]\gt ; An Determination of the Variation of Stress in a Thin Rectangular Plate Subjected to Shear .
By E. G. COKER , M.A. , D.Sc .
, Professor of Mechanical ineering in the City and Guilds of London Technical College , Finsbury .
( Communicated by Prof. Karl Pearson , F.R.S. Received December 28 , 1911 , \mdash ; Read February 8 , 1912 .
) Introduction.\mdash ; The determination of the distribution of shear stress in a body subjected to the action of an arbitrary set of forces applied to its bounding surfaces is often of great importance in constructional work .
In most cases mathematical difficulties do not allow of an exact solution , while the simplifying assumptions which are often made to reduce these difficulties lead to formulae which are probably not correct even to a first approximation .
It is important , therefore , to have experimental determinations to compare with the values calculated by approximate methods to determine to what extent the assumptions are correct , and also , if possible , to afford data for better approximations in cases which do not allow of mathematically exact solutions .
The shear stresses in riveted joints , the stiffened webs of plate girders and such like structures , afford examples of practical problems of extreme complexity from the mathematical standpoint , which are dealt with in practice by rough approximations , using large factors of safety fixed by experience of the behaviour of similar structures , in order to allow for contingencies due to ignorance of the actual stresses and accidental overloading .
The experimental determination of the state of stress in a body as inferred by measurements of the strains produced at its surface affords a means , which has been used by various investigators , to determine the condition of stress in a body subjected to the action of forces applied at its boundary .
In shear problems , with which the present paper deals , this ] involves the measurement of angular displacements , but these strains are so minute for the materials of construction at ordinary working stresses that it is usually found necessary to make experiments on other materials like indiarubber , plasticine and the like , which afford strains sufficiently large for accurate measurement , while their elastic properties are such that they may be considered to behave like more intractable materials with fair accuracy .
Another possible method of attack is to use a material possessing optical properties which depend on the applied stresses , and by measuring the change in optical property to infer the stress in the material .
For this Prof E. G. Coker .
Optical Determination of purpose glass possesses very many advantages , in that its action on light depends on the stress in the material and , to a first a linear law expresses the relation between the stress and the effect produced .
Glass can be obtained in a very perfectly elastic condition , and the strain relations resemble those of steel and wrought iron very Unfortunately , measurements of the optical effects produced usually stresses in the neighbourhood of fracture , and this , coupled difficulty of shaping the material into other than the simplest prevented its use becoming general for this kind of investigation .
At the present time other transparent bodies may be used which do offer this kind of objection , and in the investigation to be described commercial form of transparent nitro-cellulose is employed which has , a high degree , most of the desirable properties of glass and few of disadvantages .
If a beam of plane polarised light is passed through a plate of glass , or like transparent material , it is found that the relative retardation duced between the ordinary and extraordinary rays depends on difference between the principal stresses and , at any point , also on the thickness of the material .
We may express the relation the formula where is a constant .
This retardation is easily measurable , and the of may be inferred in a simple manner by a mechanical measure4 meant to be described .
The distribution of the shear stress in the cross section of a dam has , late years , been the subject of many experimental and theoretical tions , and one of the outstanding difficulties of this problem is the distribution of the shear stress in the horizontal secrions of a dam .
At the suggestion of Prof. Karl Pearson some experiments were made , the autumn of 1910 , on the related and more simple problem of distribution of shear stress in a rectangular plate , by an optical method being used by the author in another investigation .
These earlier experiments on a small model gave promise of accurate quantitative results with a large experimental apparatus , and , a preliminary to the more difficult problem of the shear stress in a an apparatus for experimental work was constructed on what appeared to a sufficiently large scale , and it was set up in the engineering of the City and Guilds of London Technical College , Finsbury .
1911 .
] of Stress in a Thin Rectangu lar Plate .
Description of the Apparatus the Method of Experiment.\mdash ; A small tension testing machine of the simple lever type was available for applying stress to a plate .
It was capable of exerting a maximum pull of and it had a clear gap between the end shackles of 20 inches .
