RP-S6-4 N um ber of Active Coils in Helical Springs By R. F. VOGT,1 MILWAUKEE, WIS. D u e to th e to r s io n a l d is p la c e m e n t b e tw e e n c ro s s -s e c ­ tio n s o f t h e h e lic a l s p r in g b a r o r w ire i n t h e s o - c a lle d d e a d o r in a c tiv e c o ils o n e a c h e n d o f t h e s p r in g , t h e t o t a l d e ­ fle c tio n o f th e s p rin g is g r e a te r t h a n t h a t w h ic h c o r r e ­ s p o n d s to th e d e fle c tio n o f t h e fre e c o ils . T h e e x a c t a m o u n t o f th e s p rin g d e fle c tio n c o rr e s p o n d in g to th e in flu e n c e o f th e “ in a c tiv e ” c o ils c a n b e c a lc u la te d fo r k n o w n lo a d in g c o n d itio n s o f t h e s p r in g a n d c a n b e c lo s e ly e s tim a te d fo r a ll s p rin g s s u b je c te d to c o n v e n tio n a l s p rin g p r a c tis e lo a d s . T h e d e fle c tio n s o f t h e e n d c o ils a r e c a lc u ­ la te d in te r m s o f th e d e fle c tio n o f a c tiv e c o ils . A te s t to d e te r m in e t h e a c tu a l n u m b e r o f a c tiv e c o ils is s u g g e s te d a n d e x a m p le s a r e g iv e n . where ju is Poisson’s ratio. for spring steel E m m G HE predetermination of the correct number of active coils in helical springs is, in many applications, very important. C e n tr if u g a l s p rin g loaded regulators, controlling the speed of prime movers, spring-loaded indica­ tors, and many other apparatus require helical springs of which the correct num­ ber of active coils is essential. If, in the use of the conventional helicalspring deflection equation T — = — = 30,000,000 lb per sq in. 0.275 1/n — 3.63 0.392E = 11,700,000 lb per sq in. Any discrepancy between these given values and experimental values is due to an error in counting the number of active coils in the helical spring under consideration. I t is the purpose of this paper to show how the correct number of active coils can be determined both by analysis and experi­ ment. The number of active coils as used in the deflection formula for helical springs does not always equal the total number of coils or the number of free coils. Particularly, in most commer­ cial helical compression springs we find th at the number of active coils must be more than the number of free coils, if we assume G = 11,700,000 ~ 12,000,000 as correct and applicable to helical springs. Tests of regulator tension springs, of which each end coil was held a t two diametrically opposite points, checked closely with G = 2/5 • E when the number of active coils was counted from the middle of the two supporting points on one end of the spring to the corresponding point on the other end, i.e., when the num­ ber of active coils was taken as the number of free coils plus V2 coil. R / r d P G n = = = = = = deflection mean radius of coil diameter of wire load modulus of elasticity in torsion number of active coils an error is made in determining the correct value of n, the factor G is usually adjusted to offset the original error. The factors /, r, d, and P are always fixed in value and can easily be mea­ sured and checked. In such cases it is erroneously assumed th a t G is a variable amount for different helical springs, the variation ranging from below 10,000,000 to 12,000,000. The fact, however, is that the modulus of elasticity in torsion G is proportional to the modulus of elasticity in tension E and is characteristic for each material and constant within the elastic range: 1 A ssistant Chief Consulting Engineer, Allis-Chalmers Mfg. Co. Mem. A.S.M .E. R obert F. Vogt was born in Geneva, Switzerland, and had his prim ary and secondary schooling a t Rom anshorn, C an­ ton School a t St. Gall, and Swiss Polytechnicum a t Zurich, Switzer­ land. His professional career began in the U nited S tates in 1903. He has been connected w ith the Allis-Chalmers Mfg. Co. as m echani­ cal engineer since 1907. C ontributed by the Special Research Com m ittee on M echanical Springs and presented a t the Mechanical Springs Session of the Annual Meeting, New York, N. Y., Dec. 5 to 9, 1932, of T h e A m e r i ­ According to Htttte, 26th edition, od U nder T w is t Let us consider, as shown in Fig. 1, a rod of the length L twisted by applying a force P per­ pendicular to a rigid arm of length r, which is perpendicular to the rod. The deflection of the point of ap­ plication of the force P in the di­ rection of P is expressed by the formula: in which co is the angle of twist in radians. H e l i c a l S p r in g W it h R ig id A rm s a t C o il E n d s If the rod referred to for Equation [1 ] is coiled into the shape of a helical spring on which the rigid arms extend from the ends of the coil to the center line of the coil and are perpendicular to the coil center line, the force P acting on the arms at the center line of the coil in the direction thereof produces a deflection / of S o c ie t y o f M e c h a n ic a l E n g in e e r s . N o t e : Statem ents and opinions advanced can in papers are to b e understood as individual expressions of their authors, and n o t those of the Society. a = pitch angle n — number of coils 467 468 TRANSACTIONS OF T H E AM ERICAN SOCIETY OF MECHANICAL ENGINEERS In this case all coils are active, i.e., subjected to the twist of the full torque moment P ■r and no coils or p art of coils are in­ active. The number of coils n is also the number of active coils. C o m m e r c ia l S p r in g s Commercial springs are not equipped with rigid arms at the ends. Many tension springs have loop ends, as shown in Fig. 2, which add a negligible amount of bending deflection to the tor­ sional deflection given in Equation [2]. In such springs all coils between the base of the loops are active coils. T h e d e ta ile d discussion of this type ;fig_ 2 of s p rin g w ill be omitted in this paper. In helical springs where the full length of the bar is within the cylindrical part of the spring, as is the case of commercial helical compression springs and also of many kinds of expansion springs, the active coils extend beyond the free coils. The free coils include the coils of the spring which are not connected with brackets or yokes through which the spring load is applied and do not make contact with the end coils. The active coils include all coils contributing to the deflection of the spring. The angle of twist of the bar changes from maximum in the free coil to zero in the end coils. The angle of tw ist within the end coils adds a certain amount to the total deflection of the spring, which can be determined in many cases with complete accuracy and in all other cases with sufficient accuracy to satisfy fully the practical applications. H e l ic a l S p r in g s W it h D if f e r e n t T ypes of L coil is as flexible as the free coils located between A and B ' and the bar between A and B will twist in accordance with the re­ spective moment acting on the bar at its cross-sections. This moment is no longer constant and equals P ■r for all cross-sections from A to B, but decreases gradually from the maximum of P /2 • 2 • r = P ■r at A to P /2 -0 = 0 at B. Between A and B the P •r acting moment is M = P /2 • (r — r cos <p) = ----- (1 — cos ip) 2 for the cross-section designated by angle ip. The total angle of twist of cross-section at A , due to the torsional resilience in the end coil from A to B and the moment P • r acting at A, is 16 • P ■r2 a = — - ■■- —, and the center 0 of the coil, which we imagine CL * (r rigidly connected with cross-section a t A , moves in reference to its original position O', when the spring is not loaded, an amount of 16 • P • r* f = r ■a = ————— (see Appendix 1). O' may be regarded d •G as the center of the end coil in its new position. O may be taken as the original location of the center. The distance between 0 and O' then eq u als/', which is also the deflection of the end coil 16 - p . r s / ' = — - ——— . This corresponds to the deflection of lU coil d4 • G which is subjected to the moment P • r. The same, of course, occurs at the end A 'B ', so th a t the total deflection of the helical spring amounts to o a d in g In order to illustrate the deflection of the end coils in helical springs, various ways of spring loading will be analyzed: 1 The Two-Point L o a d in g . To an open-wound helical spring the load is applied by means of yokes reaching on each end dia­ metrically from one side of the coil to the other (see Pig. 3). The load P is applied at the middle of the yoke by means of a pivot, so th a t each end of the yoke transm its the same pull P /2 on the spring. As is shown in d e v e lo p in g Equation [2], the effect of the pitch angle a is such th a t it may be correctly assumed that the end