P. C. RENZO Dacor of Enginering. S. KAUFMAN Chiat Engineer, Couping Dvn Paper No. 67—WA/DE-8 Gear Couplings A discussion and mathematical analysis of the operation of gear conplings at angular D. E. DE ROCKER Chief Engineer, roduc Bavelopmen! SieBath Gear Cow in North Bergen, Ke Thess are diterences of opinion as to how flexible sear couplings tet, whether they transmit true uniform motion truant ng ow sang eeth are in contac, ad what Kind of erown should be applied ‘We shall now ry #0 establish the facts pertinent to subject. Gear coupling ae unslly arrange in pais, enc individual coupling eomprcing alee with stright teth oni inside, and fs hub with tevth eronned to cooperate withthe sleeve athe Tange of anges speed ‘see, Seo uy fare shown in aligned position na frag smontary cou etin in Fg, 1 and nan ain section in Fg 2 Gerealy, the sleeve teeth have involute profes, és ring from a bate cine, bas on conventional gears, Adjacent ine olute tooth surfaces hve a constant distance p from euch other Snywhet, taken In the direction of the surface normal a In consequence, the adjacent crowned tooth surfaces of the hub Contributed by the Design Engineering Division for presentation at the Winter Annual Meeting. Pittsburgh, Pa, Novernber 12-17, er, of Tae “Autenicay Soctery or Mecuawien. ENaINEERS Manuscript received at ASME Headquarters, July 26, 1967. Paper No.t—-Wa/DES. a inisalignment: ransmission of uniform motion, tooth separation, tooth Toad ditibu- tion, coupling load capacity ath bearing, ond special tooth forms. should alse have a constant distance from each other in the dire tion of the surface normals to match the sleeve teeth. This re- quirement it no different on couplings than itis on gesrs. Teeth ‘unequal normal distance p could not be brought to match ‘and take over load smoothly fram.one another. ‘As a resillof this requirement, the tooth profiles ofthe bab x planes g perpendicular to the hub axis, should change inereas- ingly with increasing distance from the hub axis, atleast when the ‘coupling is designed for a substantial running angle, This wll be further deseribed, Fig. 2 shows & conventionsl uniformly crowned lub. ‘The hub looks like an excessively crowned gear. A8 on & gear, its shape is best defined by the shape of the rack teeth with which it can ‘mesh and run ao that its entire tooth sides got into contact. ‘This rack can be considered an extremely large, infinitely large, gear, A'hob produces the straight profile of the involute rack in the midplane G. If now the hob is fed about an axis C as if ‘uumed about this axis, it will produce the same straight profile in ‘all planes containing axis C and envelop the rack tooth shape. ‘Each tooth surface of the rack contains straight profiles that jntersect axis @ at the same point and that have a constant in- clination to axis C, In other words, the surface that would be [prduced on the reck is conical surface with axis . a2In its feed motion about axis C, the rotating hob, 69 to say repreecuts a moving rack that is the counterpart of the mek or Infinitely large goar produced ty the bel Fig. 3 shows part of the represented rack with axis C in perspective. ‘Tho straight rack profiles show up in all axial sections, asin midplane @. ‘They come to. point or apex an axis C. The cone apex of side SrisatC,, the one of side Sat. Adjacent sides S, Se are iden- tical conics) surfaces merely dispinced along the cone axis C. ‘They have a constant distance from each other everywhere in the direction of the surface normals, “The tooth sides of the hub are produced in generating motion whereby the rack represented by the hob rotls on the bub, This rolling mation is as if a eylindrioal piteh surface & of the hub would roll on the piteh plane of the rack ia contact therewith ‘The pitch surfaces intersoct the tooth sides in eurves called pitch lines. As the pitch surfaces roll on each other without sliding, the pitch lines of the hub are as if printed on ite eylindrical piteh surface by the pitch lines of the tack. In development of the cylindrical piteh surface to plane, the piteh lines ofthe hub are identical with the piteh lies ofthe rack ‘Atany intersection point @ ofa rack pte line withthe contact line PQ of the pitch susfaces, the contacting tooth surfaces have the same direction. They have a common surface normal GC (ig. 2), whose inclination to axis C ig constant and equal to the pressure angle in the midplane 6 Normsl QC remains normal to the eonical tack tooth sides in its fixed position even asthe rack moves along axis C. Tn each rack position, it intersects the rack-tooth sde ate point of contact with ‘the hub meshing therewith. Tt is a path of contact. As this is ‘rue for all points @ ead their surface normals, these surface normals