In order to adapt the machine for the purpose of applying shear stresses a steel frame was constructed to give a symmetrical double shear to a plate of the transparent material .
The arrangement adopted is shown in the accompanying sketch ( fig. 1 ) , in which , are eye-bolts provided with nuts having spherical ends fitting into the shackles , of the testing machine .
When these latter are separated by the action of the machine the lin pull passes through the centres of the pins , to which the frame proper is secured .
This frame consists of pairs of vertical planished steel bars of 2 inches by 9/ 16 inch rectangular cross-section , the outer pairs being pin-connected above to a pair of cross- bars , 3 inches by 5/ 8 inch in cross-section , and below to a similar but lighter set K. Each gripping is slightly undercut for the greater part of its width , leaving faces on each side , so that when the plate to be stressed is placed in position and the bolts adjusted in the holes provided , the bars grip the plate along its whole length on each side .
The extent of the gripping surfaces is indicated by dotted lines in fig. 1 and by the accompanying cross-section .
The central bars are constructed in the same manner , and when a pull is applied by the testing machine the free parts , of the plate are subjected to a nearly pure shear .
The maximum length of plate which can be accommodated in the apparatus is 14 inches , and horizontal cross-bars are provided to permit free widths of*inoh , 1 inch , and 2 inches respectively .
The general arrangements of the testing machi1le and frame are shown in fig. 2 , in which a simple lever A pivoted at supports the stressing frame on a spherical bearing , and the lower shackle engages with a siniilar bearing .
The load is applied by a wire attached to a spring balance hanging from the outer end of the lever .
The wire passes over a pulley to a frame provided with a nut threaded on a screw , which latter is turned by a hand-wheel J. Prof E. G. Coker .
Optical of This arrangement has some advantages over an alternative employed at an early stage of using dead weights at the outer end of testing machine lever .
This necessitated the lifting and frequent of masses of as much as 150 lbs. weight in some experiments , involving much labour and waste of time .
The same load can be applied with ease by a few turns of the hand-wheel and changed for a new load a few seconds without shock or jar .
It is , however , open to the that a spring balance is hardly ever quite accurate , and that its vary slightly from time to time owing to changes in temperature and like .
These objections are to a great extent removed by careful and these were carried out at intervals and the observations corrected from .
The differences proved to be very small , and were practically negligible .
At the commencement of an experiment the machine balanced by adjusting the position of the weight on the screw load was put on the specimen by turning the hand-wheel until the balance registered a definite pull .
The distances between the on the main lever were such that the pull on the speoimen was 20 the pull shown on the balance , while the weight of the central pair of and shackles always added a load of 15 lbs. to the plate under shear , which due account was taken .
Arrangemxnt of the Optical Apparatus.\mdash ; In the earlier experiments consisted of a pair of Nicol prisms and suitable focussing lenses 1911 .
] iation of Stress in Thin lar Plate .
producing a parallel beam of light from an arc lamp .
The field of view with such an arrangement was very small , and with the apparatus available the maximum area under observation at any one time was practically a circle of about 1S inches diameter .
To examine the optical effects produced over a long length of plnte the apparatus was supported on an optical bench and carried by a rising and falling table , which latter could be adjusted by screws to any part of the plate under observation .
This caused much waste of time due to the frequent changes of position required and the numerous minor adjustments of the optical apparatus .
The arrangement was ultimately replaced by a much more convenient apparatus designed for me by Prof. Silvanus Thompson , , which enabled a plate 10 inches in length to be examined in detail without any change in the adjustments of the optical apparatus except an alteration in height of the Nicol prism used as an eyepiece .
This new apparatus consisted of a black glass plate 24 inches by 12 inches , set at a suitable angle for polarising the light obtained from a number of glow glamps , and diffused through tissue paper screens .
Quarter wave plates of mica were also constructed about one square foot in area for obtaining circularly polarised light .