coil is in a plane perpendicular to the center line of the coil and that the forces are perpendicular to the plane of the coil. Referring to Fig. 3, the load P /2 at A has no part on the def­ ormation of the end coil between the points A and B. The load P /2 at B produces a moment of P /2 • 2 • r = P ■ r at point A which is balanced by the same moment P • r acting on the other side of the spring. All cross-sections of the spring bar between A and B ' are under the influence of this moment P • r. If the end coil between A and B would be absolutely rigid, the cross-section a t A wrould remain in the same position relative to the end coil as it had before loading took place. B ut the end A n a l y s is at T h r e e - P o in t L o a d in g Proceeding in the same manner for three-point loading, shown in Fig. 4, where three equal forces P /3 are placed 120 deg apart on the end coil, it is found th at the de­ flection, due to twist in the end coils amounts to and the total spring deflection is In these analyses the effect of bending and shear have been omitted in favor of F ig . simplicity. The error made thereby is so small as to be of no practical consequence. C o n c l u s io n s o f A n a lyses for D 4 e sig n Most commercial spring-loading conditions will conform to the two-point loading and the equivalent deflection of the two ends of a helical spring in terms of the deflection of a free coil will approximate 0.5. RESEARCH The general expression for the deflection of helical springs which are not provided with rigid arms or loops at the ends and to which the load is applied in the common manner then takes a convenient form readily available to designers: For compression springs the forces are applied in the opposite direction; the deflection is also in the reversed direction, but the calculations and results are otherwise the same. The design and application of a commerical helical compression spring are such th at the load condition ranges between the two cases given. The first condition is the more common. The foregoing calculations are based on full-bar cross-section in the end coil. This assumption applies to most commercial compression springs, for that part of the coil which must be con­ sidered in the calculation. The basic design of the spring is shown in Fig. 5. The ends of such a spring are closed, 3/< of the end coil is tapered from full cross-section to 1/ i thickness a t the end, the pitch changes at contact point B from p to d. Full cross-section of bar is maintained from B to A , suggesting a load division of P/2 at A and P /2 at B. The variation is usually not far from this load division and comes within the range of the threepoint loading of 3 X P/3, in which extreme case the difference would only correspond to 1/ 3 — 1/ i = 1/ 12 coil for each end cor­ rection. We must bear in mind th a t the resultant of these forces is in the center line of the coil. If it were to fall outside the center line, the spring would bend out sidewise, which, in most compressionspring applications, does not occur to any appreciable extent. Within the range of free coils, i.e., from B to B ' (see Fig. 5), the bar is subjected to a shear force equal to P /2 and a torque equal to P ■r. The effect of shear upon the deflection is small and can be neglected. In a well-applied compression spring the torque P ■r is uniformly the same all along the bar, P acting in line of the center line of the coil. Due to the fact th a t the length of contact between the end coils increases slightly during increase in deflection, thereby effecting a slight decrease in active coils, the assumption of the 2 X P /2 load division is more justified, as the error in allowing for slightly less active coils than would correspond to an actual, possibly different load distribution, is compensated by the ten­ dency for a slight decrease in active coils during compression. It is therefore logical and practically correct to choose the 2 X P/2 load division. E x p e r im e n t a l V e r if ic a t io n of M o d u lu s........................................... d D n' n c= - n ' + 1/2 Q I 1 .8 3 9 7 .5 3 19V s 6 .0 7 5 6 .5 7 5 8Vt P f = 8 n-D* P f-d* =3 469 The springs were tested with a testing machine of 5000-lb capacity. The dimensions were carefully taken with microme­ ters and averaged from many measurements taken of diameters perpendicular to each other along the full length of the springs. The free coils were carefully determined and fixed by inserting spacers at the contact points with respective end coils. F ig . 5 The theoretical number of active coils n and the modulus of elasticity in torsion G may easily be determined by testing for deflection two compression springs made of the same material of equal bar and coil diameter, b ut with a different number of free coils, as follows: S p rin g 1 C o il d ia m e te r ...................................... B ar d ia m e te r ....................................... N um b er of free c o ils ....................... L o a d ........................................................ D e fle c tio n ............................................. N u m b er of a c tiv e c o ils ................... M od u lu s of e la stic ity in to r sio n . m = D d n' P /i n' G Sprin g 2 D d n" ~ P = h x m = n" — G = = + + x Since both springs are alike except for number of free coils, the modulus of elasticity in torsion is the same for both springs as well as the effect of the end coils under the same load. We find A na lyses A large number of tests substantiate the mathematical analysis and the general application of the results. To illustrate, the following test records are presented. Errors in the dimensions of bar diameter, coil diameter, and deflection, on account of their large amount, are relatively small. The examples, therefore, are of especially high value as proofs of the analyses. The following springs were made by the Railway Steel Spring Company for the Allis-Chalmers Manufacturing Company: B ar diam eter (a verage), in ___ C oil diam eter (average), in ___ Free le n g th ..................................... F ree c o ils ......................................... A ctiv e c o ils ..................... .............. T o ta l num ber of coils from tip to tip of bar (ta p e r ).. L oad, lb ............................................ D eflec tio n ........................................ RP-56-4 4580 0 .7 6 0 11,800,000 II 1.122 4 .1 4 0 195/ s ll1/* 12 H»/4 4600 1 .6 6 0 11,850,0 0 0 and If there should be any variation between the two springs in di, Di, or P, th e n /i o r /2 must be corrected to correspond to values for d, D, and P adopted for the foregoing calculation. I t will be found th at x approximates the value of */» very closely and that G approximates 11,700,000 for any size and kind of steel spring bar. Appendix 1 T N order to find the total angle of twist of cross-section at A under the influence of P • r, the half-circle between A and B is divided into differential lengths A (s) = r ■ A <p. The moment 470 TRANSACTIONS OF TH E AMERICAN SOCIETY OF MECHANICAL ENGINEERS acting on any cross-section of the arc A B is M = P /2 ■ a = P /2 ■r ■ (1 — cos < p). (See Fig. 3.) The twist angle u between two cross-sections separated by the distance A(s) amounts to 32 • M ■ A(s) ml , ------- -— —— = 4 m. The total twist angle—th a t is, twist angle tt ■a* ■ G of cross-section at A in reference to cross-section at B—equals the sum of the angles of twist for all sections A(s) located between point A and point B. It is in which C is the torsional rigidity and E l the flexural rigidity. This equation is satisfactory for any value of 0 between 0 and a. T w o - P o in t L o a d in g If there are two forces P /2 acting at A and B, the deflection at B is readily obtained from the direct application of Equation [a], substituting in it P /2 for P, j (l a = 0 = d * TT it, • 7r I = -----, and C = 64 G ■——. Then the deflection of point B is: oZ If we assume the center 0 of the arc A B rigidly connected to the cross-section at A , it will move with the cross-section A the 16 • amount of r -, which distance corresponds to the di ■G deflection of the center O of arc A B from its original position under the influence of load P /2 a t point B. For the three-point loading of the end coil, we find the de­ flection of the center of the arc of the end coil, proceeding the same as in the case of the two-point loading, as follows: Point 0 deflects in which form the equation shows that the deflection of an end coil is equal to the deflection of one-quarter of a free coil. Case W hen <t>> a If 0 is larger than a, the necessary deflection can be obtained by using Saint-Venant’s equation in conjunction with the reciprocity theorem.3 From this theorem it follows th at the load P applied at C (Fig. 7) produces at B the same deflection as the deflection at C produced by the load at B. Since Equation [a] gives the deflection at any point C in Fig. 6, we can get at once the deflection at any point B for the loading shown in Fig. 7. Appendix 2 C a se W hen 4> < a T T N calculating deflections of a portion of a circular ring out of its plane by forces perpendicular to the plane of the ring, the known solution of Saint-Venant can be used.2 If an incomplete circular ring is fixed at A and loaded by force P at B (Fig. 6), h e e e - P o in t L o a d in g Take now, as an example, the case of three loads P /3 put at points A , B, and C, 120 deg apart (Fig. 8). Point A is considered as fixed. The deflection of the point B consists of the two parts: (1) Deflection produced by the load P /3 at B and (2) deflection produced at B by the load P /3 at C. The first part is obtained by substituting into Equation [a] P /3 for P and 2x/3 for the angles a and 0. PR3 5 This gives: F sduetoS = 7.50 —- (assuming E = ~G) . (j t CL Z The second part is obtained from the same equation by putting P R3 4?r/3 for a and 2ir/3 for 0. This gives: Fsdue to C = 6.70 ——• Crd F ig . 