]\gt ; The Distribution of Slide in a Right Six-face subject to Pure Shear .
By E. N. C. ANDBADE , B.Sc. , Jessel Scholar of the University of London .
( Communicated by Prof. Karl Pearson , F.R.S. Received October 14 , 1910 revised May 28 , 1911 , \mdash ; Read June 29 , 1911 .
) Introductoroy .
The question of the exact distribution of slide in the cross-section of a body subjected to transverse force has recently attracted considerable attention in connection with the much-discussed problem of the stability of dams .
* The experiments to be described were undertaken with a view to ascertaining this distribution in the simple case of a right six-face , one face of which is everywhere fastened to a fixed rigid plane , while the opposite face , also fastened to a rigid plane , is slid parallel to itself , all the other faces being free .
The slide was measured over a free face parallel to the direction of traction by an optical method described in the expenmental section .
Engineering practice assumes the distribution to be parabolic , but this is generally discredited .
The experimental work of Pearson and Pollard indicated that when the shear , was due to water pressure only , the distribution over a horizontal section had two maxima , and a minimum rather than a maximum at the centre .
The experiments of Wilson and indicated a double maximum with certain forced distribution of shear over the base , but not with others .
Hence it appeared desirable to test the matter de with a simpler load than occurs in a model dam , and a simpler contour ; the rectangular contour was adopted as of considerable bheoretical interest , and possibly amenable to detailed mathematical treatment .
The result has been to obtain experimentally curves which leave no doubt as to the real existence of a double maximum distribution of slide ; the double maximum is very marked over the central section , but , as would be expected , less strongly marked over the sections nearer to the faces constrained rennain plane .
The curves resemble somewhat those obtained theoretically by FilonS for a different system of loading , viz. , a bar of rectangular cross* See , e.g. , Atcherly and Pearson , " " On Some Disregarded Points in the Stability of Masonry Dams\ldquo ; ; and Pearson and Pollard , " " An Study of the Stresses in Masonry Dams ' Drapers ' Company Research Memoirs , ' Technical , II and .
See particularly dotted curves on Plate 8 .
Wilson and Gore , ' Institution of Civil Engineers , ' vol. 172 .
S L. N. G. Filon , ' Phil. Trans 1903 , A 334 .
See p. 126 .
Distribution of Slide in Right Six-face subject to Shear .
449 section , pressed on two parallel sides by two knife-edges not opposite to one another , and equilibrated by two shears over the free ends , which are at distances such that the principle of equipollent loading applies .
An attempt was made to see whether the solution of an analogous analytical problem could be approximately represented by using functions such as those indicated by Filon*and Dougall , depending on roots of the transcendental equation .
A number of such solutions were taken , and the constants so chosen that the shift conditions were satisfied at a number of isolated points .
The results are shown in fig. 5 , and give variations of slide of the same general nature as those found ] but the lack of closer agreement indicates that such solutions are probably inadequate to the complete solution of the problem .
A discussion of the difficulties connected with this mode of representation will be found in the analytical section of the paper .
Bxperimental Method .
To measure the slide , the optical method described in the memoir on the stresses in masonry dams by Karl Pearson and A. F. C. Pollald , already quoted , was employed .
The right six-face , whose approximate dimensions were 16 in .
in .
in .
, was cast out of the gelatine-glycerine jelly of which the model dams were made , and two of the larger parallel faces fixed to boards by means of fine copper gauze nailed to the wood and heated before the application of the jelly .
into one of the large free faces , fine needles .
were then inserted in rows 1 cm .
apart , the needles in each row being 1 cm .
apart ; this was effected by laying a sheet of squared paper on the jelly , through which the needles were pushed , after which the paper was removed .
The lower half of the face was thus prepared with five rows of needles , enabling measurements to be made along four lines .
The microgoniometer employed was the one fully described in the above memoir , used in conjunction with a transparent vertical scale , a straight portion of the filament .
of an ordinary incandescent electric lamp being used as the source .
One of the boards was firmly fixed to the table , while a horizontal pull could be applied to the upper one by a cord running over a pulley .
So that the six-face might be subjected to pure shear , without bending moment , it was arranged that a vertical pull could be applied to one end of the upper horizontal face by a weight attached to a cord passing over a pulley ; this , * L. N. G. Filon , " " On the Expansion of Polynomials in Series of Functions ' Lond. Math. Soc. Proc 1906 , Ser. 2 , vol. 4 , Parts 5 and 6 .