This apparatus was set up in the neighbourhood of the stressed plate , and the colours due to double refraction were observed by the aid of a Nicol prism , which when set at close range enabled any point to be examined in detail , while the whole plate could be viewed from a distant position at one time .
The positions of any stress maxima or minima could therefore be approximately determined from the colours by a qualitative examination , and any stress variations due to imperfect adjustments could be localised .
For quantitative work it is most convenient to determine the stress produced by reference to a colour scale , and this may be either a calibrating member stressed in the same way as the plate under experiment until the colour effects are identical , or an optical wedge might be arranged to give a graduated retardation of any fraction of a wave-length of light to correspond to that given by the stressed plate .
In the present experiments , however , the most obvious method was to take advantage of the fact that a pure shear is equivalent to a pair of equal strains at angles and equally inclined to the direction of slide , one a stretch and the other a squeeze .
If a tension member cut from the same plate is stressed in the direction of the compression stress the retardation produced by the one is annulled by the other for a definite pull on the calibration member .
Let the tension member be subjected to a stress , then the strain in Prof. E. G. Coker .
Optical of the direction of its length is given by , where is modulus ; the lateral contraction produced in the breadth of the is , where is the stretch-squeeze ratio .
These strains are equivalent to a slide and an all round stretch point , and we may express the relations between them by from which we obtain and .
Now , since the shear stress , where is the coefficient of rigidity the material , we obtain or T. The shear at any point may therefore be determined from the stress in a tension member set along the direction of compression and stressed to give total extinction at the point considered .
The tension stress in the calibration member also produces a change in the thickness of the material of an amount expressed by Experiments*have shown that the value of Young 's modulus for this material is approximately 300,000 , measured in pounds and inches , and that is .
The change in thickness er 1000 lbs. of stress intensity is therefore .
The highest stress recorded in any experiment is 4730 lbs. per square inch , corresponding to a diminution of rather less than of the thickness of the material .
In the majority of the experiments the stress rarely exceeded .
per square inch , and the joint correction for the calibration member and the shear specimen is therefore so small that it is well within the limits of error of the determinations , and no corrections have been applied .
It is necessary to determine the directions of the principal stresses at points in the plate where measurements are to be taken , and these can be mapped with considerable accuracy if the plate is examined by aid of a pair of Nicol prisms , with their principal planes at right angles .
At any point where the directions of principal stress correspond to the principal planes of the crossed Nicol prisms no light can pass to the eye , and all points in the field of view for which this condition obtains are mapped by a black band or black area , which , in general , alters in position when the prisms are rotated .
An examination of the stressed plate shows that the directions principal stress are inclined at to the line of pull of the frame 'Phil .
Mag October , 1910 .
1911 .
] Variation of Stress in a Thin Plate .
at the ends , where the directions change somewhat rapidly .
The observed changes in angular position for one end of a plate having a length of 4 inches and a breadth of 1 inch are shown in fig. 3 , from which it will be observed that for the centre line all these variations occur within inch from the edge .
Hence , if measurements are made along the centre line of this plate the tension member must be set at an angle of to the line of pull for all positions not very close to the ends .
This is also evident from a map of the lines of principal stress obtained from the curves of observation , and shown by fig. 4 .
The breadth of the plate influences the distribution at the ends , and the disturbing effect upon the positions of the lines of principal stress appears to be approximately proportional to the breadth , as appears from fig. 5 , which shows the lines of equal inclination corresponding to a plate Prof. E. G. Coker .
Optical of 2 inches in breadth and 4 inches in length .
For this breadth , of the shear stress were taken along the central line , and also along parallel thereto , and inch away from the sides .
The lines of stress and equal inclination at one end of this plate are shown in fig. from which it appears that along the central line no correction is unless the distance is less than inch from one end , while along a parallel and inch away from the central line a correction will required for this width of plate if t'ne distance is less than inch one end .
For applying load to a calibrating member set along the direction compression stress in the plate a miniature lever testing machine constructed ( fig. 7 ) , so that the specimen could be ad.iusted in position .
