaJe Rgadirvqs .
I Divi .
Table 'll ( from which the loops 1 to 6 , Fig. Ill , have been plotted).\#151 ; Scale Headings in Inches .
The arrows the sequence of loading .
Messrs. J. J. Guest and F. C. Lea .
\#163 ; S .3 \#169 ; too \#169 ; \#163 ; \#169 ; O Ph o QO o d d \lt ; M o cc S ' o CD d 00 +S4- QO I tP | QO I 00 | -121 cq HP cq ^ \#171 ; H ( M C 1^ IS J\gt ; CO CO 00 00 rft ^ QO \lt ; M H Oi 05 Ci cp cp 9 CO 00 00 CO O QO o 00 CD \lt ; N iO H to to to io -H 00 $3 \gt ; \lt ; \lt ; I QO CD 00 I\gt ; rH rH I\gt ; I\gt ; O 00 CD 00 05 Tp GO 00 r\#151 ; I rH rH r\#151 ; l I\gt ; |\gt ; Jt\gt ; 00 r\#151 ; i I\gt ; t\gt ; O CD 00 00 oT HP 00 ^ Ol 05 ft 05 X O * .
O X X \#151 ; .
O X ( N X X oo i\gt ; cq cd h 05 00 C5 00 05 X X X X X V x cq cd CD 05 CD 00 00 00 cq cq cq rH rH rH X O CD CD X t'* H lO O ^ 00 Tft 00 rfl 00 cq cq cq cq cq r\#151 ; i rH rH rH rH A. .
^ rfi rfi coto WIO cq cq cq cq OO X X 00 r-i r-^ r^\ r-i O rH O _ _ O rH O .
O -H V d d A H d Tf - !
fl iH rH pH I i\#151 ; i rH rH rH rH rH o o cq \lt ; m o x , O^G5WX(N Are \lt ; 9 7^ ?
7* 05 2 .
rfirPOO^ sq to ( M to h X 00 X X X to l O 'O to to rH rH rH rH rH CD CD O O O O H* CD Tfi CO rH CD tO to to to tp tp i\gt ; l\gt ; Ih l\gt ; in rH rH rH rH rH rH cq 05 o oo g ?
s s X X S ' v\#151 ; ?
y \#169 ; ^ 1 \#163 ; 3 ' o HP CO 05 CO cq rH S3 I s XX X to to Y to H H 1 H o rH 9-790 9-762 end o 9-790 9 -744 9-760 9-748 end o i o s S o o o o " H CD X CD ZD CD CD CD O w O O O H Y H H H 10-672(E ) 10 -618(F ) 10 -666 10 -608(G ) 10 -662 10 -640 10 -648 10 -638 10 644 10 -600 ( M ) 10 -622(H ) 10 -610(G ) 10 -660 ( L ) o rH 11 -502 11 -528 11 -500 11 -526 11 -500 11 -510 11-470 11 -482 HP X l w to H A ip ip Ip I I\gt ; X\gt ; I y -ft tP cq 05 2 1 05 Torsional Hysteresis of Mild Steel .
323 material may be left at the stress and strain represented by any point within the main hysteresis loop without the manifestation of any time \#166 ; effect .
The width of this part LM on the natural scale would show as one-twentieth of one scale division .
The material may , from this point of view , be said to have slight permanent set , although there is no evidence of a time effect .
The narrower the hysteresis loop over the range of stress considered , the more perfectly elastic the material may said to be .
The initial behaviour of the specimen upon any test then depends upon the previous history .
Thus , for example , if the specimen be left in the state represented by the point M , the initial line will be along MNP ; but left in the state G , the initial line will be along GNP .
The effect of a slight overstrain of the material is shown in Fig. IV , which is again divided into two parts , and has the secondary figures exaggerated fifty times .
Before the material was overstrained , the loops GB\amp ; B\amp ; X'WA were taken .
The whole actual stress-strain curve ( ABCDEFGHKEFGLMANPQR ) is shown completely in the upper figure and contains loops of such magnitude as to be visible upon the normal scale .