J. Dougall , " " An Analytical Theory of the Equilibrium of an Isotropic Elastic Plate ' Trans. Roy .
Soc. Edin 1904 , vol. 41 , Part 1 , No. 8 .
Mr. E. N. da C. Andrade .
The Distribution of [ May 28 , together with an equal weight placed resting on the other end of the face , gave a couple adjusted to be equal in magnitude to the shearing couple , and opposite in sign , so that there was no bending .
Preliminary experiments were made on the time-yield and recovery of the jelly under different loads , with the object of the greatest load with which complete recovery could be obtained in a reasonable time .
The goniometer was placed in position near the middle of the side and the block strained for some minutes and then unloaded , the scale readings being taken during straining and recovery .
With loads of .
and .
there was a permanent set ; with .
load complete recovery could be obtained , and the increase of slide during the second minute after was only about of the total slide .
The shift of the upper face due to this load was about mm. , as measured directly with a micrometer .
The load adopted was therefore 4 lbs. , with a corresponding couple to avoid bending .
The microgoniometer being placed on three needles at a fiven place , this load was applied for a minute , readings being taken at 15 , 30 , and 45 seconds after loading , and at intervals of a minute for 5 minutes after the removal of the load .
Complete recovery was seldom attained at the first series of readings on any particular set of needles , but generally with a second set of readings , the oniometer not being touched in between , recovery was obtained within the five minutes .
For a given position the displacements of the first loading were probably necessary to make the microgoniometer settle down to a perfectly stable position on the needles ; at any rate , two series of readings were taken at each place , the second being in general reliable .
Occasionally a third set had to be taken ; the criterion of reliability adopted was complete recovery within the five minutes .
The microgoniometer was placed successively on the different needles of the same horizontal row , and readings corresponding to both mirrors of the instrument taken in the manner described , one of the mirrors , of course , being very little deflected compared with the other .
There were 36 and 37 needles respectively in alternate rows , and as preliminary experiments had shown that the readings varied from day to day at a given place ( probably mainly owing to temperature , though hygrometric conditions may have something to do with it ) it was judged desirable to do a complete row in one day , when , as a rule , readings taken for the same place at different hours agreed .
To give the angle of slide , these readings had to be cted for the obliquity of the ' horizontal\ldquo ; mirror ray by means of the formula given on p. 27 of the Pearson and Pollard memoir .
When this had been done , the 1911 .
] Slide in Six-face subject to Pure Shear .
451 slide for the four different sections considered experimentally was plotted against the corresponding distance along the -axis .
Experimental Results .
Within the limit of experimental error the curves were found to be symmetrical about a vertical through the mid-point of the side , as might be expected ; only half of each curve is therefore shown .
It is seen at once that the general result has been to demonstrate clearly that the shear curves have a double maximum , with a minimum at the centre , as the work of Pearson and Pollard seemed to indicate .
The area of the curve , which boives a measure of the load , is not quite the same at all sections , as the elastic constants must vary with the temperature , and the different curves were btained on different days .
the greatest deflection was under 1 cm .
on the scale , so , since the fractions of a millimetre were estimated by eye , an experimental error of at least 2 per cent. is probable .
Small local variations may also be caused by the inevitable slight irregularities where the jelly was fastened to the boards , though it was trimmed as well as possible before experiments were made .
The thickness of the block was about 9 cm .
vertically .
The experimental measurements give roughly the mean slide over a strip of the face 1 cm .
thick , and we take this as the slide corresponding to the middle line of the strip .
Then fig. 1 gives the distribution of slide along a line distant cm .
from lower fixed face , and figs. 2 , 3 , and 4 the slide lines distant , and cm .
respectively .
It will be observed that the central minimum is most marked over the mid-line , and that as we approach the rigidly fixed face we get , as we should expect , an approach to a uniform slide across the greater portion of the section .
An important point is that secondary maxima are indicated , notably in figs. 1 and 3 .
FIG. 1 .
1911 .
] Slide a Right Six-face subject to Pure Angle Shear for the Four Sections .
' The positions are read from the end of the row of needles : 1 corresponds to a distance of about and 2 cm .
from the free edge respectively for alternate rows .
Analytical Treatment of an Amxlogous The following analytical attempt is concerned with the problem of an elastic right six-face , two parallel faces of which are given umform shifts in their own planes in opposite directions , the remaining faces free .
simplicity it has been necessary to take for the displacements and shifts their average values throughout the finite thickness of the block , in which case it is known*that the problem reduces to that of uniplanar stress , i.e. the problem of an indefinitely thin plate .