This consists of a right-angled frame , supported on a swivel pin carried by a , which latter slides on a vertical rod D. Clamping screws are provided to fix the block in any position along the vertical rod and to adjust the angular position of the frame .
One end of the specimen is carried by a , and the other by a similar knife-edge , in a lever , which latter is pivoted about a knife-edge , secured in the frame .
A pull on the specimen is obtained by a spring balance , one end which is attached to the outer end of the main lever , and the other through a hole in the frame , and is secured by an adjusting nut regulating the pull .
1911 .
] Variation of Str.ess in a Thin lar In order to obtain a uniform tension stress in the specimen each end is drilled , and steel rings are inserted in the holes so made to receive the , which latter pass through the rings , and are supported on each side of the specimen by cheek plates .
The weight of the inclined lever and spring balance produces a total pull on the specimen of lbs. , and the lever multiplies the pull of the balance by five , so that if the dimensions of the cross-section of the caliration member are determined the stress on the material can be calculated .
In all cases the calibration specimen was cut from a part in close proximity to the plate used as the specimen under shear stress .
Method of Experiment.\mdash ; A plate of material , of approximately the required size , was first cut from the sheet , and examined for flaws between crossed Nicol prisms .
If it proved satisfactory , the plate was afterwards drilled for clamping in the frame , and finished to the required size .
It was then covered with a network of reference squares by marking with a fine needle point , and , before setting in place , it was again examined , to make sure that the shaping processes had caused no injury to the material .
With the specimen clamped in position , the maximum load was applied , and a further optical examination was made , to discover if there were any signs of local stresses due to unequal tightening of the bolts , or due to direct pressure of a bolt on circumference of the holes .
This examination was easily carried out by observing the appearance of the plate from a considerable distance by aid of a Nicol prism set to give total extinction at no load .
If the colours indicated local stresses , the gripping forces of the bolts were adjusted until - these colours disappeared and similarly disposed colour effects were observed on both free widths under shear .
The colour indications were very sensitive to the slight adjustments which were usually found necessary at the commeneement of each experiment , and the application of stress by the gripping surfaces was probably very uniform , and a close approximation to an evenly distributed shear stress along the vertical sides .
The central arrangement of the applied loads , and the construction of the stressing frame , ensured that both lengths of the plate were loaded in exactly the same manner .
The gripping surfaces were maintained at the same distance apart by large turned bolts , fitting exactly into holes in the cross-bars\mdash ; these holes were reamered to ensure an accurate fit , and were probably true to one thousandth of an inch .
During the experiments , no tilt of the frame was observed which would tend to make the outside vertical members approaob one another , and , as a large tilt would be required to oause a very small horizontal displacement , it was assumed that the arrangement of the frame effectually held the bars at the correct distance apart at all Prof. E. G. Coker .
Optical Determination of loads .
This is an important matter , because a central load applied frame produces shear in the material , and , if the outer plates are approach each other , an unbalanced bending moment is produced .
however , as in the present instance , the tendency to bend is resisted by frame , we obtain a practically pure stress in the material .
earlier experiments , the same result was obtained on a single plate held rigidly in a support , and loaded by a weight , by supplying opposing couple of amount .
This was accomplished by using a bell-crank lever EOE'D , pivoted at , and connected to the outer plates by equal links .
A pull at was furnished by a balance , which was adjusted to give a moment of the required about the fixed centre O. Observations of material in both frames gave similar indications , while effect of an unbalanced bending moment could also be observed in arrangement .
A point-to-point examination of the stress distribution made by aid of the calibration member , which was adjusted to any height , with its line of direction at an angle of to the frame .
in this tension member was first adjusted until the colour effect at the 1911 .
] Variation of Stress in Thin Rectangular under observation was extinguished and a dark field obtained .
The mean value obtained by first slightly passing the stress at which extinction was observed , and then afterwards lowering the stress until this position was again passed , was taken as the measure of the shear stress in the plate under ervation .