determine and define the surface of progressive contact. ‘As in gearing, this surface ean be used to establish the required tooth shape of the hub. ‘As the conical rack-looth sides, such a $}, S, have a constant, normal distenee from each other, the co-generaced hub-tooth sides also have a oonstant normal distance from each other, as re= ‘quired for smooth power transinission. ‘At any point Q, the profile inclination oF pressnse angle is con: stant in plane QC (Fig. 2) that containa the cone axis C. Hence it is bovnd to be difevent in plane g that is parallel to midplane G tnd peependiciixe o the bub axis, Tt can be shown thet the pro- file inclination 9” oF pressure angle in plane g i= seleted to the pressure angle ¢ in midplane G and to angle QCP ~ ¢ as follows: tan $ = tan ¢ cos e a The term "is smaller than g. It drops off frst very slowly” and then more rapidly with inereasing angle e To deseribe the coupling action, we shal frst look at the tooth contact without appreciable load and without any’ elastic dtc tion of the contacting tooth surfaces and teeth. The effect of elastic yielding will be introduced later. We shall also show that involute gear couplings can transmit true uniform motion, even at very large coupling angles Fig. isan axial section showing a sleeve S and n hub Hat an rangle ¢_ Here the tooth contact has shifted toward the tooth tends. "Fig. § isan axial end view of the sleeve that as steuight involute teeth with base circle "The uniform motion of the sleeve and hub isk the motion of pair of bevel gears whose axes ccincide with the sleeve and hub axes ay a, that intersect at 0. The sleeve and hub move as if two imagined conical pitch surfaces of the gear members roll on each other without sliding, ‘These rolling surfaces contact along the iustant axis J that bisects the angle between the coupling axes ‘The instant axis I a-very usoful to ahow up the relative motion of oue member with respect to the other, AL any aie instant, the Inub moves briefly as if turned about the instant axis relatively to the sleeve, This defines the dirvction of relative motion of each point of the bub. The velocity of thie motion depends on the ‘orning velocity about the instant axis. Tes known to be ob- tainable from the turning velotity w of the sleeve and hub by ‘geometric addition, as expressed in formula (2) ae (2) @ ‘Transactions of the ASMEits straight path describes a line on the tooth sides asthe coupling "This line coincides with an involute profile on the sleeve tons fxs a, and when extended is tangent to base circle b (Fig. 5), eal interecte the instant axis [at ‘The crown of the hub teeth can be determined so as to place the path of contact at a desired axial dstanice 2 from the inter section 0 of the axes, st the design misalignment angle i. ‘The foregoing requirement defines the location of the contact point P, at the pitch circle &. P, generally does not le directly on the vertical V through 0, but elose enough to it that ite distance from. the instant axis dots not differ much from pitch radivs R. For this reason, the sliding velocity v, at mean contact point P, can be put down a Go eoppren) @ where » is the peripheral velocity Ra, and 4° is the coupling ‘angle measured in degrees "Each of the two sides of the teeth has two diametrically oppo- site paths of contact, One is along normal n that intersects the instant axis / at N,- The other is along normal n’ that inter fects instant axis J at "on the opposite side from 0. The ‘contact normals n,n’ intersect the cylindrical inside surface of the sleeve teeth and & spherical outer sixface of the hub teeth. The path of contact is between the two intersection points. Its length Hetermines the duration of contact. TF it were exactly equal to the normal distance p (Fig. 1) af adjacent tooth sides, then each ‘tooth starts contact when the preceding tooth leaves off. As on sears, profile overlap is desired, a length of 1.2p to 1.6p or more, ‘Thislength and the duration of contact depend on the tooth depth ‘and on the pressuce angle or profile inclination ‘After passing through the contact, @ tooth separates from its mate, 1 contact it again at a different spot after about half a ture. The maximum separation attained depends on the coupling ange Journal of Engineering for Industry Tooth Bearing Fig. 6 is a fragmentary axial setion ofa sleeve, wherein ite involite tooth profile appear as straight line. Tn the aligned Peston of the aleve and ub, the tooth contact ie along profile ye IC sero Toad, all the teeth contacting simultaneously. On both sides of line yp, the contacting toth surfaces gredunly separate ftom each other ata rate depending on the amount of erowning. They separate at fist very slowly. The cromhatched ret round point P as