The notation used is that of Pearson and Todhunter 's ' History of Elasticity .
' Let the block be of length , depth , with axes taken the mid-point .
We then have when ; when , say ; \mdash ; when In Love 's ' Treatise on Elasticity ' it is shown that for eneralised plane stress the stress can be expressed in terms of a stress-function , the formulae for the stresses * L. N. G. Filon , 'Phil .
Trans loc. .
Also Love 's ' Treatise , ' Chap. IX .
Love 's ' Treatise , ' Chap. IX . .
LXXXV.\mdash ; A. 2 I Mr. E. N. da C. Andrade .
The Distribution of [ May 28 , and for the displacements where and denotes , , and with this notation must satisfy the equation .
( 1 ) The conditions of the problem make an even function of and an uneven function of ; we assume , therefore , taking the sine in which is a known solution of ( 1 ) .
This gives , Bvhence we find where 's ratio .
, ( 2 ) ny .
For these stresses to vanish at , we have from ( 2 ) and , substituting in ( 3 ) , or 2 .
( 4 ) This is the transcendental equation for given by Filon .
* , or if nx\mdash ; ny , ( 5 ) L. N. G. Filon , ' Lond. Math. Soc. Proc 1911 .
] Slide in a Right Six-face subject to Pure Shear .
ny .
( 7 ) Since , determined from ( 4 ) , is complex , we must rationalise these values of , and for numerical interpretation .
Lt 2 , and take ( 5 ) , ( 6 ) , and ( 7 ) finally give , as rationalised values for , and where first expression in brackets second expression in brackets the suffix 1 indicates that the values are those for the first root of the equation and the summation is to be taken for all roots of this equation .
, ( 9 ) Mr. E. N. da C. Andrade .
The Distribution of [ May 28 , where first exp1ession in brackets second expression in brackets - where first expression in brackets second expression in brackets ( 8 ) and ( 9 ) are ions for and which satisfy the equations of elasticity and make over the free ends of the contour .
For numerical interpretation of these results , and were made to have given values at isolated points , thus giving equations for , , and the of these constants obtained by elimination were substituted in , thus giving as a determined function of and .
It was not found feasible to work with more than three roots of the transcendental equation ; hence there are the six constants , , , determined .
The system finally adopted was to add to a parabolic term of the usual form and corresponding terms to and ; the expression already given for is then of a corrective nature .
was made equal to the constant on the face at the three points , and at the same three points .
As is identically at , we have an extra condition , which we make at .
( The symmetry of the equations fixes and at the corresponding points on the negative side , and on the face ) The ordinary parabolic term is , where is the displacement of the rigid face .
The values of and which correspond to this are obtained from the equation , 1911 .
] Slide in a Right subject to Pure Shearand are ?
This solution makes everywhere .
We have also for the root , a telm where can be chosen at pleasure .
Hence , if 2 denote the sum for the first three roots other than zero , putting in the values , and we have , ( 11 ) .
( 12 ) To find the roots of the equation , we have , putting 2 , the two equations , ( 13 ) .
( 14 ) If is a solution of ( 4 ) , so also are , but these do not lead to any new expressions for typical terms in and .
The approximate expressions given by Filon*for the roots of large modulus are not here applicable , as for the first three roots these expressions are very rough , giving values differing from the true ones by from 5 to 1 per cent. for The distribution of the roots is seen by considering the intersections of the curves sech 'Lond .
Math. Soc. Proc Mr. E. N. da C. Andrade .
The Distribution of [ May 28 , consists of a double branch between , with a node at the origin ; as increases positively or negatively , there are two symmetrical values of approaching .
There are no values of between and and , etc. ; between and ws have two -shaped branches , the distance , , between the two branches being .
Between and and , etc. , we have a series of -shaped branches , the distance between corresponding branches always increasing as we recede from the -axis .
The diagram is repeated for negative values of gives between a curve with two branches , having a node at the and asymptotic to .
Between and and , etc. , the curve does not exist ; between and and , etc. , the branches between are repeated , and similarly for negative values of Inspection of the curves shows at once that there is no intersection , except at the origin , between , and that the roots between and are those of least modulus , the modulus of succeeding roots always increasing as we pass out from the The values of and for the first three roots were therefore found as those of smallest modulus .
After having obtained approximate values by trial and error , final corrections were made from the following formulae , given by simple expansion .