Analytical Theory.\mdash ; The mathematical theory of the distribution of stress due to shear in a rectangular section may be considered with reference to the case of a cantilever joist , fig. 9 , loaded by a weight at its outer end , the term joist denoting that the length is many times the depth , and that the thickness is small compared with the depth .
At any cross section distant from the origin there is a couple and a shear , both of which are balanced by the stresses and set up in the interior of the beam .
The couple can be balanced if the stresses where is the distance from the central plane , while the stress must -satisfy the general equations of equilibrium , , and the stresses being taken as averages throughout the thickness of the plate .
In addition , the sum of the shear stresses taken over the vertical section must have the same total magnitude as the shear load , and the value of must be zero at the upper and lower surfaces of the beam .
These conditions are fulfilled if provided that the terminal load is distributed over the end section according to this parabolic law .
In general the accurate distribution of the load over the terminal section is of no importance in practical applications to beams , since they are usually of considerable length compared with their Love , ' Mathematical Theory of Elasticity , ' p. 137 .
VOL. LXXXVI.\mdash ; A. Prof. .
G. Coker .
Optical Determination of [ Dec. depth , but it has often been assumed that the shear stress in any section is distributed according to this parabolic Iaw , even in such cases as that of the section of a reservoir wall or a dam , where the of the application of the load and the dimensions of the section are different from those assumed in the simple theory .
In the case of a plate of rectal ) gular section subjected to uniformly shear stresses over the edges , there appears to be no good for the assumption that the shear distribution follows a parabolic law , the width of the plate is comparable with the depth .
If the plate is long compared with its width , and the distribution is uniformly applied over the vertical edges , it would also appear from general considerations that the shear stress must be very uniform over a considerable length .
For if we imagine the vertical shear stresses equilibrated by shear stresses of the same intensity applied along the two horizontal ends , the condition of the plate becomes one of uniform shear throughout , and it seems improbable , if these latter applied forces be removed , that any considerable variation in shear stress can be caused thereby except near the ends .
Andrade*has considered the case of a rectangular block of length and breadth with the upper and lower faces subjected to uniform shifts in their own planes in opposite directions , all the remaining faces being free , and he obtains for the distribution of shear stress the form where A and are complex and the summation is for all values of ; and is the distance from the central plane of the block .
This gives maximum values near the ends and some maxima and minima of much less importance distributed along the central length .
Experimental Results.\mdash ; The general method of experiment having described , we now turn to the values obtained .
One of the early experi .
ments with the shear apparatus above was made on a plate which had thickness of 1/ 16 inch and a free width of 1 inch .
A length of 10 was chosen as the most convenient maximum , as all the experiments then be made without altering the adjustments of the polarising apparatu The plate was subjected to a pull of 100 lbs. per inch run , and of the testing machine showed that 99 lbs. per inch run was applied , the remaining 1 per cent. being accounted for by errors in balance and frictional resistance of knife-edges .
The weight of the ' ' The Distribution of Slide in a Right Six Face subject to Pure Shear ' Roy .
Proc , vol. 85 , 1911 .
1911 .
] of Stress in Thin Plate .
maximum value .
From this point there is a slight decline in the stress intensity until in the bourhood of the centre of the plate a minimum value is reached ; and , as we proceed in the same direction , there is a gradual recovery to a maximum value at a distance from the lower edge of the plate which corresponds very well with the position of the upper maximum , the stress finally falling to zero at the lower edge .
The general accuracy of the experimental work may be tested by a determination of the mean value of the tension stress from a summation of the area of this diagram .
If this value is compared with the mean intensity of the applied shear we ought to find that this latter is one-half the numerical value of the tensional stress , provided we may assume that the optical effect has a linear relation to both tension and shear .
In the present example this ratio was found to be In order to discover the effect of an alteration in the ratio of length to breadth , the same plate was cut to a length of 6 inches by removing equal portions from each end .
The applied load was maintained at nearly the same mean intensity , and readings similar to those of the previous experiment were obtained , as shown in the annexed table .