aseparation within a fixe, very small amount 2, such as 0.00] in. Such a separation might be overcome by lastic defection under heavy load. The area then becomes & {oth bearing ares ‘Ata coupling ange the contact has bifted away from central pition fo two mesh zones. A tooth contactsonly atone pint a ttn, at ero load, a a point sich as Q in one turing postion; td afer ball a turn at poiot Q". The erow-hatched elipeal or ‘oval area has a seperation within a given small amount, Tt ia Smvaller than the cross-hatched aren around point P. As the Caupling tur, the contact point mover long profi y- The fectangula ares around QorQ” is within a separation 2 of geting into aoth contact at aero Toad. Under load, it may become the ‘rea swept by tooth contact With the conventional taiform crowning the width of these aes around points P and @, Q”is approximately equal. When the coupling rune at at migl, however, there are fewer teeth In contact, only two at ties at the masioum design ange, and these fer teeth have lew intimate contact. Moreover, sliding increase with increasing angle In consequence, the sustained load eapacity at the design angle is only a smal fraction ofthe capacity ofthe coupling in alignment or near-alignment Fig. 7 shows the kindof tooth beating obtained at» eubetantil langle when the profile Inclination ofthe hub tooth is constant in planes ¢, Fig 2, rerpendicular tothe hub axis rather then belng ‘constant in planes QC containing axis C. The profile inclination is then too large in lanes ,s0 thatthe tooth bearing displaced toward the top of the deve teth when the coupling rune wt the design angle, This affects the smoothness of the transmitted ‘motion and causes early wearContact Cycles and Backash Let us look at the relative motion of the hub with respect to 8 sleeve maintained stationary. Instead of turning both the sleeve fing bub on their exes, an opposite turning motion about the sleeve axis ie added to the system comprising sleeve ad hl, se that the sleeve turning Motion is eancelled out and the sleeve stands still. The hub axis then deseribes a conical surface about the sleeve axis, Ie apex is at the intersection point O of the ‘We shall first consider the case where the crowning axis C in- tersects the hub skis, at-O, and look at a spherical surface Sph centered at O ahd patsing thrsugh mean pointyP of the hub. Fig. Sis. radial view taken in direction PO. Fig. 9 aside view taken in the direetion of the sleeve axis The spherical surface Sph intersects the hub-tooth surface passing through P eub- stantially in circle (") centered st 0.” tn projection, Fig. 8 it ‘appears as a straight line that coincides with the hub axis a ‘The same sphere Sph intersects the contacting tooth surface of ‘thesleave ina curve p, whose mean curvature radiusin projection, Fig. 8, can be shown to amount 10 R ein @ on curves having only & small distance 2 from O. "The cirele and curve p, contact or nenrly contact at point P". ‘yn the relative motion, the hub exis describes « conieal surface out sleeve axis a, whereby a point He of the hub axis deseribes circle k*. Ata turing angle 8, point He reaches & position H, And at turing angle @= 90 deg and @ = 180 deg, it reaches positiona 1” and £(*, respectively. Th the view in Fig. 8, the projected bub axis Of appears inclined at an angle; to sleeve axis ir Tan j can be resdily computed as tan j = tan ¢ eos 8 ‘At a turing angle of 90 deg, when Hy ia at 1”, the hub-woth side egtin appeers projected as Fig. §, «line coinciding with the projected hub axis O17, And ta tuming angle of @ = 180 deg, is appeare projected as a straight line OH". Ttappears in Fig, 8 asf swinging about radial line PO between end positions OH and OH". ‘We ahall caw enmpute the distances 2 between circle k? and curve p, an if ince #” would swing shout radia\ line PO whereby ita lane always contains the hub axis, Although this assumption in not exactly fuléilled, it provides a close enough result st the moderstowngles wiW considered. ‘The curvature canter of projected curve p fies in the plane passing through Pat right angles to the sleeve axis a,, at a dis- ‘4 the circle k” ee stright nein ‘The distance of curvature center from projected circle Kis & cos j and the distance sof circle k’ In the more general eases where the distance R, of mean point P from the crowning axis C-C (Fig. 2) differs from R, a sphere with dius A, isconsidered. ‘The foregoing formule for: applies Also when R, is sabatiusted Cor R. ‘Ac amall angles , as in common use, the formula can be Sane formed into found to amount to = from curve p, is 2 = YAR, tn 6 tan? i sint ‘The maximum separation sie attained when 8 is 0 deg where sind = 1. Hence zo fa otn $ tants w 2 = zesint o ‘The coupling runs at minimum backlash at the