7 Hence the total deflection of the point B is : PR3 then according to Saint-Venant’s solution the deflection at any point C, defined by an angle <f>,is given by the following equation: Vb = U.2 0 - 2 The S aint-V enant solution can be found in L ove’s “ M athem atical T heory of E lasticity ,” pp. 456-457, or in "S tren g th of M aterials,” vol. 2, p. 469, by S. Tim oshenko. T his m anner of solution was sug­ gested by R. L. Peek. In calculating the deflection of the point C, we again have two parts: (1) Deflection at C produced by the load at C is obtained • M ethod proposed by Prof. S. Timoshenko. RESEARCH by substituting P /3 for P and a = <t> = 4ji-/3 into Equation [a], PR3 which gives Fcdue to C — 33.10 —— and (2) deflection at C proGd duced by the load at B. This is equal to deflection at B when the load is at C and is obtained from Equation [a] by substituting in it P /3 for P and taking a = 4x/3 and </> = 2ir/3, which gives: VCdue to B = y Bdue to C = 6.70 RP-56-4 471 parallel to this center line, i.e., perpendicular to the plane. The component P /2 ■ sin a is negligible, as it does not contribute directly to the deflection considered and has very little influence on the diameter of the coil. The problem of finding the total spring deflection may, there­ fore, be illustrated by Fig. 10. PR3 Gd4 Then the total deflection at C is: I t will be seen that the method described can be used for any number of concentrated forces. I t can be easily extended also to the case of distributed loads. Appendix 3 HTHE Saint-Venant solution may be used directly to determine 1 the total deflections of helical springs under various load condi­ tions. The most simple case of its application for helical springs is a spring with the two-point loading as shown in Fig. 3, and Fig. 9. For this analysis the spring is assumed to consist of two equal parts, equally loaded, which meet at the center cross-section A of the spring bar. (See Fig. 9.) Point A is now considered the fixed point of two circular arcs. The center angles of these arcs are equal, a = (»'/2) 2ir + ir = (n ' + 1) x, 2n' being the number of coils be­ tween B and B'. If ip is the pitch angle of the spring, the actual length of arc AC is L = r • r (n' + 1) a n d th e COS <p loads on the arc perpendicular to the plane of the arc would be P /2 cos <p. But as shown by developing Equation [2], the prod, r • x (»' + 1) P uct ------------------ - — • cos eliminates cos <p. cos ip 2 Therefore, it may be assumed th at cos = 1 without interfering with end results. This means th a t it may be assumed that arc ABC is r ■it • {n' + 1) in length, located in a plane perpendicular to the spring center line and loaded by the forces P /2 which are The deflection f a of point H representing the center of the bracket BC is the mean of the deflection at B and C. The de­ flection fn of point B is composed of the deflection f u of point B due to the load P /2 at B and fn " due to the load P /2 at C, and the deflection fc of point C is composed of the deflection fc ' of point C due to the load P /2 a t C a n d /c " due to the load P /2 at B. The total deflection Sb " = fc ," according to the theorem of reciprocity, so that The total deflection of the spring is / = 2fH The d eflectio n s/s",/s', a n d /c ' can be determined by equa­ tion To this result a correction must be added to compensate for the influence of the pitch angle and the deflection due to pure shear. Saint-Venant’s equation4 for the deflection of a helical spring with rigid end levers is : 4 Love, “ M athem atical T heory of E lasticity,” p. 422; S. Timo­ shenko, “ Strength of M aterials,” p a rt 1, p. 289. 472 TRANSACTIONS OF TH E AM ERICAN SOCIETY OF MECHANICAL ENGINEERS B Sh I r a = = = = = spring load deflection length of spring bar radius of coil pitch angle and therefore the values presented in this paper may be an appro­ priate average for steel springs to be used at ordinary tempera­ tures only. If we transform this equation in terms used in Equation [2], we find In the deflection Equations [2], [3], [5], [a], [6], [7], and [8], it is assumed that the spring bar or rod is thin in comparison with the radius of the curvature, i.e., D /d « c o . The error caused by this assumption, when the equations are applied to com­ mercial helical springs where D /d = 3 or more is very small and for all practical purposes negligible. Equation [8] is given to show the influence of the pitch angle. Other influences which affect the accuracy, and are not con­ sidered in this equation are caused by preventing the free ends of the spring from turning freely about the axis of the spring during compression or expansion and the pure shear deflection. The complications affected by properly considering all these facts are too great, and the change in end results too minute to war­ rant the application of these highly refined methods of calcula­ tion in practical engineering work. As far as the deflection effect of the end coil is concerned, it is, for all practical purposes, sufficient to consider its torsional deflection only in the manner shown in Appendix 1. This is especially justified when we realize that the mathematically more complicated method employed in Saint-Venant’s solution is also not absolutely accurate because the effects of such items as pure shear deflection, coil pitch angle, spring index D/d, weight, and end conditions of the helical springs are neglected. Another factor, which demonstrates the fallacy of striving for accuracy to the extreme in calculating the deflection of the end coils, is the unavoidable variation in cross-section shape and size, coil diameter, and pitch angle in commercial springs. Equa64 • n ■r> ■P ,, 64 • («' + §) • r3 • P tions / = — ----------- and / = -----------——---------, respectively, Or * a* (j • a 4 can be regarded as being accurate for all practical purposes. Discussion T. M c L e a n J a s p e r .6 The paper by Mr. Vogt on helical springs is exceedingly interesting. I am wondering if the values of E, 0 , and l/m are as constant for spring steel in general as is assumed in the paper. My reasons for asking this go back to some tests made in 1924 which were published in the Transac­ tions of the American Society for Testing Materials of that year and some work presented in the Philosophical Magazine for Oc­ tober, 1923, which indicate that the state of the steel as well as the temperature at which the tests were made influences the values of the so-called elastic constants somewhat. The only way that this should be determined for spring applica­ tion is to make several tests on identically shaped springs made of different steels. I am not familiar with the values of O to be assumed for steel when formed into helical springs and when using different steels, W. M. A u s t i n . 