XV .
On the mathematical theory of suspension bridges , with tables for facilitating their construction .
By DAVIES GILBERT , Esq. F. P. R. S. &amp;c. Communicated March 9 , 1826 .
Read March 9 , 1826 .
MNY attention was first directed to a consideration of suspension bridges , and of the catenary curve on which their theory depends , when the plan for making such a communication across the Menai Straits was submitted to the Commissioners appointed by Parliament to improve the communication by roads and bridges through Wales .
It then appeared to me , that the proposed depth of curvature , was not sufficient for ensuring such a degree of strength and permanence as would be consistent with the due execution of a great national work .
This opinion I advanced as a Member of the Commission .
But wishing to take on myself the full responsibility for such increased expense , as must of necessity be occasioned by enlarging the curvature , I also printed some approximations hastily deduced , in the Quarterly Journal of Science and derived from them a confirmation of the opinion that had been given .
The interval between the points of support and the road-way of the Menai Bridge has in consequence been augmented to fifty feet ; and it now possesses that full measure of strength , which experience has established as requisite and sufficient for works of iron not perfectly at rest .
Since bridges of suspension are obviously adapted to very general use , I have flattered myself with the hope of doing something serviceable to the public , by expanding into tables mathematical theory , &amp;c. 203 the formula from which my approximations were derived ; adding to them other formulae and tables for the catenary of equal strength .
A curve not merely of speculative curiosity , but of practical use , where a wide horizontal extent may chance to be combined with natural facilities for obtaining a correspondent height for the attachments .
Both the ordinary catenaries , and these of equal strength , like circles , parabolas , logarithmic curves , &amp;c. have the property of being each identical with themselves in every respect but size : and as the radius , the parameter , and the subtangent give the respective magnitudes of these curves , so are the catenaries determined in magnitude by the tension ( expressed in measures of the chain ) , which takes place at the middle point , or apex of the curve , where it is a minimum .
Consequently , when this tension is determined or given , all the other relations may be expressed in the same manner as sines , cosines , &amp;c. in the circle .
I assume that the first principles of the catenary curve are known ; they will , consequently , be noted with no other view , than to derive from them ulterior properties .
For the ordinary catenary : Let a the tension at the apex , estimated in measures of the chain x the absciss , the versed sine , or depth of curvature ; y the ordinate , or semi-transverse length ; z= the length of the curve .
Then since the tension , ( a , ) acts horizontally at the apex ( A , ) since the weight of the chain ( z ) acts at right angles to the former , and the force of suspension at ( P ) acts in the direction of the tangent .
These forces must be represented in direcT , the tension at P , being obviously equal to Vaz + Zi , is equal ( Eq. A. No. 2 .
) to V za +a 42 X. The angle of suspension is derived from the common analogy of the incremental triangle , and of the forces corresponding with it .
Tables I. and II .
are constructed from these theorems , and their use will be best explained by an example .
Let the span proposed for a suspension bridge be 800 feet , and let the adjunct weight of suspension rods , road-way , &amp;c. be taken at one-half of the weight of the chains ; then , if the full tenacity of iron is represented by the modulus of 14800 feet , the virtual modulus for the whole weight must be reduced in the proportion of 2+ 1 : 2 , or to 9867 feet ; and let it be determined to load the chains at the point of their greatest strain , that is at the points of suspensions , , with onesixth part of the weight they are theoretically capable of sustaining .
Then since the semi-span is 400 feet , and y in Table I. is taken at an hundred measures , each of these measures must be four feet , and the weight expressed in the same measures to be sustained at the points of suspension will be 9867 +6x4 41 1,12 , .
Now it appears from Table I. where y is uniformly an hundred , that when T-412 a -400 measures or 16oo feet ?
x= i2.565 50.260 Z= 101.045 404.180 &lt; the angle of suspension 750 49 ' .
Having now determined a , the modulus , latus rectum .
, or parameter of the curve .