maximum angle 4. The backish is inereased by Ab when the coupling is set in ‘alignment, whereby the separation 2 is added on each side Ab = 2 © ‘The foregoing figures apply vo uniformly crowned teeth Transactions of the ASME, : mF ec we oP PF RP 10% 180?” e me , a CIT] a ‘Z Tory I | D7 a a ‘The variation of separation 2 with the turning angle Vis directly shown in dingram, Fig. 10, Tt shows a circle &, with center Oy And diameter #5 = O/0*, ‘The term shows up as the distance of any point Q' of the cirele from the straight-line element X-X tangent to the circle at 0". Point Q” corresponds to x vurning angle 9 = @'0°O". Tt can also be obtained by plotting an angle 26" from center O. Separation 2 is seen 10 vary harmonically ‘with the double turning angle 28 Example: With = 20.deg, # = Ldegand R, = 2/in.;2 = 0.0010 in, from (4), ‘This is a quantity small enough that it compares with the clastic deflection ofthe teeth underlond. Underload, then, more ‘teeth get into simultaneous contact, especially at small angles. ‘We shall now determine the number of teeth in contact under load. Load Distribution ‘The number of teeth that carry the load depends on them ‘yetreen the coupling axes, on the tooth design, and also on the load. In operation atan angle at very smaliionds, exch tooth gets into contact, eparates, and contacts again after half a tum of the coupling, We have given the maximum tooth separation 2. in Journal of Engineering for Industry formula (4) for no appreciable loud snd uniformly crowned teeth. ‘The separation # at any turning angle 6 from contae\ position is defined in formula (5). ‘To estimate the number of teeth in contact, we consider average conditions, without such iregularites in contact patter as may ‘Occur whens new tooth geta into contact oF a tooth gets out of ontact. In the ease considered hare, moreover, the individual tooth load is in direct proportion to the elastic tooth deflection, the added defection of the sleeve tooth and hub tooth, both sur- face deflection and bending. Thin proportionality is at least ‘approximately fulfilled. “These tzeth are in contact whose separation is smaller than the deflection of the teeth that carry the largest individual load P,, where Pow ce lity fyctor C depends on the material wed and ‘on the tooth design. Its computation is involved and omitted hhere.- Tt ean niao be determined reliably by tet. In the test, all but two diametrically opposite teeth of the test hub are removed. Contact, with the sleeve teeth ie established at sero soupling tangle. Then torque is applied which results in a slight relative ‘turning displacement. The displacement ia measured close to the contacting teeth. C is the proportion gf the tooth load applied to the dispeament at the pitch point, ‘The proportio‘The elastic defestion z, of loaded teeth allows adjacent mating teeth to move toward each other. “The teeth whose separation 2 was equal to the elastic defection 2, of the most inted teeth then move into eontact with each other bit carry no load, ‘The teeth whose reparation = was less than 2, reloaded in proportion, to the difference (2, — 2). Their load is aes) ‘The term z is plotted iu Fig. 11 in ternis of the angle 8, the angular distance from the mean contacting tooth where 2 ~ 0. Tt fs the vertical distance from X-X of any point (Q) whose horizontal distance from O is proportional to angle 8.” ‘The curve s so obtained is sine-curve, It repeats with every half After determining z,for alond P, that can be carried by a single tooth with » margin of safety, z, is plotted in Fig. 11 from XX panda Tine B-B is drawn parallel to X-X at a distance 2, therefrom, z,should be plotted ut the scale used for zs, the maxic ‘mum soparation at the now considered eaupling angle t Line B-B cuts off the bottom of curve s, between end points QQ. Allthe teeth within the spread QQ.’ carry somé lond, that fades out and becomes zero at these end points. The load At any point is proportional to the vertical distance within the cross-hatched area at that point. The total load carried in one ‘engagement zone is proportional tothe erost-hatched ares below the spread Q-Q/. If all the teeth within 90 deg to both Sides of O would carry the maximum load P., then the total loud within that range would be proportional to the area of the rectangle 128-4. ‘Thesaid total load amounts to 1ANP, fr double that amount all around the periphery, A” = tooth nume ber. The safe load that ean be actually carried on perfectly accurate couplings isa fraction q of the load NP, where gis the ratio of the cross-hatched area f to the aroa of rectangle 1-2-4 ‘These areas can be readily computed, Using redian or arc rmessure for the angles, area tan be shown ta amour 10 J = "oz con 28, (tan 28, — are 24.) = Hes 008 