6 The writer has had to apply helical compres­ sion springs, both large and small, to quite a variety of machinery and has often observed the influence of the end turns. Particu­ larly, he has observed that the average spring designed to be made like the author’s Fig. 6, except having a length relative to diame­ ter several times longer than Fig. 6, will usually buckle badly when fully loaded. Small springs often have their ends malformed. The spring maker winds enough wire on his mandrel to make two or more springs, and then cuts them apart. He then presses the end of the spring against the flat side of a rapidly turning dry grinding wheel. The heat generated makes the end turn red hot at some point about 3/s to Vs turn from the end of the wire. The wire bends at this red-hot place and the end of the wire moves back against the next turn. He then dips the spring in water in an attempt to restore the temper to the heated part, and finishes the grinding. The end turn, instead of tapering uniformly in thickness for 3/ 4 of a turn to V i the diameter of the wire at the end, tapers for V i turn to a thickness about l / z diameter of the wire, then in­ creases in thickness for another l/ t turn to 3/i diameter of the wire, then tapers another x/ 4 turn to V 4 diameter of wire at the end. This last taper may lie against the next turn for most of its length. The writer has often had to show the machine assembler (not a spring maker) how to cut off part of the end turn and regrind so that the spring will not buckle in service. Even if the spring were made according to the drawing as usually made, the center of gravity of the load would not be in the extended axis of the spring. If the spring is not more than three times as long as its diameter, the buckling is usually not very noticeable. The writer prefers to make the end turn so that the end of the wire does not touch the next turn until the spring is compressed solid, and instead of making the ground end exactly perpendicular to the axis of the spring, to make it a helicord of small pitch relative to the pitch of the spring. If this is done, the end of the wire will take its proper share of the load without bending beyond the plane of the part, 3A of a turn away, where the tapering of the wire began. Most springs are never completely unloaded in service, many of them never more than 1/ 2 unloaded. In cases like this the minimum load brings the end of the spring into a plane perpen­ dicular to the axis. It is probable that the center of gravity of the load is not in the axis of the spring, at the time of minimum load, but as the load increases the center of gravity of the load approaches nearer and nearer to the axis, and when maximum load is attained the ideal condition exists with the center of gravity of the load is in the axis of the spring. It is then seen that the flat-ended compression spring and its modifications is at best only a compromise, more or less suc­ cessful, so to load the spring that at no time during the compress­ ing or releasing of the spring will any part of it be stressed beyond its safe load. In tension springs provided with hooks bent up out of the end turn and having tHe same diameter as the main body of the spring, the hooks have to stand the same bending moment as the torsional moment in the body of the spring. This means that the tension stress on the inside of the hook is about twice the shearing stress on the inside of the body of the spring because, for 1 D ir e c to r o f R e s e a r c h , A . O . S m i t h C o r p o r a tio n , M ilw a u k e e , W is . M e m . A .S .M .E . Pa. 6 E n g in e e r , W e s t in g h o u s e E le c . & M fg . C o ., E a s t P i t t s b u r g h , M e m . A .S . M .E . RESEARCH round wire, the section modulus for torsion is ird3/16 and for bending is «23/32, so if S t be the maximum tension stress in the hooks and S s be the maximum shearing stress in the coils, then Thus, with the additional fact th a t the wire is often dam­ aged by making the hooks, accounts for the observed fact that tension springs, if they break, always break where the hook con­ nects to the body of the spring. There is a way to reduce the excessive stress in the hooks. I t is to make the end turn a spiral and bend up the hook from the inner end of the spiral, making the mean diameter of the hook about one-half the mean diameter of the main body of the spring. The author’s tests on the two large springs would be much more valuable if the springs had been loaded to near their maximum safe loads instead of limiting the stress as calculated by the old for­ mula to 34,400 for the small spring and 14,100 for the large one. In any event, I believe they should have both been loaded so as to produce the same stress. I t is quite generally known th a t Hooke’s law gives only the first term of a rapidly converging series, so that Young’s modulus E and the shearing modulus G both have higher values when de­ termined by stress-strain measurements using low stresses than they do when using stresses near but below the elastic limit. 16 • r3 ■P The author’s deflection r • w = ----------- due to the torsion in d4 ■G the end turn is the deflection due to torsion of the point B, Fig. 3, and not the deflection of the pivot hole in the bar connect­ ing points A and B. This would make the deflection of the 8 • r 3 ■P pivot hold only — —— • In the author’s analysis no account d r • (jt is taken of the deflection at B due to the bending of the end turn by the load P /2 at B. This would produce a deflection about as large as that due to torsion. In order to get experimental data on the deflection of the end turn, the writer had a piece of Vi-in. pretempered spring steel wire bent into 3/ t of a turn of 2n /ie mean diameter, as shown in Fig. 11. It was loaded at B with a 70-lb weight and the deflection at B was 0.29 in. If we let G = 11,400,000, the deflection per turn of the main body of the spring is 0.488, when loaded to 140 lb. The deflection at B is then seen to be 0.595 of the deflection of one turn, and the deflection at the center of the bar connecting A and B would be 29.7 per cent of the deflection of one turn, and for both ends the deflection due to the end turns is 59V2 per cent of one turn. J. P. M a h a n e y . 7 At the beginning of his paper the author shows that the value of G should be nearer 12,000,000 than 10,000,000. This is true provided Poisson’s ratio is taken as 0.30 to 0.335 rather than 0.365. In some instances attem pts have RP-56-4 473 been made to prove G equal to the lower value by substituting test data in the conventional spring formulas, but, since it is generally admitted th a t these formulas are approximate for “closely coiled” springs, such computations are not adequate proof. The author states th a t the bar in the free coils is subjected to a shear force of P/2. This is incorrect. The total load on the spring is P; consequently, the single bar must transm it this total load from one end of the coil to the other and the shear in the bar will be P instead of P /2. Since present conventional spring formulas can be proved inaccurate, it does not follow th a t illogical corrections are ac­ ceptable. Adding one-half a coil to the number of free coils admits th a t a portion of the seated end coils deflect, which is beyond comprehension. I t is true th a t torsional deformation extends beyond the free into a portion of the seated coils, for if this were not true, the first free coil at each end would not contribute its full share of deflection. To infer th at there is axial deflection derived from the seated coils is a process of creat­ ing one error to compensate for another. As the paper shows for balanced loading the load P on a com­ pression spring may be resolved into two components of P /2 each acting 180 deg apart. If the spring in Fig. 3 is loaded in compression, P /2 at B will produce torsional stress at A, and P /2 at A increases this stress to the final value within the spring. The stress in the seated coil must build up to the proper value at A in order th a t the active end coils may be completely effective in contributing deflection. The stress within the seated coil A-B produces deflection indirectly but its contribution to the total should not be counted twice. The author’s mathematical deduction clearly shows th a t the torque available in A -B is sufficient to produce deflection equivalent to one-quarter of an active coil provided it were free to move, which is of course im­ possible. R. L. P e e k , J r . 8 H o w accurately the solutions given for twoand three-point loading apply to helical springs compressed be­ tween parallel plane surfaces requires further analysis. Follow­ ing the treatm ent given in Love’s “ Mathematical Theory of Elasticity,” pp. 456-457, I have evaluated the force required to keep the extreme end of the inactive turn in contact with the point A (Fig. 3), a condition th at must be satisfied under com­ pression of this sort. I find this force trivial in comparison with the reaction at A under two-point loading, and this con­ sideration therefore does not affect the validity of applying the result for two-point loading to compression between parallel plane surfaces. On the other hand, in such compression the change in pitch angle of the active coils w'ill cause their axis to be no longer normal to the parallel plane surfaces applying load and their deformation will not be th a t corresponding to a purely axial thrust. Whether this effect will appreciably change the result, I have not ascertained. A. M. W a h l . 