In Table II .
will be found all the respective quantities for each measure of y. But as a is in this table taken at an hundred measures , and it has been found to be 400 of .
the former , each measure here must be 4 -times 4 , or i6 feet ; consequently , each gradation of y will also be i6 feet , and the whole semi-span 4-l ?
or 25 measures .
And since z will be given in the Table for each measure of y , the adjunct weights may readily be adapted to a strict preservation of the catenary form .
At 21 measures of y. z= 21.1547 20 measures of y. z 20.1335 1.0212 X 16 -16.3392 feet .
Consequently while the ordinate extends one measure , or 16 feet from the 20th to the 21st measure , the length of the curve will increase 16 feet and .
very nearly , and the adjunct weight should be increased in the same proportion .
At 21 the length of x is 2,2131 measures , or multiplied by 16 = 35,4096 feet , the length of the suspension rods to the level of the apex .
It appears from Table I. that the tension T for a given half span of 0oo measures is very nearly at .
its minimum when x 65,85 measures , almost one-third part of the whole span .
In the example taken above 65.85 x4 =.263,4 feet , an height not to be attained in practice , nor strictly upplicable if it could be reached , because of the great length of suspension .
If the span and height ( 2y and x ) were given , the other quantities would be found in a similar manner .
In the catenary of equal strength a. x. y. z remain as before ; but another symbol must now he introduced , C= the mass of the chain .
Then will the forces be represented as in the ordinary curve by the incremental triangle P rp .
But now kc : : : : a. And by a repetition of the former steps V+I / a 2 +C27 But on the principle of equal strength , As a : Va2 + lz : : z : ( therefore x=ax / aand ia2 +C Va % + CZ + Equ .
D.)z= ax not .log .
aa and by substituting ax -for in the equation , =a x , +5 , consequently , a ? .
+ Cz ~~ '~+ Equ .
E. ) x= x neat .
log .
aC Again , from the first analogy , =z substitute for X its equal axA and j=a x ; therefore , Equ .
F. ) y= the cir .
arc .
of which C is the tangent to radius a. a and y being given to find A Multiply hby 57,029578 ( the tab .
log .
1.7581226 ) and reduce the decimals of a degree into minutes and seconds ; then will the tangent of that arc , multiplied by a , be equal to A And when ( has been determined , the other columns of Tables III .
and IV .
are constructed from the above theorems , in a manner perfectly similar to that used in calculating of Tables I. and II .
; and they may be illustrated by the same example ; observing that a , now represents the uniform tension on each given magnitude of iron throughout the chains , and that the column T has the whole pull which any building or support may have to sustain in the direction of the tangent .
In Table III .
y being , as before , an hundred measures of four feet each , a must be sought 411,125 , and by proportioning between 420 and 400 X= 12.2904 49.16161 Z= 101.0020 404.0080 t. measures or l Feet .
= 102.0235 408.0940 T= 423.60wj 1694.4o76j &lt; .
760 3 ' 17 ' .
a , or the modulus of this curve being fixed at 411,125 measures of 4 feet each , or at 1644,5 feet ; and a in Table IV .
being taken at ioo measures , each one will be 16,445 feet , and all the quantities are given for each gradation of y. Thus at 21 measures of y. Z 21.1564 = 21.3142 20 measures of y. z 20.1347 C-20.2710 1,0217 1.04332 1.0217 x 16.445 16.8019 feet the increase of z ; 1.0432 X 16.445 = 17.1410 feet the increase of material in ( : consequently XO =0437 1021 , the quantity of matter in this conseque 1.0217 part of the chain to maintain uniform strength , that at the apex being unity , and the ' adjunct matter should be in the proportion of i to 1,04332 .
Moreover x the versed sine , or the length of the suspension rods to the level of the apex will be at 21 measures of y. x= 2.2214 measures x 16,445=36,553 ft. 2o measures of y. x 2.0135 measures X 16.445 33.112 ft. Assuming in the ordinary catenary that = 65.85 measures , is the height of the attachment to give a maximum extent of span with any virtual tenacity of material , a will be 85 measures , and a+ ? ?