In this test the amount of permanent overstrain AG involved at zero stress is small compared with the strain which the material can sustain without time effects .
In the loading , " creep"\#151 ; that is , slow increase of strain with time without alteration of load\#151 ; first occurred at C , 120 lb. load ( see Table III ) , and the main extension at D ( 150 lb. ) , while the strain continued to increase when the load was reduced to 140 lb. The loop EFGHKE shows no creep , but on continuing the curve to negative stress very distinct creep appears at L ( upper figure ) 60 lb. only .
An exaggerated curve e/ ghlcEfyi ( lower figure ) is drawn .
This is not closed as the loops in Fig. II were ; also E/ i^x lies within efg , or the second unloading curve lies within the first .
Secondary curves for later observations are also drawn .
These show that after overstraining the material to the slight amount shown on the main curve , the internal condition of the material is upset , and , at any rate for some time , no actual straight line exists .
At the load \#151 ; 60 lb. ( point L ) very slight creep occurs and the width rt(upper part of figure ) of the hysteresis loop SY for + 60 lb. is nearly as wide as the width xt ( lower part of figure ) of the loop BW for the \#177 ; 100 lb. cycle taken before the overstrain .
Similar curves for overstrain are given in Fig. Y , where a much larger amount of overstrain ( here negative ) is employed .
The natural curve starts from A ( \#151 ; 80 lb. ) and passes through B to C ( 60 lb. ) , returning through D to A , and then through E back to F , and again on through G to H( \#151 ; 145 lb. ) where the large overstrain ( HK ) occurs .
On the return , creep continued on the load being reduced to L ( \#151 ; 130 1b .
) , and the following stress-strain Messrs. J. J. Guest and F. C. Lea .
^"S3J45\pvii 4s j ?
d sqi^si - anfcuqi .
cpMncy ^3U|02 =*\lt ; U I - sqi -'Vu^Xjq p'ro-ja ^uS 3 \ % Torsional Hysteresis of Mild Steel .
325* ( J V CD x O X CM r- a !
X. 4i a o !
c red points on Fig. IV .
\_Toface p. 324 .
60 .
80 .
90 .
100 .
105 .
110 .
3-716 3 -694 \lt ; -- ( point W , Fig. 4 .
) 115 neg .
Torsional Hysteresis of Mild Steel .
\#166 ; \gt ; ** \#171 ; ngg--.^~Xly35~~S , 325- S 5 ?
2 2 S 3 Sr S Messrs. J. J. Guest and F. C. Lea .
line is decidedly curved .
Restoring creep began to occur at M ( \#151 ; 601b .
) , and is well marked at no load N , and at the succeeding positive loads .
After P ( + 60 lb. ) the load was reduced , a loop PQRSP being formed .
This and the succeeding loop are given in the secondary curve ( exaggerated 25 times ) pqrsi pi p2 qn r2 s2 p2 .
Thus after an overstrain HK a little less than the preparatory elastic strain , the stability of the material is upset , creep occurring at + 60 lb. load , and the width of the loop is very considerable .
In comparing the secondary loops of Fig. Y with those of Figs. Ill and IY it must be borne in mind that the magnification is only 25 as against 50 times .
j Effect of Rest on Overstrained Material .
The effect of rest upon the material is shown in Fig. YI , in which again the secondary curves are exaggerated 25 times .
Shortly after the overstrain of Fig. Y the loop , of which the secondary curve is ABCDEFG , Fig. YI , was taken , and this shows creep BC and OF at + 60 lb. loading .
After 10 days ' rest the specimen was re-tested and the secondary loop PQRSTY obtained ; no creep occurred at \#151 ; 80 lb. but a small amount ST occurred at + 80 lb. ( so that Y does not join up to P ) .
The width of the loop is reduced by this rest to less than half its previous amount .
The loops are sufficiently wide to be shown on the natural stress-strain curves .
The range of stress without creep is also considerably increased by this rest .
The natural curve taken soon after the overstrain is shown in the full lines and that after rest in dotted lines , while the secondary curves are drawn by alternate dot and dash , and by dashes respectively .