20, inv 28, ® while the area of rectangle 4 isin are measure 2, = risa? 8, = Nyre( ~ 008 28.) Henge penportion g amounts to Lem 28, (tan 28 a ‘FQ — 008 28) 4 The load eapacity of an actual coupling comes the closer to the figure gNP, at che coupling angle 4; the more accurste it is, the closer its tolerances, We may introduce a caefficient K to express the tolerances; K— 1 for absolutely accurate couplings. Tt is somewhat smaller than Ion commercially accurate couplings, the difference from. 1 inereasing with increasing, tolerances. ‘Thus the Tond eapacity P, at the eovpling angle fi P, = K@NP) o Proportion q is plotted in Fig. 12 in terms of the proportion Line B-B in Fig, 11 commesponds to ®* = 0.260 9 = 0500. I 2,> 2% then g = [Me + al) * A few examples will illustrate the use ofthe g-graph in Fig, 12 Example 1: Tn the example previously given for « coupling langle f= 1 dex, ze was computed at 2» = 0.001 (in.). s,isdeter- Wee, = 25 then 1 ‘mined from the maximum safe load P, per tooth, based on surface sess as well as bending stress, If factor ( is already known, then 2, = PU/C. Let it be assumed that 2, was determined 43 = 0.00025. Then # 0.250 ati = 1 deg. Read from the ggraph, Fig. 12,9 = 0.218, If the maximum safe tooth load at a given tooth piteh and rowning is, for instance, 1000 Ib, and the hub has 20 teeth while the accuracy factor K is 0.80, then the coupling could earry a total Jod of $0,000 Ib when in alignment, and 4(40,000 Ib) = 8720 Ib ‘when running at an angle i= 1 deg Example 2: How much load oan the same coupling cary at Iasi the angle ¥ of Example 1, at = "Vs dee? isthe sameas before, wad 2,remains 0.00025, however, is smaller, as tan ¥ in formula (4) is approximately one half of the former amount, and tant 1 is one quarter thereof. Henge 25 — 0.00025 in close approximation, and 2 = Read from the q-graph, ¢ = 0.500. This results in a total tooth load of 4(40,000 1b) = 20,000 Jb that the coupling can safely carry at. coupling angle # = ¥/s de. Bxample 3: How much load can the same coupling earry at sn angle double that ofthe first example, at? = 2 dea? Here zis approximately = 0.004 and ™ = 0.0625, "The graph shows ¢ = 0,107, 0 that the total tooth load figures q(40,000 1b) = 428010. "These figures are based on contact at a mean tooth depth. At the larger gles 4, profile action should be considered ns wel Elastic tooth defletion then decreases. in importance. The ‘couplings act more and more like gears, with inereasing angles f ‘We are aware of tnt have not directly introdviced tooth sliding in the foregoing computations, wiich increases with increasing angles sj nor intimacy of tooth contact, which decreases with in- creasing angles and thereby increases the surface stress. "Thus we have treated a somewhat simplified concept based on elastic toth deflection ‘There will hardly be any disagreement with the general con- clusion that the usual couplings ean carry much more load at ‘mall running angles than at large ones, The Vari-Crown™ ‘To make the drop in load capacity lear drastic, Sie-Bath bas developed the Veri-Crown, Fig. 15 is an axial section taken throvgh © hub. In conventional crowning, as produced by hhobbing, the feed path of the hob and the tooth bottom B pro- dduoed thereby is ordinarily a eizeular are. Te may be contered at on the hub axis, or at C or C%, depending on how much of a crown is requiced. "Other methods also aim to produce the same kind of erowning. ‘With the Vari-Crown (Trade Mark), the feed path of the hob land the tooth bottom B” are more curved in the midplane @ and less curved further toward the toothends. An axial section ofthe tooth surface through pitch point P shows a curve m almect identical with the feed psth there. The varying curvature of the tooth surfaces widens the tooth contacts with inereasing distance from midplaiveG, while atthe midplane it has the smallest width. Tetends to more equalize the load eapacity at different angular ties, taking tway from the excessive load capacity at zero angi= Insity and distributing this excess to larger angulaitien Thus the capacity at the langest angularty is more thas. twice as large ‘as with the uniform crown, End Round Tn some cases, a range of angularties is specified for action under load, and an additional range for action without ap- yprecable load, There are two common ways of meeting this * Made under patunt No. 292,204; other patents pending, Transactions of the ASME‘specification, One is designing the coupling for the whole range of angularties. "This means increased crowning nd inervased stress, The other is to design the coupling for tne nguiaities under losd, and letting the tooth contact shift to the ends and ledges of the teeth for the no-load angularities, allowing sufficient Tbacklash