A slight increase of the shear stress was observable over the middle portion of the length , with maxima spaced nearly as before .
The ratio of tension to shear in this case was found to be .
In the same manner a length of 3 inches was also examined , which gave more pronounced maxima , nearly symmetrical ] spaced ; the ratio of the stresses was The shear distribution in the 2-inch length , however , underwent a it no longer had a double maximum , but increased in value from the to the centre .
The principal result of this change in the distribution was to raise the maximum stress by a considerable amount , as the diagram shows , and a still further reduction in length to 1 inch gave a similar distribution curve which approximated to a parabolic form with a further increase of stress intensity .
These experiments show fairly well the general features of the distribution , which are characterised by a fairly uniform distribution of stress over the greater part of a long length , rising slightly to a maximum near the ends and falling rapidly to zero at the edges .
As the tithe diminishes the maximum values approach one another , until they finally merge into a maximum value at the centre for some ratio of length to breadth , which the experiments show to be in the neighbourhood of two .
A measure of the accuracy of the values contained in Table I is afforded by the accompanying Table II , from which it will be observed that the percentage differences between the experimental and calculated values range Prof. E. G. Coker .
Optical of between and per cent. , and they indicate that the measured stresses are in defect , a result which may be due to inaccuracies in observations or imperfections in the material , or both .
These percent differences are given for each set of experimental values .
Experiments were also made on a similar plate 12 inch in free and having a length of 8 inches .
The loads imposed were slightly from those of the previous case , owing to the dead weight of the middle being allowed for in the application of the load , which latter was maintain at a uniform amount of 100 lbs. per inch run of plate , corresponding a mean stress of 1584 lbs. per square inch on the specimen when due is taken of friction and other errors .
The curves of equal inclination that the principal stresses were equally inclined to the central line distances up to within 1/ 16 inch from the ends .
This plate yielded which agree very well with those described above as shown by Table 1II the accompanying fig. 11 .
Commenclng with the full length , we have same characteristic distribution as in the former case , and this is repeat with a length of 4 inches , except that the intermediate stress is slightly than before .
Further reductions in the length of the specimen to 2 and inches respectively show somewhat similar stress distributions maxima closer ether and a general increase in numerical while for a length of 1 inch we obtain a single central maximum corresponding to a stress intensity in the calibration member of 4150 per square inch .
The final length of 12 inch , for which the plate a square form , also gives a very symmetrical shear curve with a very increase in the stress .
It is interesting to compare the stresses at the of this length with that on the 8-inch length under the same mean per unit of length .
The former gives a shear stress of .
per inch as compared with .
per square inch on the longer length or of 36 per cent. in the intensity of the stress .
Variation of Stress in a Thin lar Table As the fol.egoing experiments at a high working stress could not be completed to the buckling of the thin plate , it was considered advisable to study the stress distribution at low loads , and a sheet of transparent material was therefore obtained of the greatest thickness procurable , inch , and from this three plates were cut and ruled , as described above .
They were each 10 inches in length and of 12 inch , 1 inch , and 2 inches width respectively .
The applied load was in all cases proportional to the length of the plate , and for convenience it was fixed at 15 lbs. on the spring balance reading per inch length of plate .
Allowing for all error in defect of 1 per cent. , as already mentioned , this amounted to an actual load of lbs. per inch run .
The procedure was much the same as in the earlier experiments , but the -experience of former work showed , among other details , that it was of primary importance to have all the bolts very accurately adjusted to give equal intensities of grip on the plate to avoid local stresses and consequent error of measurement .
Although the applied stress was much lower than before , the increased thickness of plate multiplied the optical effect so considerably that no difficulty was experienced in making the required measurements .
The plate of 5 inch width , when under test , showed all the principal features brought out by previous experiments .
The numerical values obtained are shown in the accompanying Table , and these values are plotted in fig. 12 .
Prof. E. G. Coker .
Optical Determination of plate in all cases except that of the full length .