If , where and are small , , The following are the values of the roots:\mdash ; Roots of the equation The functions , , , were calculated for , and for .
The hyperbolic functions were built up from and , which were worked out by logarithms .
The complete expressions for and were then computed , and the expressions for at the three points equated to , the expressions for at and , and for at equated to zero .
From these six ' equations the constants , were then obtained by elimination , the laborious arithmetic being done on a " " Brunsviga\ldquo ; machine .
The functions 11 .
] Slide in Right subject to Pure , having been evaluated for various values of , curves of along the sections , were plotted : they shown in fig. 5 .
FIG. 5 .
As will be seen , they show very markedly the double-headed character found experimentally , but otherwise they do not correspond very closely to experiment .
The maxima are more pronounced , and , the secondary maxima experimentally indicated are shown in Curve ii , the relative nitudes of the primary and secondary maxima rather from the experimental case .
The imperfect correspondence may be due partly to weakness of the analytical treatment , partly to the fact that the experimental and analytically assumed conditions are not identical .
In the analytical solution we assume the mean value of to be zero , and if we integrate the expression for , we have , where , are the perpendicular shifts at the plane faces considered , and , are the mean shifts in the plane faces .
This can , of course , be satisfied at the free faces , but it is not consistent with the constraint at the faces , where we must have , but does not vanish .
There seems no experimental possibility of giving freedom to the face shifts on the terminal planes Mr. E. N. da C. Andrade .
The Distribution of [ May 28 ; Again in the analysis form of expansion has been obtained , and it seems possible that one condition say may completely determine all the constants .
Filon*seems to suggest that one such condition accounts for the coefficients corresponding to all the roots of the transcendental equation , and in this case , if we take the constants up to , we can make the first terms of the expansion agree with each one condition at points , this is forcing the expansion to satisfy an unnatural condition .
The divergencies at points other than those selected restraint will then probably become worse and worse as we take more points , instead of , as we hoped , becoming less .
The arithmetic carried out in attempting to put into numbers forms of solution other than those given here seemed unfortunately to lend colour to this view .
In a paper published when the above work was nearly completed , F. Richardson has obtained by finite differences an arithmetical solution of the distribution of shear over the base of a dam of complicated contour .
The double maximum is indicated , but the maxima are caused by the sharp corners where the dam joins the base , where theoretically the shear is infinite .
Within the base of the dam he gives a distribution of shear falling steadily to a minimum at about a quarter span , which differs from most previous experimental work .
The boundary taken offers , of course , little resemblance to that here treated .
Summary of Results .
The main results of the work are embodied in the following summary:\mdash ; ( 1 ) The parabolic distribution hitherto assumed does not give even an approximately correct representation .
Both experimentally and theoretically the distribution of shear is given by a curve having two maxima at about one-sixth span from the end and a minimum at the mid-point .
For practical purposes a shear of twice the mean shear at about one-sixth and five-sixths of the breadth at the mid'Lond .
Math. Soc. Proc A solution was obtained making , everywhere along the rigid faces , but leaving , undetermined at .
Equations making at isolated points gave the constants , and was then plotted for different sections .
But the curves gave unequal areas over the different sections , showing that there must have been large shears on the free faces between the points at which they were fixed to be zero . .
F. Richardsou , " " The Approximate Arithmetical Solution by Finit Differences of Physical Problems , etc 'Phil .
Trans 1910 , A 467 .
( For the curves referred to , see p. 349 .
) 1911 .
] Slide in Right Six-face subject to Pure section may probably be taken as the safe limit to maximum shear in the case of blocks of approximately the present ratio of length and breadth .
( 3 ) Secondary maxima are indicated both experimentally and analytically .
Such secondary maxima in the shear across the base of their dam are indicated in the work of Pearson and Pollard already referred to .
( 4 ) The failure of the parabolic distribution of shear , which involves and is involved by the linear distribution of normal stresses , shows how inadequate ordinary engineering theory of both dams and abutments is to meet the requiremehts , experimental as well as theoretical , of the science of elasticity .
( 5 ) While solutions in functions depending on the roots of the transcendental equation can be made to yield numerical results in some degree corresponding with experiment , it seems at present doubtful whether they are adapted to five a complete and satisfactory solution of the \mdash ; problem when both and must have known values over two opposite faoes .
The work was carried out in the Department of Applied Mathematics of University College , London , at the suggestion of Prof. Karl Pearson , F.R.S. , for whose interest , encouragement , and constant assistance the author returns his warmest thanks .