s0 shit the coupling can take the largest angulerity Hee Journal of Engineering for Industry Without binding. Neither of these ways is quite satisfactory. For such cases, Sigr-Bath introduces the Bnd Round. It is itlustnted in Fig. 14. Crowning is sharply inereased at the tooth fends, at. ‘The main crowning is figured for the angularties under load. ‘At the maximum load-angularity, the center of the tooth coutact ie plaved at Q, where it would be normally placed. The End Round starts further out, at, » disanee from Q about half the ‘width of the contact area. It is sufficiently curved to prevent contact to shift to the end edges, keeping it on the voothside suctaces. This is tangent contact eapable of carrying moderate load, without any tendency to chew up the sleeve tooth sides. Extended Contact Design With spindle-type couplings, sometimes large angulaites are specified, but exch coupling af x group is run only within a re- stricted angular range which is diferent for different couplings. Jn general, all these couplings are made alike and are designed to run up to the largest anpulasty. Por simplicity of replacement, this arrangement is idesl. But we can make much sronger ‘couplings and multiply the coupling life perhaps 10 times if the ‘couplings do not all have to interchange with etch other For instance, if the range of adjustment of a coupling with ‘maximum angularity of 4 deg is cut down to 1 deg, 30 thet the coupling may be designed to run only between 3 and 4 deg angularity, the fond capacity based on surface stress ean be in= creased three 10 four times, and the coupling life at « given load Ingreased many more times, ‘The tooth contact can be Teh widened. Ia the extra simplicity of the prosent setup worth wasting coupling life that much? "Tt isnot dificult to have two or three diflerem sospling designs to take the place of «single design for the whole range. ‘They could be marked, for ine stance, in different colors Wearing In and Wearing Out [emightbe thought that the couplings wear ito the right shape anyhow, and that there is no need for refinements or even for faccuraty. But this we consider no more valid for couplings than It is or gears. Improper shape eauses excessive eure stresses ‘and tonds to damage the tooth surfaces, Tt is the first stage fof wearing out: We suggest that the tooth shape should be as neatly comet and adequate as possible. History and Conclusion Gear couplings started out with straight teeth on the hub. And they worked at the small angolasities where they were used. ‘As the need for large angulartiew arose and grew, it was recog hined that the tooth ends of the hub had to beeased of. Crown- fig wax invented. It was also recognized thet at the larger rngiarities the tooth contact ia confiued to two diametrically ‘opposite zones and that then centering was desirable or required. ‘A spherical outside eurface was intreduoed on the hub, centered fu the hub xis, to let the hub teeth besr against the tooth bottoms of the sleeve for centering. The contact between the spherical outside surface and the cylindrical inside surface of the tooth bottoms provides accurate centering at all angularties, ‘This may have been suggestive of the thought that the sides of the hub testh shout lio be crowned about center O (Fig. 2). ‘A while later came the realization that there was no complsion or natural law for erowning about center 0. Other crowning ‘enters were used as well to achieve different amounts of erowne ‘And now we know that we do not even need s center for crowning, that crowning may be varied along the tooth. "The stepwise progress was accompanied by an increased ‘understanding of how the teeth act. At large angularities, they fre like gears with imtersal mesh. Like gears, they transmit crue uniform motion. ‘They have, however, the peculiarity that tooth gets into contact in two zones of mesh. Between these 1Zones, the teeth separate. Their separation is much smaller than fon gears even with internal mesh. And at «mall angulariies, it is very small, 6o small that the elastic deflection of the teeth plays an important part. In this article, we have treated couplings like gears, starting out ‘by assaring vigid bodies, and formulated the separation of the teeth for rigid bodies. And then we have considered the elastic deflections of the contacting teeth, obtaining modified results therefor, We have confirmed mathematically that the number tet in contact increases with decreasing angularty and have decribed implied and approximete computation of the nl- ter of teeth in contact ae diferent angulaiien, ‘Acknowledgment We express our thanks to Emest Wildhaber, our consultant, {or his valuable advice and help in preparing this pape. Printed ‘Transactions af the ASME. wea