The change to the with a maximum at the centre is not complete for a length of 2 but becomes so for a length of inches , a result which agrees positions of the other maxima in all but the full length , but is not agreement with other determinations .
The minimum values are very defined , and are practically at the centre in every case where they This set of experiments shows a much closer approach to the standard , as will be observed by the results collected in Table which it appears that the percentage differences are with one very small and negative .
The experiments made on a plate of 2 inches width were directed to determine the stress distribution not only along the central section , but also at sections midway between this and the sides of the gripping bars .
The results obtained ( Table IX ) show that the distribution near the sides is very similar to that of the central section , for some distance remote from the ends , and in most cases the differences are within the limits of experimental error .
At the ends , however , there are considerable changes ( fig. 14 ) , especially as the length of the specimen is diminished , and the change in the directions of the lines of principal stress becomes an increasingly important factor .
Except close to the ends , however , the general features of the distribution are shown to remain unchanged ; figs. 6 and 7 indicate that the observations on each side of the central line require correction for distances within inch from the ends .
This slightly affects few values at a distance of S inch from the ends , where the lines of principal stress are inclined at approximately and to the central line .
Subsequen measurements on another specimen , when the calibration member was in accordance with the angles indicated above , showed that these correction amounted to approximately 4 per cent. , and the curves of shear stress 1911 .
] Variation of Stress in a Thin Plate .
observation and setting are somewhat variable , and are greater at longer lengths .
This latter circumstance may be partly accounted for by the want of uniformity in the application of ] at the ends of the full length .
The rapid change in the intensity of the shear at tho ends of the shear plates , and the apparent rity of these portions of the curves for different lengths , is one 01 considerable interest , and it is ) tant to enquire if the distributions { ollow the same law for all lengths .
methods of experiment adopted for the general snrvey of a long plate do not adapt themselves very well for a very detailed examination of small areas , but the ) ental values recorded afford some opportunity of .
the character of the end distril ) ution curves , and it appears that the observations for a plate in which ratio of length to readth is the hbourhood of unity follow a law approximately .
may be shown by comparing the ) rved values with the corresponding values of a trne parabola , and this has been carried out for a thick plate by applying Simpsoll'S rules to obtain the of the experimental shear curves .
this way the mean average values of the sheal stress been found along the central lines of } ) lates of square form , inch , 1 inch , and 2 inches side respectively .
values are mately 1413 lbs. , , and 1556 lbs. respectively , and correspond to parabolic distributions with maximum ordinates of ) average values .
True parabolic curves were plotted and compared with the experimental values , and one is shown in the accompanying fig. 11 from which it appears that the correspondence is a fairly close one .
The accompanying Table XI , showing approximately the percentage difference from a parabolic distribution , indicates a very fair agreement , with the exception of two observations at points close to the edges of the plate .
VOL. LXXXVI.\mdash ; A. 1911 .
] Variation of Stress in Thin Plate .
general distribution of the shear stress that there is any want of uniformity in the results .
It is shown that the shear stress in a long thin plate , subjected to a uniform shear load applied to the long parallel edges , rises rapidly from a zero value at each end to a maximum value , which latter is usually attained at a distance rather less than the free breadth of the plate ; the stress then decreases in value until it reaches a minimum at the centre .
As the length of the plate is diminished , the maximum and minimum stresses become more pronounced , and when the ratio of the length of the plate to the free width is in the neighbourhood of two the distribution changes in such a manner that there is a maximum at the centre and a rapid fall to the ends .
The experiments show that a parabolic distribution of shear is probably only true within narrow limits .
The distribution of shear stress over a long rectangular section may be nately represented by a lmiform shear over the central section with a rapid fall towards the ends .
For rough approximations the shear stress over a long ular section may be taken as uniform throughout .
The experiments appeal to show that the tion is of the sam cter for the same ratio of length to width .
In sion the author desires to express his warmest thanks to Prof. Karl Pearson , his valuable advice and assistance during the progress of the work , and also to Prof. Silvanus Thompson , F.B.S. , for much help in optical matters .