9 The exact solution of the additional deflection produced by the end turns of a helical compression spring is undoubtedly a very complicated problem, since it depends on the exact shape of the end turns and on the distribution of load thereon. The author has simplified the problem by assuming the end turns to have the full bar cross-section throughout their length. In addition he assumes various distributions of load on the end turns, finally choosing th a t which seems to agree best with test results. Since, in most practical cases, the deflection due to the end 8 Bell Telephone Laboratories, New York, N. Y. 7 Assistant Professor, Industrial Engineering, Virginia Polytechnic 9 W estinghouse Research L aboratories, E ast Pittsburgh, Pa. Institute, Blacksburg, Va. Jun. A.S.M .E. Assoc-Mem. A.S.M .E. 474 TRANSACTIONS OF TH E AMERICAN SOCIETY OF MECHANICAL ENGINEERS turns is but a relatively small part of the total deflection of the spring, a considerable error in estimating the effect of the end turns could be made without introducing much relative error in the total deflection of the spring. For this reason a rough ap­ proximation, such as the author has introduced, might be of value in practical work, provided it has been confirmed by a number of accurate tests. The question of the effect of the end coils is closely bound up with that of the modulus of rigidity of the material. For years some spring manufacturers have used modulus values of 10.5 X 10® or 10 X 106 lb per sq in., as the author points out. It is well known that these values do not agree with modulus values obtained by means of torsion tests on ordinary spring steels. It has been the writer’s opinion that these modulus values have been used largely to compensate for inaccuracy in estimating the effect of the end turns and possibly for errors in the spring dimensions. To illustrate this point, some tests made on different springs at the Westinghouse Research Laboratories will be mentioned. The method used was to measure deflections between prickpunch marks on diametrically opposite points of the coil in the body of a helical spring and is described in a previous publica­ tion.10 The coil diameter and wire diameter were carefully measured at several points on each coil and the results averaged. By measuring deflections in the body of the spring, the effect of the end turns was eliminated. The values of “effective” modu­ lus 0 could then be found from the known formula Three springs from one manufacturer, having indexes of about ten, when tested in this manner, yielded the following values for the modulus: Spring N o ........................................... G X 1 0 -« lb per sq in .................. 1 1 1 .4 5 2 1 1 .4 6 3 1 1 .5 0 Three springs having indexes of about 6.5 from another manu­ facturer gave the following values: Spring N o ........................................... Q X 10" M b per sq in .................. A 1 1 .1 9 B 1 1 .1 2 C 1 1 .3 0 These values are all definitely higher than the value of 10 or 10.5 X 106 as assumed by some spring manufacturers. It should be noted that this method of determining the modulus assumes that the effect of the spring curvature is small, i.e., that the spring acts like a straight bar subjected to a torsion moment Pr. This of course becomes more nearly true for springs of large index. As far as spring deflections are concerned, this assumption is born out by previous tests by the writer,10 wherein it was found that the ordinary deflection formula for helical roundwire springs was correct within 3 per cent for springs having indexes varying from 2.7 to 9.5. In other words, a fourfold in­ crease in curvature of a spring having a given wire diameter did not seem to have an appreciable effect on the modulus. The same thing is known to be true of curved bars in bending; i.e., in general, a curved bar in bending may be computed within a few per cent accuracy as far as deflections are concerned by using the fundamental methods applied to straight bars, although this is not true when stress calculations are made. The effect of curvature on deflection was also found to be small in the case of helical springs of circular wire by O. Gohner,11 who used more exact methods of calculation involving the theory of elasticity. The effect of curvature may be checked up experimentally by using the following method suggested by R. E. Peterson, of the Westinghouse Company. A heat-treated round bar of spring material is first tested in torsion, thus determining the technical value of the modulus G. This bar would then be wound into a spring, and heat treated, after which deflections would be mea­ sured in the body of the spring between prick-punch marks, so that the “effective” value of G could be found by use of the ordi­ nary spring-deflection formula. The two values of G thus found should be nearly the same if the effect of curvature is small. The writer would like to suggest that in determining the num­ ber of coils to add to the free coils to find the active coils, it is necessary to know the “effective” value of G accurately; in other words, a small error in G would produce a big error in the number of added turns. For example, in the case of the author’s spring II, if G is assumed 11.85 X 10®, then from it is found that n = 12, whence the added coils become 12 — llV i = Vs- But suppose G = 11.6 X 10® instead of 11.85 X 10* (a variation not at all unreasonable). Then we would find n = 11.65, from which the added coils would be found to be 11.65 — 11.5 = 0.15, a value which differs greatly from ‘/a as found by the author. This example shows the necessity for an accurate knowledge of the “effective” value of G. This could be determined, as mentioned previously, by measurements between prick-punch marks in the body of the spring, after which the average dimensions of the spring would be accurately mea­ sured. In this connection, the writer has found it to be extremely difficult to obtain accurately the average wire diameter of a spring, without cutting it up after the test, since, due to coiling, the wire section becomes slightly oval. The method of determining the number of active coils, as proposed by the author, consisting of using two springs similar in every respect except in number of turns, would no doubt give an approximation which would be useful in practical work. For purposes of checking the theory, however, it would be neces­ sary to find the average dimensions of each spring accurately. This would involve more labor than would the testing of one spring, as suggested above. Furthermore, there is a possibility that the modulus would vary some between the two springs, and this again would involve an additional error. For these reasons it is the writer’s opinion that tests on one spring would be preferable in order to confirm the theory. The value of Poisson’s ratio 1/m = 0.363 reported in the paper seems rather high for steel. Using G = 11.7 X 10a, E = 30 X 106, this would give 1/m = E/2G — 1 = 0.283. Taking 1/m — 0.3 (a value commonly used for steel) and E = 30 X 10®, this would give G = 11.53 X 106, which is not far from the values ob­ tained in the writer’s tests mentioned above. A u th o r' s C losure Answering Mr. McLean Jasper’s discussion in regard to the constancy of the modulus of elasticity E and Poisson’s ratio m for spring steel at various temperatures, we may, according to Hiitte, for all practical purposes assume E and m and therefore G constant at temperatures between 0 F and 400 F. Examples of springs applied at high temperatures are springs in steam indicators and on valves for internal-combustion engines and steam engines. As far as the author knows, the steam-indicator springs which are used for high-temperature steams and *• A. M. Wahl, “Further Research on Helical Springs of Round gases as well as for cold air have been accepted as accurate for and Square Wire,” Trana. A.S.M.E., 1930, paper APM-52-18, practical purposes without using any correction factors for the p. 217. 