Effect of Heating Overstrained Material .
The effect of boiling ( at 100 'C .
) upon the material is shown in Fig. YII , in which the exaggeration is again 25 times ; 18 days after the previous overstraining , a test running between + 80 lb. gave the natural loop ABCD , with no creep , which is shown exaggerated at \#171 ; BcD and is to be compared with the loop of Fig. YI taken after a shorter rest , 10 days .
The test was continued\#151 ; showing creep first at E ( 100 lb.)\#151 ; in order to give the overstrain indicated by EFGG'HK ( mainly at 140 lb. ) when the load was reduced and a loop obtained as indicated by the line KLMNPQRSN .
To keep this figure within bounds G has been transferred to G ' , whence the curve continues along G'H'K .
In the loop NPQR8N creep occurs .
During reduction of load from + 145 lb. creep occurred at L ( + 80 lb. ) and continuously afterwards , at 60 lb. , at 40 lb. and lower loadings .
It also persists in the following loops , as is clearly shown in the secondary figure NpgRwN .
The specimen was then boiled in water for one hour and afterwards the loops ( shown in the figure on Torsional Hysteresis of Mild Steel .
sgi ?
ci \#166 ; anfauoi sqiXpMl 05 - qi I Messrs. J. J. Guest and F. C. Lea .
- gg2s| s |ntuqyq|^02 \#187 ; ( g| - ggi XmyVio p^- ] \#151 ; S___f t__11 o o o GO O Torsional Hysteresis of Mild Steel .
329 the right ) between 60 and 80 lb. were taken for comparison with the loops \#171 ; BcD\#171 ; and N^RmN respectively .
The loops obtained after boiling are very much narrower than those shown on the left of the figure , the width on the natural curve being covered by the width of the tapering line WS .
Compared with the loop taken after 18 days ' rest , the width of the loop Wa^S^UW is about two-thirds that of aBcDa .
The same loop is shown at BDAC in Fig. VIII but the magnification here is 100 .
After taking the loop BDACB , Fig. VIII , the curve is continued along B'CDEF , showing creep at C and a considerable permanent set at OF .
The unloading curve is FG-H and is followed by the loop HKLMK clearly seen on the natural curve .
As before , return creep took place on the unloading curve , the condition of the material having been upset .
The specimen was then tempered , or as it is sometimes termed normalised , by being maintained at 330 ' C. for one hour in an electrically heated furnace .
It was then re-tested .
In the natural curve obtained after heating all the observations are included in the width of the thickness of the taper line PX .
The secondary curves PQRSVP of the loop for + 80 lb. loading is drawn for comparison with the loop ABCD , Fig. VIII , obtained after boiling , its width being about three-quarters of the latter .
A second exaggerated loop TUVWT between + 40 lb. was then taken and the loading was afterwards continued as shown on the natural curve EX up to 140 lb. load , when a sudden yield XY occurred without signs of previous creep .
The secondary figure for this loading is shown at WTRZXA ' .
A very small overstrain was permitted and the load gradually reversed , giving the line YB'C'D'E ' .
Creep occurred at D ' ( \#151 ; 110 lb. ) and a yield at E ' \#151 ; 140 lb. loading .
In all the above cases considerable yield occurred at about 140 lb. ( + or \#151 ; ) and sometimes yielding continued after the load had been reduced , but after heating at 330 ' C. no creep took place before the yield at 140 lb. After overstraining followed by rest or boiling at 100 ' C. creep occurred at less than 140 lb. , so that normalising at the above temperature considerably extended the range in which the hysteresis loop is narrow and in which the material shows no sign of , creep .
It has not been thought necessary to include Tables from which all curves have been plotted , as the three given are quite representative of the results obtained during the tests . .
The particulars given in this paper are but a commencement of the researches it is proposed to carry out , but they clearly demonstrate the existence of a hysteresis loop at stresses considerably below the generally accepted elastic limits for mild steel , and they further show the beneficial effect of rest and of mild heat treatment .