8 465.85 , or 15o,85 measures equal the given virtual tenacity .
This taken as before at ?
of O of 14800 feet , will give I0,87j feet for each mea3 6 sure , and the whole span at 25-1= 21 75 feet .
Chains merely supporting themselves , and at the utmost of their tenacity will extend nine times further , or to 19575 feet .
In the catenary of equal strength , the semi-span being equal to the circular arc of which is the tangent to radius a , it is obvious that ax semi-cir .
arc must be the limit of the span .
Therefore if a= 3-of I of 14800 feet , or 1644,44 36a 5154 feet .
And if the chains merely sustain themselves at their utmost tenacity , 5154 x9 will give 46385 feet , equal to 8,785 miles , or somewhat more than 8 miles and three-quarters .
But this case is purely hypothetical , for the purpose of ascertaining a limit , since 4 ; the mass or weight of the chain must be infinite , and consequently its length : the figure approaching indefinitely near to that of a chain sustaining itself from an infinite height , which figure is identical with that of a building , capable so far as pressure and the strength of materials are alone concerned of being carried to any elevation whatever .
This figure is readily determined Let a= the section of such a building at its base , y= the section at any height , Xthat height ; Then , since the ; section and the superincumbent pressure must always be in the same proportion to each other , X and are in a constant ratio .
Let then x =_ 3S -Y where m is the modulus of pressure in the given material x but when x= o , y = a , therefore .
' the not .
log. .
; X ~~~~~~a or A.mi the tab .
Slog ' , Y A =2,3025851 ; but if e and v the homologous sides or diameters of these sections ; then , X -tab .
log .
Finally , I would notice a correction of frequent use in practical surveying , to be deduced from the properties of the catenary curve .
When the measuring chain is extended over ground un-t even , intersected by ditches , or made soft by water , it cannot be laid flat , but must be elevated at both its extremities while the middle just touches the surface : thus giving the measurement too great by the difference between the whole perifery and the double ordinate .
Let z= the half length of the chain .
x the elevation at each end equal to the depths of curvature .
Then Equ .
B. No. 2 .
y=ax not .
log .
+ Ij , And Equ .
A. No. 3 .
a=t therefore zx : .
g%_ $2 vX Z+X2X not .
log .
Zj .
But when x is very small in comparison of z , the not .
log .
of X+$ becomes and Z-At z 2x th is therefore the difference between half the chain and the ordinate .
If x be expressed in parts of the whole chain , 4 xwill be the correction for the difference between the perifery and double ordinate .
If x ( the elevation at each end be one link of the common measuring chain , 4 xiof a link , of .
A6 .
of a foot 25 25 100 _ o.3168 of an inch , varying as the squares of a &gt; .
If half the chain were considered as a straight line , and the hypothenuse of a right angled triangle , the horizontal distance would be z 2x giving but one half of the true difference , 0,1584 parts of an inch .
And if the chain were supposed to be in the arc of a circle , ZBy + L , &amp;c. Andy-= Va4ax-e ( when x is very 6 az ' small in comparison with a ) = 2a Therefore = Ye And since y is also small in comparison of a , the second term of the series 6 will be the difference between ( 6 a " the ordinate and the arc .
Then substituting 4 for an zX~~~~~~~~~4 -6 0= 3 ; or if x be expressed in parts of the whole chain , 83 will be the whole correction , = 0.2112 parts of an 3 inchl- , or two-thirds of the true difference , Formulae might readily be constructed for different elevations of the extremities of the chain , but they would prove much too complicated for practical use .
One further observation may be applicable to suspension bridges , wholly unconnected with the preceding investigations .
In the event of their wanting stability to counteract and restrain undulatory motion , the balustrades may be carried