11 O. Gohner, “Die Berechnung zylindrischer Schraubenfedern,” various temperatures to which they are exposed. The author, however, mainly considered springs used in atZ.V.D.I., March 12, 1932. RESEARCH mospheric temperatures where accuracy in deflection values are essential. Mr. Austin calls attention to irregularities in the shape of commercial compression springs especially in small sizes. How16 • r3 ■P ever, he errs in his conclusion th at the deflection r ■o> = ------- -— dl ■G is the deflection of point B (see Fig. 3). This deflection is derived from the product r ■a which (as is clearly explained in the paper and in Appendix 1) is nothing else than the deflection of the original end-coil center 0, which, as well as th a t of the pivot hole between points A and B, is at a distance r from the center of the bar cross-section subjected to torsion. Mr. Austin’s claim that the bending effect of load P /2 on the deflection of point 0 would be as large as th a t of torsion only is unfounded as may be seen from the Saint-Venant solution (see Appendixes 2 and 3) which includes the bending effect of load P /2 at B and shows that the deflection of point O is even somewhat less than that given by the author for torsion only. In Mr. Austin’s experiment shown in Fig. 11 the deflection of point B is claimed to have been 0.29 in. for a load of 70 lb at B; but according to Saint-Venant’s solution this deflection should have been 0.231 in. for G = 11.4 X 106 or 0.225 in. for G = 11.7 X 10«. Mr. Austin would have found more accurate and reliable re­ sults had he arranged his experiment according to Fig. 12. This arrangement consists of a helically and closely coiled springsteel wire of one and a fraction of a turn. The coil diameter is about 20 or more times the diameter of the wire which latter should be about V< in. The wire and coil diameter and the de­ flection should be large enough to make unavoidable errors negligible in reference to the deflection. In Fig. 12 B A B ’ is exactly one full coil and BA = A B ' and each is one-half coil. In order to eliminate errors due to initial tension or deflection, the deflection ei — e2 for the load Qi — Q2 is determined. The 1 --- 62 deflection of a point B in reference to point A is / = --------2 for the load Qi — Q2 at B. In order that the stress in the wire is within the elastic limit of the spring steel Qi must be less than 15,000 lb where d and D are given in inches. The sum of the differential deflections in the two half coils BA and A B ' is alike and opposite in direction. For this reason the wire cross-section at A does not turn and therefore does not cause a change in the true deflection of B in reference to point A. In Mr. Austin’s test, however, the cross-section at A, Fig. 11, RP-56-4 475 will turn and thereby increase the deflection of B, an amount corresponding to the torsional twist in the wire within the copper clamp near point This clamped portion of the wire cannot be held securely enough by the comparatively soft copper clamp to prevent twisting of the wire and consequently the turning of the wire cross-section at A. This torsional displacement of cross-section A of course increases the actual deflection due to the twist in half coil BA which stamps a test made according to Mr. Austin’s arrangement, shown in Fig. 11, as unreliable. The author has made a number of experiments according to Fig. 12 in which he found the deflections to check very closely with the Saint-Venant results. J. P. Mahaney mentions th a t the pure shearing force in the free coils due to the load P must be equal to P which is quite cor­ rect. However, according to the explanation given by the au­ thor in answer to A. M. Wahl’s discussion, the shearing force P is divided into halves. One-half balances an excess of the sum of the torsional shearing-force components parallel to the axis of the spring and acting in a direction opposite to P, while the other half adds a pure shear deflection to the torsional deflection 8 ■n ■D 3 ■P as given by the conventional deflection equation / = ----- -——----a 4 ■G Mr. Mahaney, after admitting th at torsional deformation ex­ tends beyond the free coils into a portion of the seated coils, elabo­ rates considerably on his conception th at since the end coils in a compression spring are not free to move they cannot contribute to the deflection of the spring. The deformation of the end coil, Mr. Mahaney claims, makes it possible for the first free coil to contribute its full share of deflection. If this statement were true, the conventional spring-deflection Equation [2] as de­ veloped in the paper under the heading “Helical Spring With Rigid Arms at Coil Ends” would be faulty, as the first free coils in this case do not have the benefit of torsional deformation in end coils and therefore wrould not contribute their full share of deflection. Obviously, such a contention is against sound reason­ ing as the development of the deflection Equation [2] includes the contribution of the full share of deflection of all coils. The fact th at the end coils are held so th at they can move only axially and parallel to their plane does not prevent the bar of the end coils from twisting due to the torque applied. Thus the axial deflection of the spring is increased proportionally to this twist and corresponds to one-fourth of an additional free coil per spring end beyond the deflection of a spring with rigid arms at free coil ends. The author fully agrees with R. L. Peek, Jr., th at in cases of compressing helical springs between parallel plane surfaces, the deformation of the spring as a whole and in particular of the end coil, will be different from the deformation as calculated in ac­ cordance with assumptions made in the analyses in the paper. This difference will vary with the different shapes of the spring ends as furnished in commercial helical compression springs. However, w^hen we consider the error range due to (a) using the conventional spring-deflection equation instead of the Saint-Venant equation given in Appendix 3, (6) unavoidable variations in spring-bar and coil diameters of commercial springs which appear in the equation in the fourth and third power, re­ spectively, (c) change in pitch angle and coil diameter during compression, (d) uncertainty as to spring end loading conditions, (e) neglecting the influence of the spring index D/d, and (/) un­ certainty as to the actual value of the modulus of elasticity E or G, respectively, the variation of the actual deflection of the end coil from the one calculated, and given as being equal to the de­ flection of Vi coil due to maximum torque, is so small in compari­ son to other discrepancies th at its disregard is fully justified. This is very apparent when we realize th a t a 5 per cent error in determining the end-coil deflection results in an error of less than 476 TRANSACTIONS OF TH E AMERICAN SOCIETY OF MECHANICAL ENGINEERS 0.5 per cent in reference to the total deflection of a spring with five active coils. An objection-free determination of the actual deflection of the end coil of a commercial helical compression spring with an accuracy within such a small error range would be very difficult. Referring to A. M. Wahl’s discussion, the values of E, G, and m were taken from the latest edition of Hiitte, 1931, first volume, p. 689, where the following data for spring steels is given: E = G = G /E = E — G = m = 2 ,1 0 0 ,0 0 0 kg per sq cm or 8 2 2,000 kg per sq cm 0 .392 v 30 ,0 0 0 ,0 0 0 lb per sq in . 11,700,000 lb per sq in . 3 .6 3 , from G = E / 2 (1 + 1/m) These data have always corresponded with spring tests made under the consideration of the proper number of active coils (re­ gardless of small or large number of active coils), as given in the author’s paper, and were therefore accepted by the author as being dependable. Hiitte is considered one of the outstanding sources of reliable engineering information. Mr. Wahl questions the accuracy of determining the value of G by testing two springs as suggested by the author, and in his example assumes G = 11.5 X 10® instead of 11.85 X 106, in which case Mr. Wahl calculates the effect of the end coils to be th a t of 0.15 free coils instead of 0.5 as demonstrated in this paper. Mr. Wahl’s analysis is, on this point, incorrect and de­ ceiving. In the author’s example, G is determined from actual values of n ' = 11.5, d ± 1.122, D = 4.14,/ = 1.66, and P — 4600 and (in conformity with the theory developed in the paper) n = n ' + ' / 2. If, in the example, the value of G had been different, say 11.5 X 10°, then the deflection/ would have been 1.715 in. in­ stead of 1.66 in. as it actually showed in the test, and n = 12 and not 11.65. The number of effective coils is fixed by the spring design and does not depend on the value of G. The value of G cannot vary much for commercial spring steel. The skeptical engineer, however, can determine its value and concurrently the actual effect of the end coils, with satisfactory accuracy, by the method of testing two springs of equal dimen­ sions but with greatly differing numbers of coils, as suggested by the author in the last part of his paper. Mr. Wahl, in referring to the influence of the spring index on spring deflection, mentions th a t the conventional spring-deflection equation for helical round-wire springs is correct within 3 per cent for springs having indexes varying from 2.7 to 9.5. The author determines the effect of the spring index on the spring deflection definitely by adding the direct shear deflection to the torsional deflection of the helical spring. The deflection t P ■L of direct or pure shear for the spring is /" = yL = - L = ———, G Fi ■G where L = 2Rirn, d2 * 7T Fa = ------- (for circular cross-sections from 4- 1. 