Messrs. J. J. Guest and F. C. Lea .
S o -.gpVq\gt ; l o \lt ; ~\gt ; ~o o o 1 o Cvi O OQ \lt ; J3 O CM S S S S S S. Torsional Hysteresis of Mild . .
One of the authors* has shown that the elastic limit in compression , as measured by an ordinary extensometer , of cold drawn steel tubes is very considerably raised by tempering at temperatures from 300 ' to 550 ' C. During the severe treatment to which mild steel is subjected in cutting machinery the material is locally stressed so as to leave conditions of overstrain .
After overstrain the material gradually , with time , acquires a more " normal " state , but a more perfect state is rapidly produced by slight heat treatment as above described .
In several establishments manufacturing precision machinery the parts are allowed a considerable period of rest before the final operations are performed upon them , but one of the authors has used such heat treatment ( temperature of boiling water and oils ) for the past 15 years in the production of accurate machine details .
Another point that the experiments indicate is that the yield of a bar under torsion may be much more sudden than it is generally thought to be .
After tempering the specimen at 330 ' C. the tendency to creep at less torsion loads than 140 lb. was nearly eliminated ( entirely in the example of Fig. VIII ) , but at 140 lb. a remarkably sudden yield occurred .
The observation of this was very striking .
The load was applied to the arm by small instalments and slowly .
Upon reaching what proved to be the yield point load , for a short time no sign of yield occurred ; the specimen then began to give way slowly and soon the rate of yielding became rapid and the bar twisted through a considerable range .
The abruptness of yield appears , as is to be expected , to depend upon the uniformity of the material .
As the stress in a shaft under torsion reaches a maximum at the outer surface , the suddenness of this yield ( after normalising ) indicates that when the stress is close to the yield point the material is really unstable .
The yielding of the outside thus starts the yield of the adjacent material which is not so highly stressed and which thus apparently yields at a lower stress than that at which the outside material yielded .
If this were not so the yield of the outer material could only produce a small twist in the specimen as a whole .
Since fatigue effects depend upon the gradual increase of the width of the hysteresis loop with repetition , it would appear that boiling and tempering at comparatively low temperatures remove initial strains , and thus considerably increase the resistance of the steel to repetition stress .
Hence parts to be subjected to such stresses , and materials , such as tubes , subjected to cold drawing can , with advantage , be normalised .
One of the authorsf has for some years advocated the adoption of this treatment for such articles as crank shafts .
* 'Recent Researches on Tubular Steel Struts , ' + ' Proc. Inst. Automobile Engineers , ' vol. 6 p. 365 .
VOL. XCIII.\#151 ; A. 2 C Torsional Hysteresis oj Mild Steel .
One more point is worthy of note , After overstrain it has been stated that negative creep occurred at diminishing loads , or , in other words , as the loads were decreased the twist of the bar diminished while the load was steady at a given amount .
This is accounted for , if it be assumed that the stress near the surface of the bar is beyond the elastic limit , while nearer to the centre it is still within the elastic limit .
Hollow specimens of various kinds of steel are to be tested as soon as opportunity offers .
The results obtained in these preliminary experiments are not such as to allow of definite conclusions being drawn as to the behaviour of material in general , but there seems to be no doubt that , for the mild steel used in these tests , the torsion stress-strain , curve for a rising or falling load does not obey Hooke 's Law , and that for a given range of torsion , well below that which produces a stress equal to what is generally considered the elastic limit of a mild steel , whether wholly positive , negative , or changing sign , a hysteresis loop is obtained and the elastic limit as at present defined would appear to be zero .
An examination of the curves given in the paper shows that for any given value of the torque , the slope of the stress-strain curve , and thus the modulus of rigidity , depends upon the way the stress is changing , and at any given stress , including zero stress , is not a constant quantity , but depends upon the previous history of the material .
The thanks of the authors are due to Mr. M. Parks , Stud .
Mem. Inst. C.E. , for his skilled assistance in preparing the drawings from which the curves have been reproduced .