2 Hiitte), P = spring load, or /" .= ———----------- . About half G • o2 of this deflection is already included in the conventional spring equation, as in helical springs under load P about one-half the shear load P is balanced by the total sum of torsional shearing- stress components parallel to the spring axis, as explained by Dr.-Ing. A. Rover in Z.V.D .I., Nov. 20, 1913, p. 1907. By using the conventional spring-deflection equation, it is assumed th at a curved bar has the same torsional deflection as a straight bar of the same length. The stress distribution in the cross-sections of the straight and curved bars is, however, slightly different and causes a small difference in deflection, amounting to one-half the deflection due to pure shear. The deflection of the curved bar is less than th at of the straight bar, when the pure shear deflection is considered for both. The total deflection of the helical spring under load P is: or or 7 t R D d n P G / L /" = = = = = = = = = = = shear angle in radians shearing stress mean radius of coil mean diameter of coil diameter of spring wire or bar number of active coils spring load torsional modulus of elasticity total spring deflection effective length of wire or bar deflection due to pure shear. Applying the extended deflection accurate spring tests will result in for G. Plotting the shear deflection, in deflection, against the spring index in Pig. 13. equation in conjunction with finding more uniform values per cent of torsional spring D /d we find the curve given Depositories for A.S.M .E. Transactions in the United States B OUND copies of the complete Transactions of The Ameri­ can Society of Mechanical Engineers will be found in the libraries in the United States and other countries which are listed on the following pages. Alabama Auburn................. Engineering Library, Alabama Poly. Inst. Birmingham.........Public Library Arkansas Fayetteville..........Engineering Library, Arkansas University California Berkeley............... Library, University of California Long Beach..........Public Library Los Angeles..........Publio Library University of Southern California Oakland................Oakland City Library Teachers' Professional Library Pasadena..............Library, California Institute of Technology Santa C lara..........Library, University of Santa Clara San Diego............ Public Library San Francisco. . . . Public Library (Civic Center) Engineers Club of Sail Francisco Mechanics Institute Stanford Univ__ Library, Stanford University Colorado Boulder.................Library, University of Colorado Denver..................Public Library Fort Collins......... Colorado State Agricultural College Connecticut Bridgeport............Public Library H artford...............Public Library New H aven..........Public Library and Yale University W aterbury........... Silas Bronson Library Delaware Newark.................University of Delaware Wilmington..........Wilmington Free Institute District of Columbia Washington..........Scientific Library, U. S. Patent Office Library of Congress Bureau of Standards Library George Washington University Catholic University Florida Gainesville........... University of Florida Jacksonville......... Free Public Library Miami...................Public Library T am pa..................Publio Library Georgia A tlanta................. Carnegie Public Library Georgia School of Technology Savannah............. Public Library Idaho Moscow................ University of Idaho Illinois Chicago................ John Crerar Library Western Society of Engineers Library, Armour Institute of Technology Public Library of Chicago Moline.................. Public Library U rbana................. University of Illinois Indiana Evansville............ Public Library Fort W ayne......... Public Library Indianapolis.........Public Library and Indiana State Library Notre Dame.........Library, University of Notre Dame Terre H aute.........Rose Polytechnic Institute West Lafayette.. .Library, Purdue University Iowa Ames..................... Iowa State College Des Moines..........Public Library Iowa C ity.............State University of Iowa Kansas Kansas C ity .........Public Library, Huron Park Lawrence..............Library, University of Kansas M anhattan...........Kansas State Agricultural College W ichita.................Wichita City Library Kentucky Lexington.............University of Kentucky Louisville..............Speed Scientific School University of Louisville Louisiana Baton Rouge....... Louisiana State University New Orleans........Howard Memorial Library Louisiana Engineering Society Public Library Maine Orono....................University of Maine Maryland Annapolis..............United States Naval Academy Baltimore............. Johns Hopkins University Engineers Club of Baltimore Public Library Massachusetts Boston.................. Northeastern University Boston Public Library Cambridge........... Harvard University (Engineering Library) Massachusetts Institute of Technology Fall R iver............ Public Library Lowell................... Lowell Textile Institute Lynn..................... Free Public Library New Bedford....... Free Public Library Springfield............Springfield City Library Tufts College....... Tufts College Worcester............. Worcester Polytechnic Institute Free Public Library Michigan Ann A rbor............University of Michigan D etroit..................Public Library Cass Technical High School Highland Park Public Library University of Detroit E ast Lansing....... Michigan State College F lint...................... Public Library Grand R apids. . . . Public Library Houghton............. Michigan College of Mining & Technology Jackson.................Public Library Minnesota D uluth.................. Public Library Minneapolis......... University of Minnesota Minneapolis Public Library (Engineering and Circulating Libraries) St. P au l................ James Jerome Hill Reference Library Mississippi State College........Mississippi State College Missouri Columbia..............University of Missouri Kansas C ity.........Public Library Rolla..................... Missouri School of Mines and Metallurgy St. Louis............... Engineers Club of St. Louis Public Library Washington University Mercantile Library Montana Bozeman...............Montana State College TRANSACTIONS OP TH E AMERICAN SOCIETY OP MECHANICAL ENGINEERS Nebraska Lincoln..................University of Nebraska Omaha.................. Public Library Nevada Reno..................... University of Nevada Library New Hampshire D urham ................University of New Hampshire New Jersey Bayonne............... Free Public Library Camden................ Free Public Library Elizabeth..............Free Public Library Hoboken...............Stevens Institute of Technology Jersey C ity...........Free Public Library Newark.................Newark College of Engineering Free Public Library New Brunswick... Rutgers University Paterson............... Free Public Library Princeton..............Princeton University Trenton................ Free Public Library New York Albany.................. New York State Library Brooklyn.............. Polytechnic Institute P ra tt Institute Brooklyn Public Library Buffalo.................. The Grosvenor Library Engineering Society of Buffalo Buffalo Public Library Ithaca................... Cornell University Jamaica, L. I ....... Queens Borough Public Library New Y ork............ Engineering Societies Library Public Library College of the City of New York Cooper Union Columbia University New York Museum of Science and Industry New York University Library Potsdam ............... Clarkson College of Technology Rochester............. Rochester Engineering Society Schenectady.........Union College Syracuse............... Syracuse University Public Library T roy...................... Rensselaer Polytechnic Institute Utica..................... Public Library Yonkers................ Public Library North Carolina Chapel Hill...........University of North Carolina (Engineering Library) Raleigh................. North Carolina State College North Dakota Fargo.................... N orth Dakota State Agricultural College G rand Forks........ University of N orth Dakota Ohio A da........................Ohio Northern University Akron....................Public Library University of Akron C anton..................Public Library Cincinnati............ University of Cincinnati Public Library Engineers Club of Cincinnati Cleveland............. Public Library Case School of Applied Science Cleveland Engineering Society Columbus............. State of Ohio Library Public Library Ohio State University D ayton................. Engineers Club of Dayton Toledo...................Public Library University of Toledo Youngstown......... Public Library Oklahoma N orm an............... Oklahoma University Oklahoma C ity .. .Public Library Stillwater..............Oklahoma Agricultural and Mechanical College T ulsa.....................Public Library Oregon Corvallis............... Oregon State Agricultural College Portland............... Portland Library Association Pennsylvania Allentown.............Free Library Bethlehem............Lehigh University E aston.................. Public Library Lafayette College E rie....................... Public Library Lewisburg............ Bucknell University Philadelphia.........Engineers Club Drexel Institute University of Pennsylvania Franklin Institute Pittsburgh............ University of Pittsburgh Engineers' Society of Western Pennsylvania Carnegie Institute of Technology Carnegie Library (Schenley Park) Carnegie Free Library of Allegheny Reading................Public Library Soranton...............Publio Library State College........Pennsylvania State College Swarthmore......... Swarthmore College Villanova..............Villanova College Wilkes-Barre........Public Library Rhode Island Kingston.......... .Rhode Island State College Providence........... Brown University Providence Engineering Society Public Library South Carolina Clemson College. .Library, Clemson College Tennessee Kingsport....... ... .Public Library Knoxville..............University of Tennessee Memphis.............. Goodwin Institute Nashville.............. Vanderbilt University Texas Austin................... University of Texas College S tatio n .. .Agricultural & Mechanical College of Texas Dallas................... Public Library El Paso.................Public Library Forth W orth........Carnegie Public Library Houston................Rice Institute Public Library Lubbock............... Texas Technological College (School of Engineering) San Antonio.........Carnegie Library Utah Salt Lake C ity .. .University of Utah Public Library Vermont Burlington............University of Vermont Virginia Blacksburg...........Virginia Polytechnic Institute Charlottesville.. . . University of Virginia Norfolk.................Public Library Richmond............ Virginia State Library Washington Pullman................State College of Washington Seattle.................. Public Library Engineers Club University of Washington Spokane................ Public Library Tacoma................ Public Library West Virginia Morgantown........ West Virginia University Wisconsin Madison............... Library, University of Wisconsin Milwaukee............Public Library Board of Industrial Education, Vocational School Library M arquette University Wyoming Laram ie................ Wyoming University Depositories for A.S.M .E. Transactions Outside the United States Argentine Buenos Aires........Biblioteca de la Sociedad Cientifica Australia Adelaide............... Public Library of Adelaide Melbourne............Publio Library of Victoria P erth.....................University of Western Australia Library Sydney..................Public Library, N. S. W., Sydney Brazil Rio de Janeiro— Bibliotheca da Escola Polytechnica Bibliotheca Nacional Sao Paulo............. Bibliotheca da Esoola Polytechnica Canada Montreal.............. McGill University Engineering Institute of Canada Toronto................ University of Toronto, Library Chile Santiago............... Universidad de Chile, Facultad de Ciencias Fisicas y Matematicas (Engrg. School) Cuba H avana.................Cuban Society of Engineers Czechoslovakia Prague.................. Masaryko va Akademie Prace Society of Czechoslovak Engineers Danzig Free City............. Bibliothek der Technischen Hochschule Denmark Copenhagen......... The Royal Technical College England Birmingham.........Birmingham Public Libraries B ristol.................. University of Bristol Cambridge........... University of Cambridge Leeds.................... University of Leeds Liverpool..............Public Library of Liverpool • Liverpool Engineering Society London................. City & Guild Engineering College Institution of Automobile Engineers Institution of Mechanical Engineers Institution of Civil Engineers Institution of Electrical Engineers The Junior Institution of Engineers The Royal Aeronautical Society M anchester..........Manchester Public Libraries (Reference Library) Oxford.................. University of Oxford Newcastle-uponT yne................. The North East Coast Institution of Engineers and Shipbuilders Sheffield................Sheffield Public Libraries WaZes Cardiff.................. Cardiff Public Library France Lyons....................University of Lyons Paris..................... IScole Nationale des Arts et Metiers Ecole Nationale Supfirieure de L ’Aeronautique ficole Centrale des Arts et Manufactures de Paris Soci6t6 des Ingfinieurs Civils de France Germany Berlin....................Verein deutseher Ingenieure Bibliothek der Technischen Hochschule Breslau................. Bibliothek der Technischen Hochschule Cologne (Koln).. ,Universita,ts- und Stadtbibliothek Dresden................Bibliothek der Technischen Hochschule Dusseldorf............Bucherei des Vereines deutsoher Eisenhuttenleute Frankfort............. Technische Zentralbibliothek Germany {continued) H am burg.............. Bibliothek der Technischen Staatslehranstalten H anover................Bibliothek der Technischen Hochsohule K arlsruhe............. Bibliothek der Technischen Hochsohule Leipsic.................. Stadtbibliothek M unich.................Bibliothek der Technischen Hochschule Bibliothek des Deutschen Museums S tu ttg art.............. Bibliothek der Technischen Hochschule Hawaii Honolulu...............University of Hawaii Library Holland Amsterdam...........Koninklijke Akademie von Wetenschappen D elft......................Bibliotheek der Technische Hoogesohool The Hague........... Koninklijk Instituut van Ingenieurs R otterdam ............Nationaal Technisch Scheepvaartkundig Institut India Bangalore............. Mysore Engineers Association C alcutta................Bengal Engineering College Poona....................Poona College of Engineering Rangoon............... University of Rangoon Ireland Belfast.................. Queen’s University of Belfast Italy M ilan.................... Biblioteca della R. Scuola d’Ingegneria Comitato Autonomo per l’Esame della Invenzioni Naples...................Biblioteca della R. Scuola d’Ingegneria Rome.................... Biblioteca della R. Scuola d’Ingegneria Consiglio Nazionale delle Ricerche presso il Ministero della Educazione Nazionale T urin.....................Biblioteca della R. Scuola d ’Ingegneria Japan Kobe..................... Kobe Technical College Tokyo................... Imperial University Library The Society of Mechanical Engineers Yokohama............Library of Yokohama Mexico Mexico C ity......... Asociacion de Ingenieros y Arquiteotos de Mexico Library of the Escuela de Ingenieros Mecanicos y Electricistas Norway Oslo....................... Den Polytekniske Forening Poland W arsaw.................Bibljoteka Publicazna Porto Rico Mayaguez.............University of Porto Rico Portugal Lisbon...................Institute Superior Technico Roumania Bucharest............. Scoala Polytechnica din Bucharest Scotland Glasgow................Royal Technical College Mitchell Library South Africa Cape Town.......... University of Cape Town Johannesburg.......South African Institute of Engineers Sweden Stockholm............ Kungl. Tekniska Hogskolan Svenska Teknologforeninger Gothenburg..........Chalmers Tekniska Institut TRANSACTIONS OF T H E AMERICAN SOCIETY OF MECHANICAL ENGINEERS Switzerland Zurich................... Eidgenossische Technische Hochschule Turkey Istanbul................ Robert College U.S.S.R. K harkov............... Supreme Economic Council of Ukraine Leningrad.............Leningrad Polytechnic Institute Moscow................ Supreme Council of National Economy Tomsk..................Tomsk Polytechnic Institute